Latest Michael Burry (Cassandra Unchained) reports. My own summary + every ticker he names, the price, size and reason he gave for each position, with PnL performance if you followed his call. The Mentions column flags who else is in each name — Big Investors (I), Serenity (S), Activists (A), Politicians (P), Trump (T), Citrini (C).
69 reports · 329 positions · #001–#069 · Jun 24, 2026 – Nov 16, 2026 · updated Nov 17, 2026
#069
Trading Post
Nov 16, 2026
Trading Post November 16, 2026: Adobe, Oracle, Caterpillar, lululemon, Alphabet & More
Burry is rotating from a short put book that had grown large and expensive toward high-conviction longs. Puts on Oracle and Caterpillar were closed after the recent downward volatility, and with implied volatility rising he is weighing outright shorts instead of buying more puts. He added to Adobe, arguing that AI clones validate rather than threaten its workflow lock-in through file formats, installed base, enterprise contracts and security, and that its valuation more than pays for the AI-cost and subscription-growth risks. He also added to MercadoLibre, QXO and lululemon, and is donating much of his Alphabet stake, which he describes as akin to a sell on a weaker five-to-six-year outlook.
aineocloudairobotaisemiconsumershoppingenergygridfinancefintechmag5saas
Positions & performance
-0.1%avg return56% win · 9 tracked · L +0.1% · S -0.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LADBE:NASDAQ saas Adobe ADDcommon · now a full position | NASDAQS&PBitget | IPT | 246.0his fillNov 16 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | +1.0% |
| WhyAdobe's edge is that creative workflow still runs through its industry-default file formats, installed base, enterprise contracts and information security, not that its AI is better; AI clones like ArtCraft validate the need for its functionality. It remains the cheapest software name in the Cassandra Unchained universe and the share count keeps falling on bona fide buybacks. |
| LQXO:NYSE energygrid QXO ADDcommon | — | IT | 12.0his fillNov 16 '26 | 12.12 ● | 10.0427.61 | -0.7%APY -4.1% | +1.0% |
| WhyBought more QXO common; no further rationale given in this post. |
| SPLTR:NASDAQ saas Palantir HOLDputs | NASDAQS&PBitget | ISPTC | 160.39closeNov 16 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +0.6% |
| WhyRetained put position on Palantir; no new rationale given in this post. |
| SNVDA:NASDAQ aisemi NVIDIA HOLDputs | NASDAQS&PBitget | ISPTC | 156.14closeNov 16 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +0.3% |
| WhyRetained, consistent with his recent posts questioning NVIDIA's reporting changes and the durability of AI demand. |
| LGOOGL:NASDAQ mag5 Alphabet TRIMcommon (GOOG and GOOGL) · a good portion of appreciated shares (donated) | NASDAQS&PBitget | ISPTC | 283.72closeNov 16 '26 | 283.9 ● | 272.11408.61 | +17.2%APY +217.9% | +0.1% |
| WhyDonating a good portion of Alphabet shares held since COVID; he says this is akin to a sell and reflects a less promising take on Alphabet's next five or six years. |
| SSMH:NASDAQ aisemi VanEck Semiconductor ETF HOLDputs · large position | NASDAQBitget | I | 461.9closeNov 16 '26 | 461.68 ● | 315.05671.83 | +21.9%APY +643.5% | 0.0% |
| WhyRetained as a large position, the core of his bearish semiconductor / AI-capex view. |
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon | NASDAQS&P | ISAPT | 106.0his fillNov 16 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -0.6% |
| WhyLULU does not report until the second week of December, so the information vacuum keeps pressure on the stock, which he treats as an opportunity to accumulate. |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon · now largest position (with TPW AU) | NASDAQBitget | ISAPC | 1480.0his fillNov 16 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -1.2% |
| WhyKept adding into weakness; MercadoLibre is now tied with Temple & Webster as his largest position, with a full write-up in the works. |
| SMET:NYSE financefintech MetLife HOLDputs | S&P | IPT | 101.45closeNov 16 '26 | 103.15 ● | 67.33106.72 | -6.5%APY -43.3% | -1.7% |
| WhyRetained put position on the life insurer; no new rationale given in this post. |
| SORCL:NYSE aineocloud Oracle COVERputs · closed out all puts | S&PBitget | ISAPTC | 119.09closeNov 16 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +0.1%closed |
| WhyClosed the Oracle puts after continued downward volatility in the name, trimming a short put book that had grown large. |
| SCAT:NYSE energygrid Caterpillar COVERputs · closed the last bit of the puts | S&PBitget | IAPT | 570.49closeNov 16 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | -0.0%closed |
| WhyClosed the remaining Caterpillar puts on the continued downward volatility, completing the exit begun in earlier Trading Posts. |
| SNDX NASDAQ 100 Index HOLDputs | — | — | —Nov 16 '26 | — | — | — | n/a |
| WhyIndex-level puts retained as a broad hedge on AI-heavy tech alongside the single-name semis shorts. |
| LTPW AU Temple & Webster HOLDcommon · largest position (with MELI) | — | — | —Nov 16 '26 | — | — | — | n/a |
| WhyNamed alongside MercadoLibre as his largest position; a write-up is in the works. |
Full summary 490 words
This Trading Post follows a run of November posts in which Burry built up MercadoLibre, Adobe and lululemon while keeping a large bearish options book on semiconductors and AI-linked names. The headline is a rebalance: he closed some put positions and added to longs, saying the short put book had grown big enough that trimming made sense.
Most of the post defends Adobe. He bought more at about $246, which makes it a full position, and points to a chart of a share count that keeps falling on bona fide buybacks, referring readers to his May piece "Software & Payments Stocks Part II" (whose universe, the SW50, is 50 stocks). Adobe is often called the poster child for incumbents caught in AI's headlights, with copycats such as ArtCraft, a free AI-coded clone of seven Adobe apps, getting attention. Burry reads that the other way: no one clones an obsolete workflow, so the clones show the functionality is still needed. Free Photoshop alternatives have existed for decades, yet Adobe's revenue grew 10% or more in each of the last four years. He does not expect Adobe's AI to be better (it even sells third-party models such as Google Gemini inside its apps). Instead, its edge is that its file formats are the industry default, plus an installed base and enterprise contracts that open-source rivals can't reach, plus information security that corporate buyers prefer. The real risks are the cost of third-party models and the growth rate of new subscriptions. Both can be monitored, and he says the valuation more than compensates; Adobe is the cheapest software name in his universe. He also points readers to a video from last week that he thinks shows Adobe adapting.
On the short side, he closed his Oracle puts and the last of his Caterpillar puts, citing continued downward volatility in both. He keeps SMH puts as a large position, along with puts on the NASDAQ 100, NVIDIA, MetLife and Palantir. Because higher volatility is making puts expensive, he says he is looking at outright shorts instead, which would change how he expresses the bearish view.
On the long side, he bought more MercadoLibre at about $1,480. It is now his largest position, tied with Australia's Temple & Webster, and write-ups on both are in the works. He added QXO common at about $12 and lululemon at about $106. With LULU not reporting until the second week of December, he expects the information vacuum to keep pressure on the stock and give him more chances to buy. Finally, he is donating a good portion of the Alphabet (GOOG/GOOGL) shares he has held since COVID. He says plainly that this is akin to a sell and reflects a less promising view of Alphabet's next five or six years. If Adobe's subscription growth slows or its model costs rise, his long case would be challenged; a sharp rebound in semis would hurt the puts he still holds.
Key metrics
- Bought more ADBE at ~$246 — now a full position
- Adobe revenue grew 10% or more in each of the last four years
- Software & payments universe (SW50) is 50 stocks; ADBE is the cheapest of them
- Bought more MELI at ~$1480 — now largest position alongside TPW AU
- Bought more QXO common at ~$12
- Bought more LULU at ~$106; earnings not until the second week of December
- Closed all ORCL puts and the last of the CAT puts
- Still holding SMH puts (large), NDX, NVDA, MET and PLTR puts
Key insights
- Adobe is now a full position after a buy at ~$246; Burry calls it the cheapest name in his 50-stock software and payments universe (SW50).
- He reads AI clones of Adobe apps such as ArtCraft as validation: nobody clones an obsolete workflow, and free Photoshop alternatives have existed for decades while Adobe grew revenue 10%+ for four straight years.
- Adobe's moat, in his framing, is its industry-default file formats, installed base, enterprise contracts and information security, not better AI. The risks to monitor are third-party model costs and new-subscription growth.
- He covered all ORCL puts and the remainder of the CAT puts after downward volatility, trimming a put book that had grown large.
- Rising implied volatility is making puts expensive, so he is considering outright shorts instead, a possible change in how he expresses bearish views.
- Still short via puts: SMH (large), NDX, NVDA, MET and PLTR.
- MercadoLibre (bought ~$1,480) and Temple & Webster are now tied as his largest positions, with write-ups on both coming; he also added QXO at ~$12 and LULU at ~$106 ahead of December earnings.
- Donating much of his post-COVID Alphabet stake is explicitly 'akin to a sell' and signals a weaker view of Alphabet's next five to six years.
#068
Community
Nov 15, 2026
Medicinal Whiskey Charity: Making Whiskey Pay in the Fight Against Childhood Illness
This is Burry's November Charity of the Month post, not an investment memo, and it states no positions. He backs Medicinal Whiskey Charity, a volunteer, physician-led nonprofit in South Carolina and Kentucky that turns bourbon collectors' enthusiasm into grants for children's hospitals and transplant programs. He calls its Form 990s the cleanest he has vetted since starting the newsletter, and he is donating 5% of the newsletter's November gross revenue to it.
Positions & performance
No single-stock positions stated in this report.
Full summary 306 words
Burry's November 2026 Charity of the Month is Medicinal Whiskey Charity (MWC). It was formed during the COVID pandemic by a physician-heavy group of bourbon enthusiasts and operates in South Carolina and Kentucky. It has funded the University of Kentucky's organ transplant program, Kentucky Children's Hospital, Duke Children's and the Burn Foundation of America, among others. Burry knows one of its doctors, Dr. Malay Shah, surgical director of UK HealthCare's liver transplant program. Distilleries such as Willett, Buffalo Trace and Old Forester provide barrels at cost and donate rare bottles, and master distillers Fred Noe and Freddie Johnson donate their time. Donors give in exchange for barrel picks, galas and one-of-a-kind bottles, and the charity itself never sells liquor.
He treats the vetting like a filing review. MWC has $0 compensation expense, grants are 95.9% of total expenses, and after four years only 6.9 cents of every dollar raised is still on the balance sheet. All four Form 990s show zero liabilities, zero compensation, and no grants, loans or related-party dealings with insiders. He calls them the cleanest filings he has vetted since starting Cassandra Unchained. In 2024 the charity gave $527,350 in itemized grants: $300,000 to the UK transplant program, $80,550 to Prisma Health, $75,000 to Clement's Kindness Fund, $30,000 to Duke Children's, and smaller grants including $8,000 for Hurricane Helene relief. About $7,000 more went in grants below the IRS itemization threshold.
He adds some history: whiskey was official medicine for centuries before the AMA turned against it in 1917, and the word comes from aqua vitae, 'water of life.' MWC's fifth annual Winter Gala is at the Carrick House in Lexington on December 11, 2026. The gala has raised about $1.4 million over four years. Cassandra Unchained will donate 5% of its November gross revenue to MWC. No trades or tickers are discussed.
Key metrics
- $0 compensation expense; run entirely by volunteers
- Grants are 95.9% of total expenses
- Only 6.9 cents of every dollar raised remains on the balance sheet after four years
- 2024 itemized grants: $527,350, plus about $7,000 below the IRS $5,000 itemization threshold
- Largest 2024 grant: $300,000 to the University of Kentucky transplant program
- Winter Gala has raised about $1.4 million over four years
- 5% of Cassandra Unchained's November gross revenue goes to the charity
Key insights
- This is a charity post with no positions or tickers; the newsletter is donating 5% of its November gross revenue to Medicinal Whiskey Charity.
- Burry checked the Form 990s: $0 compensation, zero liabilities and no related-party transactions in all four years. He calls them the cleanest he has vetted.
- Grants are 95.9% of total expenses, and only 6.9 cents of each dollar raised remains on the balance sheet.
- 2024 grants totaled $527,350 itemized, led by $300,000 to the University of Kentucky transplant program.
- Distilleries supply barrels at cost and donate rare bottles; the charity never sells liquor and raises money through donations for access.
#067
Short Thoughts
Nov 15, 2026
How NVIDIA’s Changed Asian Revenue Reporting “Disappeared” Singapore - and What Else?
NVIDIA changed how it reports revenue by geography in late 2025, from billing location to customer headquarters, and recast FY2024-FY2025 to match. That made pre-ban China revenue look higher, so the export-ban drop looked cleaner, while Singapore, a fast-growing billing hub, vanished as a line. Burry argues the new method makes it harder to see where chips were actually installed, just as WSJ and Culper Research reporting points to a Southeast Asian back door to China. Understated China revenue would most likely mean fines rather than accounting fraud, but any revenue misstatement could bring restatements and delayed filings. So he keeps his NVDA puts and says he would not buy the stock at any price in the near term.
aisemi
Positions & performance
+4.5%avg return100% win · 1 tracked · S +4.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SNVDA:NASDAQ aisemi NVIDIA HOLDputs | NASDAQS&PBitget | ISPTC | 162.85his fillNov 15 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +4.5% |
| WhyHe keeps his NVDA puts because he thinks NVIDIA's China exposure is understated and its 2025 switch to reporting revenue by customer headquarters hides where chips end up. He fears purchase commitments, ecosystem fragility and margin pressure from customers and competitors will still surface, and says he will not buy the stock at any price in the near term. |
Full summary 580 words
Burry wrote this Short Thoughts post on the morning the Wall Street Journal published an investigation into how Jensen Huang built a 'backdoor route' that keeps Beijing in the AI race. The story describes huge Southeast Asian data centers that used a gap in U.S. sanctions to give Chinese firms computing power and to cover alleged chip smuggling. Burry says the press has finally started doing its job on NVIDIA, sadly only after a lurid affair. He names Kakashii and Gary Marcus as fellow early skeptics and says he will now step back from the story. He confirms he still holds NVDA put options and will report any trades. The post follows his November 13 piece 'Say It Ain't So - NVIDIA on Fire' and his earlier essay on Jensen Huang's fall.
His new point is about reporting. Many people remember that China's share of NVIDIA revenue seemed to drop before the export ban, but fewer noticed that Singapore's share kept growing fast until it disappeared. In Q3 of FY2026 (late 2025), NVIDIA stopped counting revenue by billing or invoicing location and started counting it by customer headquarters. Burry notes this says much less about where chips are actually installed. The change was disclosed, and FY2024 and FY2025 were recast. The recast raised China revenue in the pre-ban years, so it no longer seemed to be falling before the ban. That made the real export-ban drop look even bigger, just as the Singapore line vanished. Burry calls it a 'mild form of manipulation': recasting is standard, but the effect was convenient, a compliance picture 'perhaps too clean.'
As evidence, he shows charts of where the Singapore revenue went: mostly to U.S. headquarters, plus $7.9 billion to China. By NVIDIA's own disclosure, $7.9 billion of FY2025 China-headquartered revenue was invoiced through entities outside China. That was before the ban, when most of it was legal, but it shows the offshore channel was already in use. He also cites the WSJ profile of Culper Research's Christian Lamarco, whose May 2026 report flagged the relationship between Jensen Huang and Alice Huang. Culper estimates China was over 20% of FY2026 compute revenue, more than 15% of total revenue. Burry says the incentive and the means for 'unsavory actions' were both present.
He is careful about what this proves. Understated China revenue would mainly risk fines, not accounting fraud, and would not by itself force a restatement. If NVIDIA misstated revenue in any way, though, that would be serious: restatements, delayed filings and more. There is no such evidence yet, partly because the new method makes it harder to see. The October quarter is already closed, so NVIDIA only has to report well and hold guidance. If the 10-Q is on time and guidance holds, he worries that just means nothing has surfaced yet, and he sees the 10-K, one quarter away, as the bigger test.
On positioning, NVDA closed Friday at $162.85, about $4 trillion fully diluted, near the low end of this year's range. Burry says he will not buy at any price in the near term. He points to purchase commitments, the fate of the 'Byzantine' ecosystem NVIDIA sits at the center of, and customers and competitors squeezing margins by exploiting the gap between chip delivery cycles and data-center build times. He keeps the puts. A timely 10-Q and 10-K with no restatement or regulatory fallout would weaken the near-term case. He ends with a reading list of his earlier NVIDIA pieces.
Key metrics
- NVDA closed Friday at $162.85, about $4 trillion fully diluted market cap, near the low end of this year's range
- Reporting switched in Q3 FY2026 (late 2025) from billing/invoicing location to customer headquarters location
- FY2024 and FY2025 recast under the new method, raising China revenue in the pre-ban years
- FY2025: $7.9 billion of China-headquartered revenue was invoiced through entities outside China, per NVIDIA's own disclosure
- Singapore's former revenue line mostly moved to U.S. headquarters, with $7.9 billion moved to China
- Culper Research estimates China at >20% of FY2026 compute revenue (>15% of total revenue)
- Fiscal Q3 2027 (October quarter) already closed; the 10-K is one quarter away
Key insights
- In Q3 FY2026 NVIDIA switched geographic revenue from billing location to customer HQ and recast FY2024-FY2025. Pre-ban China revenue went up, so the ban-era drop looked bigger and cleaner.
- Singapore, a fast-growing billing hub, disappeared as a reporting line. Its revenue mostly moved to U.S.-headquartered customers, with $7.9 billion moved to China.
- By NVIDIA's own figures, $7.9 billion of FY2025 China-HQ revenue was invoiced through entities outside China, evidence the offshore channel existed before the ban.
- Culper Research estimates China at over 20% of FY2026 compute revenue (over 15% of total), above what NVIDIA reports.
- Burry separates the risks: understated China revenue alone points to fines, not fraud. Only an actual revenue misstatement would bring restatements and delayed filings.
- Watch the filings: a delayed 10-Q or withdrawn guidance would be a red flag, and the FY2027 10-K one quarter away is the bigger test.
- He still holds NVDA puts, with the stock at $162.85 (about $4 trillion), and says he will not buy at any price in the near term.
- Other worries he lists: purchase commitments, ecosystem fragility, and margin pressure from the gap between chip delivery cycles and data-center build times.
#066
Short Thoughts
Nov 13, 2026
Say It Ain’t So - NVIDIA on Fire - Read All About It!
The Wall Street Journal's story on Jensen Huang's ouster leads with affairs and boardroom drama, but Burry says the real news is buried in it. This summer, major cloud providers were already moving away from Nvidia chips. At the same time, the wafers and memory Nvidia had committed to buy were piling up like an "avalanche," and Nvidia was offering billions in guarantees to customers that couldn't afford its chips. That is the supply-side capital cycle Burry has described since his first post in November 2025: Nvidia added operating leverage just as demand shifted toward cheaper, non-GPU inference hardware. He concludes the AI bubble started to come apart for real this year. Nvidia may look cheap at about $4T, but analysts' free-cash-flow projections missed the turn.
aisemi
Positions & performance
+5.2%avg return100% win · 1 tracked · S +5.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSMH:NASDAQ aisemi VanEck Semiconductor ETF HOLDputs | NASDAQBitget | I | 486.75closeNov 13 '26 | 461.68 ● | 315.05671.83 | +21.9%APY +643.5% | +5.2% |
| WhyBurry says his SMH puts, along with his NASDAQ 100 puts, are now his only exposure to NVIDIA. He holds no NVDA shares or puts directly. The semis ETF puts let him bet on the AI capital-cycle bust without a single-name position in NVIDIA. |
| SNDX NASDAQ 100 Index HOLDputs | — | — | —Nov 13 '26 | — | — | — | n/a |
| WhyThese index puts are his broad-market hedge on the AI bubble coming apart, and with the SMH puts they are his only remaining NVIDIA exposure. He argues the fallout from the AI bubble is only starting now. |
Full summary 585 words
This Short Thoughts note came out the night after the Wall Street Journal's lead story on Jensen Huang's ouster as NVIDIA CEO, titled "Before His Ouster, Jensen Huang Belittled the Executive Replacing Him." It follows Burry's previous post on Huang's fall. Most of the coverage was about affairs and about Huang's harsh words for incoming CEO Colette Kress. Burry says the press is missing the real story, which is already in the Journal's own paragraphs, and quotes them at length with his own emphasis.
The passages he highlights describe an August meeting Huang called with his finance team. The AI economy, "backstopped by Nvidia's increasingly outrageous bets," was teetering. Nvidia was in talks to provide billions of dollars of guarantees to companies that couldn't afford its chips. Some of these deals were led by Huang's son-in-law, Nico Caprez, who offered financing to upstart cloud providers and data-center developers in hopes of making them long-term customers. Huang reportedly worried that his dealmaking spree had gone too far. He said major cloud providers were "running away" from Nvidia toward alternative chips, and that the wafers and memory Nvidia had committed to buy were coming at him like an "avalanche." Burry says this is why the stock lagged and kept getting "cheaper" all year, even before the affairs came out, while CNBC and sell-side analysts kept pitching it to retail investors.
Burry argues that his NVIDIA thesis "goes far beyond illicit affairs." He links back to the first Cassandra Unchained article, "The Cardinal Sign of a Bubble: Supply-Side Gluttony" from November 23, 2025, which introduced the capital-cycle framework as part one of "The Heretic's Guide to AI's Stars." He jokes that it might have been better titled "Beware: Avalanche Territory." He also points to the follow-ups: "Unicorns & Cockroaches: The Blessed Fraud," "The Supply Side Gluttony Recurrence," and "The Blessed Fraud Recurrence." In his view, a predictable paradigm shift caused Huang's "avalanche" of worries, and Nvidia added operating leverage right into it. He lists that alongside the duration mismatch he has discussed before as one of several long-term problems for the company. He adds that GPUs were not originally designed for AI, and calls them "overkill" resource hogs that most inference, and even training of small language models, does not need.
On valuation, he notes NVDA bounced hard off $150 and spent most of the day and after hours in the mid-$160s, about $4.0 trillion in market cap, with retail piling in. He grants that the buyers "are not necessarily wrong": the valuation looks low, a WSJ Heard on the Street column highlights growing free cash flow, and "there is a price for everything." But he warns that those projections missed the shift because analysts' job is "to predict rosy numbers." His conclusion is that the AI bubble started to come apart this year and the fallout is only beginning. He says the clues were in the stock price, the bubble's financing, and SEC and regulatory filings, and he expects reporters to dig out the rest.
On positioning, Burry states he is not currently involved in NVIDIA shares or puts. His only exposure is through puts on the VanEck Semiconductor ETF (SMH) and NASDAQ 100 puts, both carried over from earlier posts. He closes with thanks to Gary Marcus and Kakashii for pushing back on the consensus. The view would be invalidated if hyperscaler demand for Nvidia GPUs held up and the committed supply cleared without the guarantees and vendor financing he sees as signs of a bubble.
Key metrics
- NVDA bounced hard off $150, traded in the mid-$160s intraday and after hours
- NVDA market cap about $4.0 trillion at the mid-$160s
- Huang's August finance meeting: Nvidia was in talks to provide billions of dollars in guarantees to customers that couldn't afford its chips
- Committed wafer and memory purchases described by Huang as coming at him like an "avalanche"
- Thesis dates back to Burry's November 23, 2025 piece "The Cardinal Sign of a Bubble: Supply-Side Gluttony"
- Burry has no direct NVDA shares or puts; his only NVIDIA exposure is through SMH puts and NASDAQ 100 puts
Key insights
- Burry says the key detail in the WSJ ouster story is that, by summer 2026, major cloud providers were planning alternative chips to reduce their reliance on Nvidia.
- Huang himself reportedly described Nvidia's committed wafer and memory purchases as an "avalanche." Burry calls this the supply-side gluttony he first warned about in November 2025.
- Nvidia was in talks to guarantee billions of dollars for customers that couldn't afford its chips, including deals led by Huang's son-in-law. Burry treats this vendor financing as a classic late-cycle sign.
- Burry says the stock's weakness all year, while CNBC and analysts kept recommending it, was the market pricing in this deterioration before the scandal broke.
- He grants that NVDA at about $4.0T in the mid-$160s, after bouncing off $150, looks cheap on free cash flow. He warns that sell-side FCF projections missed the shift to cheaper, non-GPU compute.
- He calls GPUs "overkill" resource hogs that most inference and small-model training don't need. That is the structural reason he expects Nvidia's demand to erode.
- Burry has no direct NVDA position. His only exposure is SMH puts plus NASDAQ 100 puts, so this is a sector and index bet rather than a single-name short.
#065
Single Stock
Nov 12, 2026
The Don Goes Down: What Jensen’s Fall Means
Jensen Huang's fall removes not just NVIDIA's key man but the enforcer of the terms of trade and financing architecture that Burry argues supercharged its results. Without him, customers have cover to slow purchases, renegotiate or defect, and a Board in crisis — with DOJ already investigating tying and coercion, and Congressional scrutiny of the Megaspeed export-control episode looming — will rein in circular financing and pressure on buyers. That exposes the duration mismatch between NVIDIA's fast GPU cycle and the slow build-out of data centers, so growth may slow and margins fall. The coming quarter is already booked and will look great, which he sees as bulls' last good chance to sell.
aisemi
Positions & performance
No single-stock positions stated in this report.
Full summary 503 words
Burry responds to the news that NVIDIA's CEO Jensen Huang — 'the Don' — has fallen. He casts Huang as both the boss and the 'black-jacketed enforcer' of what he calls a protection racket, noting that the DOJ was already investigating whether NVIDIA abused its dominance by forcing customers to bundle GPUs with other products, restricting choices, or charging more to customers who did not use its GPUs. Citing TechCrunch and Axios, he says Huang had as many as 60 direct reports and delegated little, so CFO Colette Kress may become interim CEO because no operating executive has global knowledge of the company. He does not believe any successor can replace Huang seamlessly; the key-man risk he always worried about has arrived.
The essay then walks through why the loss matters more than a normal CEO change. First, circular financing: NVIDIA bought $42.4 billion of equity securities in the first half of 2026, roughly 30 cents of every revenue dollar recycled into the AI ecosystem, alongside third-party platforms with Apollo, BlackRock, Blackstone, KKR, Brookfield, Morgan Stanley and Goldman Sachs that funnel pension, insurer, sovereign and household money into off-balance-sheet SPVs buying GPUs for cloud providers. He says he has found leverage of up to 76x in the system. Huang orchestrated it; a Board in crisis will rein in risk, and new investment may become tentative or even be written off.
Second, channel stuffing: he repeats the allegation that NVIDIA pushed forward sales — under threat of smaller future allocations — onto buyers unable to run the chips, which then sit undepreciated in warehouses as construction in progress. Huang's exit gives customers cover to slow deals, renegotiate or move to competitors, and any heavy-handed attempt to keep them in line would collide with antitrust exposure while the DOJ is 'already in the house'. Third, Congressional risk: he highlights Megaspeed's GPU purchases just before an export-control regime took effect, and the company's employment of Huang's mistress while Huang was close to Trump, arguing that evidence, or even just headlines and allegations, will surface. He generalizes that this is how key-man risk shows up now, and boards everywhere will reassess it.
On what comes next, he argues NVIDIA's business will not disappear but growth may slow and, critically, margins may fall. The core problem is the duration mismatch between NVIDIA's rapid GPU cycle and the much longer time needed to develop, finance and build data centers to house the chips; if channel stuffing, circular financing and coercion can no longer paper over it, the mismatch becomes obvious. The terms of trade and financing architecture were the 'secret sauce', and also exactly what he and a few others have flagged as the real problem. He notes that the quarter due in under a week is already booked and will be strong, and ends bluntly: bulls will get 'another fluffy big chance to get out. Get out.' No new trade is stated in this post; it is a bearish reiteration on NVDA rather than a trade log entry.
Key metrics
- NVIDIA bought $42.4B of equity securities in 1H 2026 alone
- Roughly 30 cents of every NVIDIA revenue dollar recycled into the AI buildout ecosystem
- Leverage in off-balance-sheet GPU financing structures found as high as 76x
- Jensen reportedly had as many as 60 direct reports (TechCrunch, Axios)
- CFO Colette Kress may step in as interim CEO
- NVIDIA reports its quarter in less than a week
Key insights
- Burry frames Huang as NVIDIA's enforcer: his exit weakens the leverage NVIDIA used over customers on allocations, bundling and terms.
- Circular financing is large: $42.4B of equity purchases in 1H26, about 30% of revenue recycled into the ecosystem, with up to 76x leverage in off-balance-sheet GPU SPVs.
- He expects a crisis-mode Board to pull back on that financing, making new ecosystem investments tentative and raising write-off risk.
- Customers now have cover to slow orders, renegotiate or defect — and DOJ's existing tying/coercion probe constrains NVIDIA's ability to push back.
- Megaspeed's pre-export-control GPU buying and its links to Huang are flagged as a likely Congressional and headline risk.
- The duration mismatch between GPU cycles and data-center build timelines is the crux; without the financing 'secret sauce', growth slows and margins fall.
- The imminent quarter is already booked and will look great — he calls it a chance for bulls to exit: 'Get out.'
#064
Trading Post
Nov 11, 2026
Trading Post & Prattle: November 11, 2026
Burry is leaning into a cluster of quality franchises whose share prices he thinks have been punished for temporary reasons — currency, a squeezed consumer, quiet management, or being out of favor as the 'anti-AI trade'. Several of them (SFM, PYPL, JD) are shrinking their share counts at depressed prices, which compounds per-share value while the market waits. Adobe's crackdown on perpetual licenses should convert holdouts into subscribers, much as Netflix's password crackdown did. Separately, he is bullish on copper because AI buildouts will expose aging power and grid infrastructure worldwide, and Ero Copper's S&P upgrade confirms a deleveraging miner executing on its plan.
airobotconsumerhealthconsumershoppingfinancefintechmag5saas
Positions & performance
-1.0%avg return29% win · 7 tracked · L -1.0%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADD | NASDAQ | IAPT | 48.0his fillNov 11 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | +5.2% |
| WhyBought more in the $48s after the stock lost its growth multiple. Share count has shrunk from 133M to 92M over a decade and buybacks at these prices compound that; he sees the weak-consumer, grocery-inflation squeeze (which favors COST/WMT) as temporary. |
| LADBE:NASDAQ saas Adobe ADD | NASDAQS&PBitget | IPT | 237.6closeNov 11 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | +4.5% |
| WhyAdobe's hard countdown disabling perpetual (e.g. Acrobat 2020) licenses forced him personally onto a subscription; he expects a Netflix-password-crackdown-style revenue lift, sees a stronger moat than the market assumes plus AI tooling upside, and loves the valuation. |
| LPYPL:NASDAQ financefintech PayPal ADDsmaller but still full position | NASDAQS&PBitget | ISPT | 59.0his fillNov 11 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | -0.5% |
| WhyBought more in the $59s. Market cap is shrinking fast from aggressive buybacks below the price at which PayPal rejected a takeover bid; 'PayPal's not going away' and will grow — a concrete value case. |
| LLULU:NASDAQ consumershopping lululemon athletica NEW | NASDAQS&P | ISAPT | 106.0his fillNov 11 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -0.6% |
| WhyBought back in after a ~4% drop to the $106s. New management has not fixed the marketing/messaging black hole, leaving the stock forgotten, but he trusts LULU's research and product development and stands by his June 'Ode to Bad Management' analysis — a good bet with little downside. |
| LJD:NASDAQ airobot JD.com ADD | NASDAQBitget | IS | 27.85closeNov 11 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | -1.6% |
| WhyCalls JD the most undervalued stock in his universe: share count down 220M+ since 2023 to 1.34B, cash conversion cycle now negative ~3.5 weeks giving large working-capital float, and HK tech is out of favor as the 'anti-AI trade'. Wishes it bought back more. |
| LMELI:NASDAQ airobot MercadoLibre ADD | NASDAQBitget | ISAPC | 1500.0his fillNov 11 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -2.5% |
| WhyBought more under $1,500. The falling Brazilian real is hurting the bottom line and the credit portfolio worries the market, but he thinks MELI reserves appropriately for its retained first-loss securitization slices and the stock has been punished too harshly. |
| LERO Ero Copper ADD | — | I | 30.92closeNov 11 '26 | 27.46 ● | 20.6140.83 | -20.9%APY -77.7% | -11.2% |
| WhyAdded on the pullback after S&P upgraded Ero to BB- (stable) on Tucuma ramp, ~80k oz gold, EBITDA ~+75% to ~$760M and leverage falling below 1.0x. He is bullish copper as AI exposes aging grid infrastructure, creating a long-term supply/demand gap. |
Full summary 525 words
This Trading Post follows the previous day's 'Portpourri' portfolio update and records a busy day of buying across his existing long book. Burry frames almost every trade the same way: a temporary problem has produced a pricing opportunity, and he is adding into it. He bought more MercadoLibre under $1,500, more Sprouts Farmers Market in the $48s, more Adobe, more PayPal in the $59s, more JD.com, and more Ero Copper, and he bought back into lululemon after a nearly 4% drop to the $106s.
On MercadoLibre he acknowledges the two bear points — the falling Brazilian real hitting the bottom line (only partly offset by making Temu, Shein and Amazon less competitive) and worry over the credit book. He explains that MELI securitizes its loans but retains a significant equity/first-loss slice, and argues it is already reserving against that piece and other bad loans, building a finance arm 'with its eyes wide open'. On Sprouts he notes the stock briefly earned a growth multiple and has been over-punished; shares outstanding have fallen from 133 million to 92 million over a decade and cheap buybacks will continue. Grocery inflation and a weaker consumer favor Costco and Walmart today, but he expects fiat debasement to become 'comfortable' again, as after the 1970s.
The Adobe section is anecdotal: after years running Acrobat 2020, he hit an undismissable warning, lost access to his files and had to pay for a subscription, with no perpetual option. He expects this to play out widely, likening it to the Netflix password crackdown, and thinks Adobe's moat is stronger than most believe, with AI tools likely to enhance its products. On lululemon he says new management has done little to fix the brand's marketing and messaging 'black hole', leaving the stock forgotten, but he trusts its R&D and stands by his June article. PayPal is now 'a smaller but still full position'; he points to a market cap shrinking quickly as the company buys back stock below the price at which it rejected a takeover bid.
JD.com he calls the most undervalued stock in his universe. The share count is down more than 220 million since 2023 to 1.34 billion, though he wishes the company would use its cash pile to buy back more; the cash conversion cycle is now about negative 3.5 weeks, giving JD substantial working-capital float. He argues the whole HK tech group is out of favor as the 'anti-AI trade'.
The longest section is Ero Copper, where he quotes S&P's upgrade of the Vancouver-based Brazilian miner to BB-: Tucuma reached commercial production in mid-2025, consolidated copper output is up over 70% since 2024, 2026 copper should top 150 Mlbs with about 80,000 oz of gold from stockpiles, adjusted EBITDA should rise roughly 75% to about $760 million, and debt/EBITDA should fall to just under 1.0x from 1.8x. He adds that the stock moves with copper in exaggerated fashion, and that he follows miners that deliver until he has reason to doubt the results. His broader copper view is that AI development will reveal insufficient, aging infrastructure across many countries and drive years of upgrades, creating a long-term supply/demand problem.
Key metrics
- MELI bought under $1,500
- SFM bought in the $48s; shares outstanding down from 133M to 92M over the last decade
- LULU down almost 4% to the $106s on the day
- PYPL bought in the $59s — now a smaller but still full position
- JD share count down 220M+ since 2023 to 1.34B; cash conversion cycle about -3.5 weeks
- S&P upgraded Ero Copper to BB- (stable); copper output up 70%+ since 2024, 2026 production >150 Mlbs rising toward 175 Mlbs
- S&P: Ero 2026 adj. EBITDA ~$760M (+75% y/y), ~80,000 oz gold, debt/EBITDA just below 1.0x vs 1.8x in 2025, ~$130M debt repaid YTD
- S&P: copper averaging $6.10/lb YTD (+35% vs 2025); gold $4,500/oz (+30%)
Key insights
- MELI: he defends the credit book — MELI keeps a first-loss slice of its securitizations but already reserves against it — and added under $1,500 despite BRL weakness.
- SFM: share count has fallen from 133M to 92M in a decade; buying in the $48s lets buybacks retire stock cheaply while the consumer squeeze (favoring COST/WMT) is temporary.
- ADBE: forced migration of perpetual licenses (he was personally locked out of Acrobat 2020) could produce a Netflix-password-crackdown-style subscription lift.
- LULU: re-entered in the $106s after a ~4% drop; the issue is marketing/messaging under new management, not product — he stands by his June thesis.
- PYPL: added in the $59s; buybacks below the rejected takeover price are shrinking market cap quickly — now a smaller but still full position.
- JD: 'most undervalued stock in my universe' — 1.34B shares (down 220M+ since 2023) and a negative ~3.5-week cash conversion cycle; HK tech trades as the anti-AI trade.
- ERO: S&P upgrade to BB- with debt/EBITDA heading below 1.0x and EBITDA ~$760M; he is a copper bull because AI will expose aging grid infrastructure globally.
#063
Trading Post
Nov 10, 2026
Portpourri November 10, 2026 - New Portfolio Feature!
Burry is running a concentrated value book in beaten-down consumer and software names, led by Temple & Webster and MercadoLibre at over 10% each, and is buying more as prices fall (MELI, ADBE, JD, SFM LEAPs). He argues these companies earn plenty relative to their valuations and are backed by buybacks, so weak charts are an opportunity rather than a warning. He offsets this with a basket of puts on semis, the Nasdaq 100 and richly valued names, consistent with his view that AI capex and GPU depreciation set up a bear market. On Nvidia, he sees the CEO scandal reviving China-trade and antitrust scrutiny that could speed up the erosion of its moat, so he is not a long-term buyer despite a likely bounce trade. The midterm results make him think Trump will try to settle Fannie and Freddie, and he is watching for a moment to move from the junior preferreds back into the common.
aineocloudairobotaisemiconsumerhealthconsumershoppingenergygridfinancefintechhealthsaas
Positions & performance
+0.7%avg return44% win · 16 tracked · L 0.0% · S +2.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LMOH:NYSE health Molina Healthcare HOLDcommon · large, just behind the top two | — | ITC | 206.35closeNov 10 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +13.0% |
| WhyNamed among his next-largest holdings behind TPW and MELI; no new action stated. |
| SPLTR:NASDAQ saas Palantir HOLDputs | NASDAQS&PBitget | ISPTC | 168.82closeNov 10 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +5.5% |
| WhyRemains in the put basket, one of his longest-running bearish views. |
| LADBE:NASDAQ saas Adobe ADDcommon | NASDAQS&PBitget | IPT | 236.7closeNov 10 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | +4.9% |
| WhyAdded on the report date. Adobe is forcing perpetual-license users onto the cloud product (he had to pay up himself), and buybacks at these low prices should be supportive, so he is fine with it sitting in the low $200s for a while. |
| SSMH:NASDAQ aisemi VanEck Semiconductor ETF HOLDputs | NASDAQBitget | I | 483.63closeNov 10 '26 | 461.68 ● | 315.05671.83 | +21.9%APY +643.5% | +4.5% |
| WhyPart of his standing put basket against AI/semis and richly valued names, consistent with his GPU-depreciation and capex-cycle arguments. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADDLEAP calls | NASDAQ | IAPT | 48.87closeNov 10 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | +3.3% |
| WhyAdded to the call LEAPs on the report date. Part of a consumer retail cluster (with LULU, DECK, BIRK, BBW) whose concentration he says now limits his appetite for more in the space. |
| SCAT:NYSE energygrid Caterpillar HOLDputs | S&PBitget | IAPT | 584.57closeNov 10 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +2.4% |
| WhyRemains in the put basket (trimmed in the prior Trading Post). |
| LHCA:NYSE health HCA Healthcare HOLDcommon · large, just behind the top two | S&P | IAPT | 504.42closeNov 10 '26 | 511.78 ● | 353.99556.52 | +41.4%APY +174.8% | +1.5% |
| WhyNamed among his next-largest holdings behind TPW and MELI; no new action stated. |
| SMET:NYSE financefintech MetLife HOLDputs | S&P | IPT | 102.44closeNov 10 '26 | 103.15 ● | 67.33106.72 | -6.5%APY -43.3% | -0.7% |
| WhyListed in the current put basket; no new rationale given in this post. |
| SORCL:NYSE aineocloud Oracle HOLDputs | S&PBitget | ISAPTC | 117.98closeNov 10 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | -0.9% |
| WhyRemains in the put basket, tied to his AI-cloud capex and hyperscaler forensic work. |
| LFISV:NASDAQ financefintech Fiserv HOLDcommon · large, just behind the top two | NASDAQS&PBitget | ISAPT | 63.32closeNov 10 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | -1.3% |
| WhyNamed among his next-largest holdings behind TPW and MELI; no new action stated. |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon · over 10% of portfolio (co-largest) | NASDAQBitget | ISAPC | 1490.24closeNov 10 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -1.9% |
| WhyAdded on the report date and a few sessions earlier, both times under $1,500; he likes the long-term prospects from the $1,400s. Says it may head lower and plans another buy if it falls a further 20%. |
| LBBW:NYSE consumershopping Build-A-Bear Workshop HOLDcommon | — | S | 30.09closeNov 10 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | -2.2% |
| WhyListed as part of the consumer retail concentration; he says he will be careful about adding to this space further. |
| LVEEV:NYSE health Veeva Systems HOLDcommon (sold calls removed) | S&PBitget | IT | 305.55closeNov 10 '26 | 298.04 ● | 148.05325.99 | +18.0%APY +87.4% | -2.5% |
| WhyStill owns it after a significant run and has closed out the calls he had sold against it, so it is now an unencumbered long. |
| LJD:NASDAQ airobot JD.com ADDcommon | NASDAQBitget | IS | 28.31closeNov 10 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | -3.2% |
| WhyAdded on the report date; he says the stock is sitting forgotten and far too cheap. Also part of his China-consumer basket alongside the Hong Kong names. |
| LBIRK Birkenstock HOLDcommon · large, just behind the top two | — | IA | 37.27closeNov 10 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | -4.5% |
| WhyListed among the positions following closely behind TPW and MELI. Part of the consumer retail concentration he flags. |
| LDECK:NYSE consumershopping Deckers Outdoor HOLDcommon · large, just behind the top two | S&P | IPT | 89.79closeNov 10 '26 | 83.46 ● | 77.20122.29 | +4.1%APY +40.1% | -7.0% |
| WhyHas not swapped back into LULU and may keep Deckers, which he likes very much. Both have crashed toward the high double digits but make plenty of money relative to valuation and will be buying back stock. |
| LTPW AU Temple & Webster HOLDcommon (also OTC TPLWF) · over 10% of portfolio (co-largest) | — | — | —Nov 10 '26 | — | — | — | n/a |
| WhyOne of his two largest positions, mostly written up for a forthcoming piece. He reads the chart as a high-volume bottom that has now turned. |
| L700 HK Tencent HOLDcommon · a little over 2% | — | C | —Nov 10 '26 | — | — | — | n/a |
| WhyPart of an unchanged Hong Kong basket dependent on the China consumer; stocks have been lackadaisical but the companies keep making money and gaining share. |
| L3690 HK Meituan HOLDcommon · a little over 2% | — | — | —Nov 10 '26 | — | — | — | n/a |
| WhyPart of the unchanged Hong Kong China-consumer basket; he notes the sticky consumer backdrop but continued earnings and share gains. |
| L6690 HK Haier Smart Home HOLDcommon · a little over 2% | — | — | —Nov 10 '26 | — | — | — | n/a |
| WhyPart of the unchanged Hong Kong China-consumer basket. |
| L6862 HK Haidilao HOLDcommon · a little over 2% | — | — | —Nov 10 '26 | — | — | — | n/a |
| WhyPart of the unchanged Hong Kong China-consumer basket. |
| L1211 HK BYD HOLDcommon (also OTC BYDDDF) · a little over 2% | — | — | —Nov 10 '26 | — | — | — | n/a |
| WhyPart of the unchanged Hong Kong China-consumer basket. |
| LFNMA Fannie Mae HOLDjunior preferred shares | NASDAQ | — | —Nov 10 '26 | — | — | — | n/a |
| WhySwapped common for junior preferreds about a month ago. The midterm results lead him to expect Trump to use his limited remaining power to wrap up the GSE saga; he is monitoring whether to switch back into the common but has not yet. |
| LFMCC Freddie Mac HOLDjunior preferred shares | NASDAQ | — | —Nov 10 '26 | — | — | — | n/a |
| WhySame as Fannie: held via junior preferreds after the October swap; he may move back into the common if a resolution under Trump looks likely. |
| SNDX Nasdaq 100 Index HOLDputs | — | — | —Nov 10 '26 | — | — | — | n/a |
| WhyIndex puts in the same basket, a broad hedge tied to his bear-market and market-structure Foundations pieces. |
Full summary 616 words
This post launches Portpourri, a new informal format named after the classifieds paper Burry delivered as a kid in San Jose. It is a quick rundown of what is moving in his portfolio plus short takes on the news, sitting alongside his Trading Posts, Short Thoughts and feature articles. It follows the October 2026 Trading Posts in which he added to MercadoLibre and cut parts of his put book.
He opens with position sizing. His two largest holdings are Temple & Webster (TPW AU / TPLWF) and MercadoLibre, each over 10% of the portfolio and both mostly written up for upcoming pieces. He added to MELI on the day and a few sessions earlier, both times under $1,500. He likes it from the $1,400s for the long term, warns it may go lower, and says he will buy again if it falls another 20%. Temple & Webster, he says, made a high-volume bottom and has turned. Next in size are Birkenstock, JD.com, Molina, HCA, Fiserv, Deckers and Adobe. He added to Adobe and JD on the day. Adobe recently forced perpetual-license users like him onto the cloud product, and he reads that as the company playing hardball. With buybacks at low prices he is content for the stock to sit in the low $200s. JD he calls forgotten and far too cheap.
On consumer retail, he has not swapped Deckers back into lululemon and may simply add LULU at a good moment while keeping DECK. Both stocks have slumped toward the high double digits, but both earn plenty relative to valuation and will buy back stock. He acknowledges a concentration in LULU, DECK, BIRK, Build-A-Bear and the Sprouts call LEAPs (added to on the day), and says this limits his appetite for more in the space. LVMH stays on his radar, but he has no room for it at its current price. He still owns Veeva after a big run and has closed the calls he had sold against it. The short side is a basket of puts on SMH, the Nasdaq 100, Caterpillar, Oracle, MetLife and Palantir. His Hong Kong basket is unchanged: Tencent, Meituan, Haier Smart Home, Haidilao and BYD at a little over 2% each, plus JD. All lean on a China consumer that has been sticky, but the companies keep making money and taking share.
He points subscribers back to recent essays as the framework for current markets. These are the two Foundations pieces on bear-market performance and on busted S&P 500 stocks and the Y2K analogue, the GPU depreciation and useful-lives article, the July market-structure and volatility-targeting piece, and Part V of the Heretic's Guide on hyperscaler capex.
On Nvidia, he says the news about the CEO surprised him and so did how hard the market turned. He does not think the scandal itself will badly hurt the business economics, but he still expects Nvidia's competitive advantage to shrink over five years. The risk is that the affair revives scrutiny of advanced-GPU trade into China and the Justice Department antitrust probe. He describes Nvidia's practice of investing in, and setting requirements for, AI ecosystem players as mafia-like and ripe for prosecution. He is not tempted as a long-term holder but expects the market to see it as a trade soon.
On Fannie Mae and Freddie Mac, he recalls swapping the common for junior preferreds a little over a month ago. He believes the midterm results leave Trump little room in his last two years, and settling the GSE saga is one of the few things he can and will do. That does not yet require switching back to the common, but he is monitoring it and promises to report any move.
Key metrics
- TPW AU and MELI each over 10% of the portfolio
- MELI adds made twice, both under $1,500; he likes it from the $1,400s
- Would buy MELI again on a further 20% decline
- ADBE: comfortable with the stock in the low $200s while buybacks run
- LULU and DECK have fallen toward the high double digits
- Hong Kong basket names each a little over 2% of the portfolio
- Put basket: SMH, NDX, CAT, ORCL, MET, PLTR
- Fannie/Freddie common swapped into junior preferreds a bit over a month ago
Key insights
- Two names, Temple & Webster and MercadoLibre, are each over 10% of the book. That is concentration on conviction, not diversification.
- MELI is being averaged down below $1,500 with a stated plan to buy again after another 20% drop. He expects volatility and is sizing for it.
- Adobe forcing perpetual-license users to the cloud is, to Burry, pricing power. He accepts a low-$200s stock while buybacks shrink the share count.
- He openly flags the concentration in consumer retail (LULU, DECK, BIRK, BBW, SFM LEAPs) and says it caps further adds there. LVMH is wanted but squeezed out on price.
- The short side is entirely puts on SMH, NDX, CAT, ORCL, MET and PLTR, a defined-risk hedge against the AI and capex bubble he has written about.
- He frames Nvidia's CEO scandal as a possible catalyst for China-export and antitrust enforcement that could speed up moat erosion. He calls it a likely trade but not a long-term hold.
- He reads the midterm results as making a Fannie/Freddie resolution one of Trump's few achievable goals. A switch from the junior preferreds back to the common is the next move to watch.
- The Hong Kong basket of about 2% positions (Tencent, Meituan, Haier, Haidilao, BYD) plus JD is unchanged. It is a patient bet that China-consumer earnings and share gains eventually get rewarded.
#062
Trading Post
Nov 06, 2026
Trading Post November 6, 2026
Burry's Caterpillar puts, a bet that the AI data-center buildout's pick-and-shovel beneficiary was over-extended, have roughly quadrupled. Rather than let a winning option position grow into an outsized risk, he sold three-quarters of it. What remains is worth about what he originally paid, so he keeps exposure to further downside in CAT while the cost of being wrong from here is effectively covered by realized gains.
energygrid
Positions & performance
-2.2%avg return0% win · 1 tracked · S -2.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SCAT:NYSE energygrid Caterpillar TRIMputs · took the puts down 75% | S&PBitget | IAPT | 558.63closeNov 06 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | -2.2% |
| WhyThe Caterpillar puts had roughly quadrupled since he put them on at the end of Q2, so he sold 75% of them. That takes the position's market value back to roughly the same dollar amount he originally committed: he has taken his principal-plus-gains off the table and left the rest running at no net added risk. |
Full summary 179 words
A very brief Trading Post ("A quick one") with a single action. Burry reports that his Caterpillar puts have roughly quadrupled in value, and on November 6 he reduced the position by 75%.
He frames the trim in dollar terms: after selling three-quarters, the market value of what remains is roughly the same number of dollars as when he put the position on at the end of the second quarter. In effect he has banked the gains and is now playing with the market's money on the remaining quarter of the puts.
Caterpillar has been a recurring short in his book. Earlier posts first shorted the common at the June 30 quarter-end (on a 200-day moving-average extension signal) and added through the summer, arguing that data center plans were "already aging". By October the exposure was being carried as long-dated, well out-of-the-money December 2027 puts, which he said on November 5 he intended to keep holding. This post gives no new fundamental view on Caterpillar; it is a risk-management step after a large gain, not a change of thesis.
Key metrics
- Caterpillar puts up roughly 4x since inception
- 75% of the put position sold on 2026-11-06
- Remaining position's market value roughly equal to the original dollar cost
- Position originally put on at Q2 2026 quarter-end
Key insights
- The CAT puts roughly quadrupled from the Q2-end entry, one of the clearest winners in his AI-capex-linked short book.
- He sold 75% of the puts, so the remaining position is worth about his original dollar outlay: a classic 'take the principal off the table' trim.
- The bearish view on Caterpillar is not abandoned; a quarter of the puts remain in place.
- The trim comes one day after he listed CAT among put positions he intended to keep, showing he is willing to monetize quickly once gains get large.
- No new price, strike or fundamental rationale is given in this post; the earlier reasoning (data center plans aging, extended chart) still stands.
#061
Trading Post
Nov 05, 2026
Trading Post November 5, 2026: MercadoLibre & Semiconductors & More
MercadoLibre's latest miss comes from deliberate investment, not a weaker business. It grew logistics capacity 20% ahead of peak season, kept spending on affiliates, and absorbed some fuel costs, while its credit book grew 62% with stable delinquencies. Burry thinks the market is wrong to punish the stock, so with the price below his $1,433-ish IV15 zone, he increased the long by 50%. On the other side of the book, the day's market move made his semiconductor puts very large, so he took a quarter of the SMH puts off but kept the bearish AI/tech hedge in SMH, QQQ, PLTR, ORCL and CAT.
aineocloudairobotaisemienergygridindexussaas
Positions & performance
-3.2%avg return0% win · 6 tracked · L -4.5% · S -3.0%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SQQQ:NASDAQ indexus Invesco QQQ Trust HOLDputs | NASDAQBitget | SC | 647.05closeNov 05 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | -2.4% |
| WhyIndex-level Nasdaq put he intends to keep holding with the SMH puts, keeping the short book focused on the AI/tech trade. No new reasoning is given in this post. |
| SPLTR:NASDAQ saas Palantir Technologies HOLDputs | NASDAQS&PBitget | ISPTC | 154.99closeNov 05 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | -2.9% |
| WhyLong-running put on a richly valued AI software name that he intends to keep holding. No new reasoning is given in this post. |
| SSMH:NASDAQ aisemi VanEck Semiconductor ETF TRIMputs · sold 25% of the SMH put positions | NASDAQBitget | I | 447.73closeNov 05 '26 | 461.68 ● | 315.05671.83 | +21.9%APY +643.5% | -3.1% |
| WhyTook some profit after the day's market move had made the semiconductor puts very large positions. Even after the sale they remain very large, and he intends to keep holding the rest. |
| SCAT:NYSE energygrid Caterpillar HOLDputs | S&PBitget | IAPT | 553.64closeNov 05 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | -3.1% |
| WhyListed among the other put positions he intends to keep holding alongside the remaining SMH puts. No new reasoning is given in this post. |
| SORCL:NYSE aineocloud Oracle HOLDputs | S&PBitget | ISAPTC | 115.16closeNov 05 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | -3.4% |
| WhyPut on Oracle's debt-funded AI data-center buildout that he intends to keep holding. No new reasoning is given in this post. |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon · increased position by 50% | NASDAQBitget | ISAPC | 1530.28closeNov 05 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -4.5% |
| WhyBought under $1,500/share after the post-earnings drop; his adjusted IV15 is $1,433 and he had been waiting for sub-$1,500. He sees the margin squeeze as deliberate investment (a 20% logistics capacity expansion ahead of peak season, affiliate spend) behind durable moats, and the credit book as disciplined with 15-90 day NPLs stable at 7%. |
Full summary 399 words
This is a short, quickly written Trading Post that Burry says he wanted to get out because of the day's market moves. It covers two things: adding to MercadoLibre after its third-quarter earnings and trimming his large semiconductor put position.
On MercadoLibre, Burry says the results do not concern him. The company has disappointed on the bottom line for several quarters, but he believes it is building businesses with moats that will protect it for a long time. His adjusted IV15 (intrinsic value estimate) after earnings is $1,433, and he had been waiting for the stock to fall below $1,500. He wonders what other analysts are seeing, because almost everything the company is doing makes sense to him.
To support this, he quotes two executives from the earnings call. CFO Martin de los Santos breaks down roughly 60 basis points of shipping margin compression: 40bp in COGS and 20bp in OpEx. Most of it came from expanding the logistics footprint, which MercadoLibre does every third quarter ahead of the fourth-quarter peak. Capacity grew 20% quarter-on-quarter this year, versus 10% a year earlier, and part of that cost should come back in Q4. Fuel and energy costs were only partly passed on to customers. A further 40bp went into affiliate marketing, though sales and marketing stays within the 11-12% of revenue range of recent years. Fintech President Osvaldo Gimenez says Brazil's 15-90 day NPL ratio was 7%, stable quarter-on-quarter and year-on-year, while the credit book grew 62%. He adds that Argentine NPLs are near historical lows and that marketplace data on how users buy and pay lets Mercado Pago underwrite cheaply and pull back quickly if behavior worsens. Burry concludes that even the finance business is fine if it is run with eyes wide open, as it seems to be.
On positioning, he increased his MercadoLibre position by 50% at under $1,500 per share, showing the five-year chart against price/sales. He sold 25% of his VanEck Semiconductor ETF (SMH) puts, noting that even after the sale they have become very large positions. He intends to keep the rest of those puts and his other puts in Caterpillar, the Invesco QQQ Trust, Palantir and Oracle. The bearish view on AI-linked equities is unchanged. The trim manages position size after a sharp move. The MELI add is a valuation buy, and a sustained break in credit quality or margins would undercut it.
Key metrics
- MELI adjusted IV15 $1,433 post-earnings; he had been waiting for sub-$1,500
- MELI position increased 50% at under $1,500/share
- MELI shipping margin compression ~60bp q/q (40bp COGS, 20bp OpEx)
- MELI logistics capacity +20% q/q this year vs +10% a year ago
- MELI affiliate investment +40bp vs last quarter; sales & marketing steady at 11-12% of revenue
- Mercado Pago Brazil 15-90 day NPL 7%, flat q/q and y/y, with credit book growing 62%
- Sold 25% of SMH put positions; the remainder is still very large
Key insights
- Burry added 50% to MercadoLibre under $1,500/share; his adjusted post-earnings IV15 is $1,433, so he is buying close to his intrinsic value estimate.
- He treats the MELI margin miss as a timing issue: about 60bp of shipping compression came from a 20% q/q logistics capacity build ahead of Q4 peak season, double last year's 10%, and the CFO expects part of it to reverse in Q4.
- Fuel and energy cost pressure was only partly passed to consumers and management expects it to persist. That is the one margin headwind not framed as temporary.
- Mercado Pago's Brazil 15-90 day NPL held at 7% (flat q/q and y/y) while credit grew 62%, which supports Burry's view that the lending arm is run with eyes wide open.
- Management says the marketplace gives it data on how users buy, pay and use installments before it lends, so it can tighten by product, cohort, credit line or duration.
- He sold 25% of his SMH puts because the market move had made them very large positions, but they remain very large after the trim.
- The bearish AI/tech book is intact: he plans to keep the puts in CAT, QQQ, PLTR and ORCL along with the remaining SMH puts.
#060
Short Thoughts
Nov 04, 2026
Anthropic’s Amended Prospectus Form S-1/A: 3rd Quarter Results & More Questions
Anthropic's amended S-1 confirms that its compute and infrastructure commitments are far larger than the named contracts suggested: $895.4 billion filed, and $1.05-1.2 trillion once unpriced TPU/AMD capacity and SpaceXAI are counted, which is 13-25x revenue and heavily front-loaded. Third-quarter growth slowed sharply to 46.7% as tokenmaxxing faded, while compute costs jumped to 58% of revenue because Anthropic had to buy expensive short-term capacity. Its own Anthropic Direct buildout turns the company into a non-cancellable take-or-pay customer of itself, funded partly by hyperscaler and Broadcom vendor financing that rests on leveraged insurance-company structures. Supranormal growth is therefore mandatory just to break even, and Burry thinks the market will only see the risk clearly in hindsight.
aisemifinancefintechmag5
Positions & performance
Full summary 539 words
This Short Thoughts note follows Burry's October 27 teardown of Anthropic's S-1. In that piece he hoped the amended filing, due before the IPO roadshow, would show third-quarter results and the October revenue run-rate. The S-1/A, filed November 2, includes the quarter, but the October run-rate is left blank.
The headline is commitments. Burry had added up about $537 billion of named contracts. Anthropic now discloses "an aggregate committed spend of approximately $895.4 billion": $320 billion of contractual obligations, $17.4 billion of short-term commitments, $451.2 billion of leases not yet commenced (up $130 billion in the quarter, averaging ~10 years), and $106.8 billion signed after September 30. The named counterparties are SpaceXAI up to $120.5 billion, Google $121.3 billion (up from $114.5 billion), Amazon $110 billion, Microsoft $44.9 billion, Broadcom $161.2 billion of rent, Nscale $44.6 billion and Lambda $41.3 billion. On top of that, $150-175 billion of Google TPU and AMD capacity sits outside the total, which takes it to $1.05-1.07 trillion. That is 25x trailing revenue of $35.3 billion and 13x the Q3 annualized $67.8 billion, with ~$447.5 billion due before 2031. He argues the SpaceXAI contracts, which are cancellable on 90 days' notice from 2027 and worth ~$3.1 billion a month at full capacity in December, probably are not fully in the $69.8 billion next-12-month figure. If they were, near-term obligations could reach ~$100.8 billion and the total ~$1.2 trillion.
He then turns to Anthropic Direct, the company's own leased data centers and directly procured accelerators. With $132 billion of cash, Anthropic is effectively becoming a hyperscaler, and it warns the buildout will be a significant use of cash "for the foreseeable future," which undercuts bulls pointing to positive operating cash flow. Burry describes it as a non-cancellable take-or-pay deal with itself as the only customer, carrying five-year Broadcom rack leases, data center leases of up to 20 years, and obsolescence risk.
Third-quarter revenue was $16.956 billion, up 46.7% sequentially, close to his ~45% estimate and down from 132% and 144% in the first two "tokenmaxxing" quarters. Compute expense rose 77.5% to 58% of revenue, and revenue per compute dollar fell from $2.08 to $1.72, the first decline on record, because of costly short-term compute that the filing says will keep growing and squeeze margins. He also flags several items. Amazon's convertible notes are marked at $130.5 billion ($70.91 per share vs the $58.90 Series H), which produces large non-cash losses; his estimate puts Amazon at ~20.6% fully converted against a 12% cap. There is $77 billion of "funding support" from Google, Amazon and Broadcom, who take Anthropic securities as payment. Channel-partner revenue share fell from 48% to ~40% while sales and marketing expense dropped from 22% to 14% of revenue, after what the filing calls changed partner terms, which he sees as possible IPO window dressing. Finally, Broadcom's financings run through Apollo's Athene and Blackstone-linked insurers in leverage-on-leverage structures; he says "this time the bubble's patsies are the insurance companies."
His model puts payments at $101/$112/$127 billion for 2027-29 as filed, or $120/$174/$202 billion with SpaceXAI and TPU. Just breaking even requires 11.6% monthly revenue growth through 2027, and a 20% GAAP margin requires 16.1% a month. He has no plans to buy Anthropic shares.
Key metrics
- Aggregate committed spend ~$895.4B per S-1/A (vs Burry's prior $537B tally of named contracts)
- Plus $150-175B of unpriced Google TPU (2028-31) and AMD capacity = $1.05-1.07T; possibly $1.17-1.19T including SpaceXAI
- Leases not yet commenced +$130B in Q3 to $451.2B
- Commitments 25x TTM revenue of $35.3B; 13x Q3 annualized revenue of $67.8B
- Q3 revenue $16,956M, +46.7% q/q vs +132% and +144% in Q1/Q2
- Compute & infrastructure expense +77.5% q/q to 58% of revenue (from 48%); revenue per compute dollar $2.08 to $1.72
- Amazon convertible notes carried at $130.5B ($70.91/share vs $58.90 Series H); est. 20.6% fully converted vs 12% cap
- $77.0B partner financing capacity: Google $20B, Amazon $15B, Broadcom $42B
- Break-even requires 11.6% monthly revenue growth through 2027; 16.1%/month for a 20% GAAP margin
Key insights
- The S-1/A discloses ~$895.4B of committed spend, 67% above Burry's $537B tally of named contracts; adding the $150-175B of TPU/AMD capacity gives $1.05-1.07T, about 25x trailing revenue.
- SpaceXAI's up-to-$120.5B contract (~$3.1B a month at full capacity in December, cancellable on 90 days' notice from 2027) appears largely excluded from the $69.8B next-12-month figure; including it could lift near-term obligations to ~$100.8B.
- Q3 revenue growth slowed to 46.7% q/q from 132-144% in H1 as tokenmaxxing faded, and the October run-rate was left blank in the filing.
- Unit economics worsened: compute expense reached 58% of revenue and revenue per compute dollar fell to $1.72 from $2.08, the first decline disclosed, with the filing guiding to more costly short-term compute.
- Amazon's convertible notes are marked at $70.91 per share (+20% vs the $58.90 Series H), creating non-cash losses; Burry estimates Amazon would hold ~20.6% fully converted against a 12% cap, so expect earnings noise and dilution after the IPO.
- Sales and marketing expense fell from 22% to 14% of revenue partly on 'changes in the terms of our channel partner distribution arrangements', which Burry flags as possible pre-IPO window dressing helped by hyperscaler partners.
- Broadcom's rack financing runs through Apollo/Athene and likely Blackstone-affiliated insurers with layered guarantees, so the AI buildout depends on leveraged insurance balance sheets.
- Breaking even on the commitments requires ~11.6% monthly revenue growth through 2027 (16.1% for a 20% GAAP margin); Burry will not buy Anthropic shares.
#059
Trading Post
Nov 02, 2026
Trading Post November 2, 2026: Global Payments Buyout Report Juices Payments Stocks, The Retail Trade, Hong Kong Calling Once Again
A reported Clayton Dubilier & Rice take-private of Global Payments, backed by 15% holder GTCR, puts a financial-buyer price on legacy payment processors, and because financial buyers pay less than strategics, that price is a usable valuation floor for the comparables. Applying the buyout math, Global Payments itself is already at roughly twice its control-premium IV15 and FIS remains far above its implied takeout value, but Fiserv at $50 sits near its control-comp price and stays a full Castle position, while PayPal at 0.80x IV15 is a Fat Pitch. Separately, record-low consumer sentiment coexisting with a calm bull market has turned high-quality discretionary retailers into reliable pod-shop shorts, which is why he keeps accumulating best-of-breed retail rejects (Sprouts LEAPs added) on the 2000-2005 bombed-out-retail playbook. In Hong Kong, JD.com trading at half IV15 with net cash equal to 71% of its market cap and BYD at post-2021 lows show businesses that have outperformed their stocks for years, so he added to both.
airobotconsumerhealthconsumershoppingfinancefintech
Positions & performance
+3.6%avg return50% win · 8 tracked · L +3.6%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFISV:NASDAQ financefintech Fiserv REITcommon · full position | NASDAQS&PBitget | ISAPT | 50.02his fillNov 02 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +25.0% |
| WhyApplying the assumptions implied by a CD&R buyout of Global Payments (~7x pro-forma adjusted EBITDA for a financial buyer) to Fiserv gives a control-comp price of $100-$120 per share, so at $50.02 the stock sits roughly at its control-premium IV15 and offers a 15% implied return to a control buyer. He still rates it a Castle, says it is far more profitable with better assets than Global Payments, notes significant insider buying at these lows, and shortens his show-me timeline now that PE activity has arrived in the space. |
| LJD:NASDAQ airobot JD.com ADD2029 calls · large position | NASDAQBitget | IS | 25.68closeOct 30 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | +6.7% |
| WhyJD holds $24.3 billion of net cash, about 71% of its market cap, with gross cash and marketable securities near $40 billion, more than the whole market cap, so the operating stub trades at roughly 3x owners' earnings and the stock sits at half his IV15. He says this gap between business performance and stock performance is now common in Hong Kong, where tech multiples have been compressed severely despite record profitability. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADD2028 LEAP calls struck in the low $100s | NASDAQ | IAPT | 49.7closeOct 30 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | +1.6% |
| WhyOn Friday he added to his Sprouts 2028 LEAP calls struck in the low 100s with the stock in the high $40s, right on top of his IV15. He sees the retail-rejects basket as the kind of bombed-out, understandable, staple-ish names that led the 2000-2005 post-bubble market, and notes Sprouts' high ROIC comes partly from its niche mid-size fresh-produce store format. |
| LBIRK Birkenstock HOLDcommon | — | IA | 35.54closeNov 02 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +0.2% |
| WhyHeld in the retail basket for its supra-normal returns on investment and steadfast profitability from a relatively protected market position. He has added to some basket names as they fell further. |
| LPYPL:NASDAQ financefintech PayPal HOLDcommon · small to normal sized position | NASDAQS&PBitget | ISPT | 58.78closeNov 02 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | -0.1% |
| WhyA different kind of business and not an easy comp to Global Payments, but he has it trading at about 0.80x IV15 and calls it a Fat Pitch in the $50s. He flags that, unlike Fiserv, PayPal insiders have not been buyers. |
| LBBW:NYSE consumershopping Build-A-Bear Workshop HOLDcommon | — | S | 29.54closeNov 02 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | -0.4% |
| WhyHeld in the retail basket for its one-of-one experiential economics and high returns on invested capital, following his two-part Build-A-Bear write-up. The chart shows it tracking the same consumer-sentiment path as the other basket names. |
| LDECK:NYSE consumershopping Deckers Outdoor HOLDcommon | S&P | IPT | 85.0closeNov 02 '26 | 83.46 ● | 77.20122.29 | +4.1%APY +40.1% | -1.8% |
| WhyPart of the retail-rejects basket he accumulated this year after large drawdowns, chosen as best-of-breed with real advantages, high ROIC and one-of-one economics. He frames the sell-off as consumer-sentiment driven and amplified by pod-shop momentum shorts and vol-targeting outflows. |
| LLULU:NASDAQ consumershopping lululemon athletica HOLDcommon | NASDAQS&P | ISAPT | 107.83closeNov 02 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -2.3% |
| WhyHeld in the retail basket alongside Deckers, Birkenstock and Build-A-Bear, all down and to the right for two years while tracking record-low consumer sentiment. He expects the best-of-breed names to recover as bombed-out retail did after the 2000 bubble. |
| L1211 HK BYD ADDcommon · mid-sized position | — | — | —Oct 30 '26 | — | — | — | n/a |
| WhyBought more BYD (1211 HK / BYDFF) alongside the JD add. He notes BYD and JD have both been dead money since the Hang Seng's 2021 top, with Hong Kong tech 'pushed underwater so long and so deep' despite continued solid profitability, and signals another Hong Kong stocks piece is coming. |
Full summary 715 words
This Trading Post responds to late-week Financial Times and Bloomberg reports that Clayton Dubilier & Rice, a roughly $90 billion private-equity firm, is pursuing a take-private of Global Payments together with GTCR, the company's largest holder at about 15%. The shares jumped as much as 18% and lifted the group (FIS +6.2%, Shift4 +4.7%, Fiserv +4.3%, Flywire +1.6%, PayPal +1.1%). Burry links back to his January piece on the payments giants, where he judged Global Payments to have little minority-shareholder value as a serial acquirer running a WorldCom-style adjusted-earnings playbook. At $96.84 the company carries a ~$25.6 billion market cap on top of ~$17 billion of net financial debt, with a near-$1 billion GAAP loss over the last twelve months against wildly profitable adjusted figures, and he reproduces a slide from its non-GAAP reconciliation deck to make the point.
He then builds a control-premium version of his IV15, the price at which a 15% annualized return is implied if better capital allocators ran the company. For Global Payments he gets $45-$60, so the current price is about twice that and implies only ~10% for a control buyer, and only if the sponsor fixes the acquisition record. If CD&R is anchoring on Worldpay at 9.8x EBITDA or the Issuer-business sale to FIS at 12.3x as exit comps, he expects disappointment. The methodological point is that a financial buyer underwrites cash flow and strategy change rather than synergies, so financial-buyer takeouts are a decent valuation comparison for minority holders in similar companies. GTCR, likely underwater on the Worldpay stock it received, has every reason to help the deal along.
Running the same math across the group gives a comp price of $18-$26 for FIS, which at $38.55 remains 45% to over 100% overvalued, especially given its own expensive M&A habit (it paid 12.3x for the Issuer business while CD&R is paying about 7x for all of Global Payments' $5.7 billion of pro-forma adjusted EBITDA). For Fiserv the comp price is $100-$120; at $50.02 it sits roughly at its control-premium IV15, and he keeps it a full Castle position, arguing it is far more profitable with better assets than Global Payments, that the penalty box reflects prior management's over-promising and poorly disclosed Argentine inflation gains, and that insiders have been significant buyers. Fresh M&A shortens the show-me timeline he had previously stretched. PayPal is a poor comp, but at about 0.80x IV15 it is a Fat Pitch in the $50s and he holds a small-to-normal position, noting insiders there have not bought. Shift4, Block and Toast lack the legacy processors' breadth, so the buyout says little about them.
The second section covers the retail trade. On Friday he added to his Sprouts 2028 LEAP calls struck in the low 100s, with the stock in the high $40s right on his IV15. Charts of Sprouts, Deckers, lululemon, Birkenstock and Build-A-Bear show two years of decline tracking consumer sentiment. The University of Michigan index hit 44.8 in May, the lowest since 1952, with five of the eight worst months in 46 years all in 2026, while the S&P 500 Equal Weight index is up 24% over two years and the VIX has averaged 15-17. He calls the 2.59-sigma gap between confidence and sentiment the widest on record and describes an employed-but-miserable consumer. Citing his Foundations post on market structure, he says pod shops treat these names as reliable momentum shorts while vol-targeting strategies dump them automatically. His basket is best of breed, with high ROIC and one-of-one economics, and the 2000-2005 bombed-out-retail rebound is the model.
Finally, Hong Kong. He added to his JD.com 2029 calls, a large position, and bought more BYD, a mid-sized one; both have been dead money since the Hang Seng's 2021 top. JD has $24.3 billion of net cash, about 71% of its market cap, and gross cash above the market cap, so the operating stub trades at roughly 3x owners' earnings and the stock at half his IV15. He says this disconnect is now common in Hong Kong tech despite record profitability and promises another Hong Kong piece soon. The view would be challenged if the CD&R deal collapses without a premium, if Fiserv's new management fails to rebuild credibility, or if weak sentiment erodes the retail basket's earnings rather than just its multiples.
Key metrics
- GPN rallied as much as 18% on the FT/Bloomberg report of a CD&R (~$90B AUM) take-private; other payments names: FIS +6.2%, FOUR +4.7%, FISV +4.3%, FLYW +1.6%, PYPL +1.1%
- GPN at $96.84: market cap ~$25.6B plus ~$17B net financial debt; LTM GAAP loss of nearly $1B
- GPN control-premium IV15 (CPIV15) $45-$60; current price ~2x CPIV15, implying ~10% return for a control stake
- CD&R paying ~7x GPN's $5.7B pro-forma adjusted EBITDA vs FIS paying 12.3x for GPN's Issuer business in January and Worldpay bought at 9.8x
- GTCR owns ~15% of GPN, received when GPN bought Worldpay from GTCR and FIS
- FIS comp price $18-$26 vs $38.55 stock: ~45% to 100%+ above implied takeout value
- FISV comp price $100-$120; at $50.02 roughly at its CPIV15; ~1.6x IV15 on a minority basis
- PYPL at ~0.80x IV15, Fat Pitch in the $50s
- UMich Consumer Sentiment hit 44.8 in May, an all-time low since 1952; 5 of the 8 lowest months in 46 years are in 2026
- S&P 500 Equal Weight up 24% over 2 years, VIX averaging 15-17 the last 3 months; Confidence vs Sentiment divergence at 2.59 sigma (100th percentile)
- JD net cash $24.3B (~71% of market cap); gross cash and securities ~$40B, above the market cap; operating stub ~3x owners' earnings, stock at half IV15
Key insights
- The CD&R bid is a valuation anchor, not a thesis change: financial buyers pay less than strategics, so a Global Payments take-private sets a floor for the other legacy processors, and the group's rally on the news is the market recognizing that.
- Global Payments itself is fully valued on the buyout math: its control-premium IV15 is $45-$60 versus a $96.84 stock, implying only ~10% for a control buyer even if CD&R fixes the serial-acquisition problem.
- FIS screens as the overvalued one, with a comp price of $18-$26 against $38.55, especially since it paid 12.3x EBITDA for GPN's Issuer business while CD&R pays ~7x for all of Global Payments.
- Fiserv is the opportunity: a $100-$120 control-comp price, a $50.02 stock near its CPIV15, better assets and profitability than GPN, insider buying, and PE activity that shortens the timeline he had previously lengthened. It stays a full Castle position.
- PayPal is not a GPN comp but trades at ~0.80x IV15 and is a Fat Pitch in the $50s, with the caveat that insiders have not bought; Shift4, Block and Toast get little read-through.
- The retail basket is a market-structure trade as much as a value trade: record-low UMich sentiment (44.8 in May) during a calm bull market has made quality discretionary names reliable pod-shop shorts, which is the setup he is buying into, adding Sprouts 2028 LEAPs at an IV15 price.
- JD.com's net cash is 71% of its market cap and gross cash exceeds it, leaving the operating stub at ~3x owners' earnings and the stock at half IV15; he added to 2029 calls and bought more BYD, with a Hong Kong stocks piece coming.
- Trade log this post: added SFM 2028 LEAP calls (strikes in the low 100s), added JD 2029 calls (large position), bought more BYD (mid-sized); reiterated full FISV and small-to-normal PYPL; holds DECK, BIRK, LULU, BBW.
#058
Single Stock
Oct 29, 2026
Build-A-Bear (BBW) Part II: The Scion Capital Analyst Template and the All Map
Build-A-Bear is a one-of-a-kind experiential retailer whose store economics were rebuilt over a decade (short leases, less mall exposure, capital-light partner and concourse formats, ~25% store margins), yet the market prices it as a fading mall dinosaur. Two guidance cuts, tariffs, the loss of a Walmart wholesale program, an AI-search hit to e-commerce traffic and a CEO transition have pushed the stock from $75 to ~$25, about 7x earnings and under 4x EBITDA. Because the balance sheet is debt-free and management keeps retiring shares, the company itself absorbs the selling, with a buyback authorization equal to ~13% of the market cap. With 25% of the float short and catalysts ahead (holiday season, 50 new locations, 30th anniversary, international expansion, possible CEO buying), any good news could re-rate the stock sharply toward the $40s, while a 10x EBITDA buyout implies $60-$65.
airobotconsumerhealthconsumershoppingfinancefintechhealthmag5saas
Positions & performance
+5.3%avg return80% win · 5 tracked · L +5.3%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LBBW:NYSE consumershopping Build-A-Bear Workshop REITcommon · rates it a full 6% position in the low-mid 20s, but sized below that because of his other consumer names | — | S | 25.0his fillOct 29 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | +17.7% |
| WhyAt ~7x FY26 EPS (~$3.50), <4x EV/EBITDA, <2x tangible book, a 3.6% yield and a buyback authorization worth ~13% of the market cap, he calls it a fat pitch versus an IV15 of $23-$32. Store economics were rebuilt over a decade, and he sees the current slump (tariffs, lost Walmart program, AI-search hit to e-commerce, CEO transition) as temporary, with 25% of the float short. |
| LBIRK Birkenstock HOLDcommon | — | IA | 33.39closeOct 29 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +6.6% |
| WhyAnother leg of the countercyclical consumer basket; he caps the basket at 25-33% of the portfolio because these names will correlate, and he is somewhat below that now including BBW. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market HOLDLEAP calls | NASDAQ | IAPT | 48.0closeOct 29 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | +5.2% |
| WhyHeld through LEAPs as part of the consumer basket; he updated its IV15 and All Map position after its after-hours earnings move. |
| LDECK:NYSE consumershopping Deckers Outdoor HOLDcommon | S&P | IPT | 82.76closeOct 29 '26 | 83.46 ● | 77.20122.29 | +4.1%APY +40.1% | +0.8% |
| WhyPart of his countercyclical consumer basket; he is playing beaten-down discretionary names on value, expecting them to rebound the way bombed-out retail did from 2000-2005. |
| LLULU:NASDAQ consumershopping lululemon athletica HOLDcommon | NASDAQS&P | ISAPT | 109.82closeOct 29 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -4.0% |
| WhyListed alongside DECK in his consumer-recovery basket; he notes LULU traded near 20x its shares outstanding during its fall, a sign of the washout in out-of-favor retail. |
Full summary 641 words
In Part II of his Build-A-Bear (BBW) coverage, Burry runs the stock through the Scion Capital Analyst Template he wrote in 2006 to get his analysts working the same way (he jokes they ignored it). He calls BBW a mall dinosaur that has survived the GFC, smartphones and now AI search, and which once again faces a weak consumer trading down on discretionary purchases. The stock fell from over $75 last year to about $25. It is a micro-cap under $300 million with no debt, roughly $80 million of EBITDA, 25% of the float short and almost 9 days to cover.
The business notes trace the turnaround. In 2012, 20% of stores were unprofitable and 85% sat in malls. Management shortened the lease book (over 70% under 3 years by 2022), cut mall exposure to 65%, refreshed stores into the Discovery format and added funded partners such as Carnival, Landry's and the Girl Scouts. Wholesale locations grew from 60 to 177. A 2023 shareholder call showed corporate stores earning about $1 million of revenue at a 25% margin on a $400-500K build, a two-year payback, with concourse stores paying back in a year. Maintenance capex is under $10 million against $80 million of EBITDA. New CEO Chris Hurt, an 11-year insider, built wholesale and licensing and lifted gross margin from 45% to 55%. Burry describes the moat as the in-store ritual itself (building, naming, a birth certificate), noting that 40% of buyers are teens and adults and a third of revenue is tied to birthdays.
The bad news is plain. In Q2, revenue fell 7.2% to $115 million, comps fell 9.5%, gross margin dropped 335 bps to 54.2%, e-commerce demand fell 15.6%, and contribution margins fell in all segments. FY26 revenue guidance was cut to $500-525 million and GAAP pre-tax to $60-68 million, and the Walmart program did not repeat. The Chief Growth Officer was terminated 11 weeks after promotion. Still, he finds the accounting clean: low stock comp, no options, no goodwill, a full tax rate, a $13.2 million tariff refund received and a share count down 25% since 2021. Governance checks out too. The board withheld 80% of target bonuses in 2023-2025, the CEO must own 5x salary in stock, and Cannell Capital's long activism instilled a buyback culture. Buybacks plus dividends of $154 million over 2022-2025 exceed half the current market cap, and tangible book per share tripled to $13.02.
On valuation, BBW trades at ~7x EPS, under 4x EBITDA and under 2x tangible book with a 3.6% yield, against toy wholesalers at up to 15x and Pop Mart above 30x. His IV15 range is $23-$32, implying about a 14% compounded return. A 10x EBITDA buyout would be worth $60-$65. The stock has traded 10x its shares outstanding since the high, with a volume-weighted close of $42.13, so a rally could run toward $40 before meeting overhead supply. With heavy short interest, a 50% move could come fast on positive news. Catalysts include the holiday season, 50 new locations, the 30th anniversary, a UK McDonald's Happy Meal tie-in, an Orlando flagship in early 2027, international expansion (85% of revenue is still North American) and eventual open-market CEO buying.
On positioning, he rates BBW a full 6% position in the low-to-mid $20s but keeps it smaller because he also holds DECK, LULU, SFM (through LEAPs) and BIRK, which will correlate. He caps that consumer basket at 25-33% of the portfolio and is a bit below it now. He is watching LVMH but says it is not there yet. He expects a couple more heavy quarters and a weak consumer, with UMich sentiment below GFC levels. His model is 2000-2005, when bombed-out retail names were top performers. He closes by posting an updated All Map with new IV15s after earnings from GOOG, MSFT, SFM, NOW, MOH, JD, PYPL and ADBE.
Key metrics
- BBW ~$25 vs $75.85 closing high (Sep 15, 2025) and $0.89 COVID low
- Market cap under $300M, no financial debt, EBITDA ~$80M, FY26 EPS est ~$3.50
- ~7x EPS, <4x EV/EBITDA, <2x tangible book, 0.6x sales, 3.6% dividend yield
- IV15 range $23-$32; implied 15-year compounded return ~14% at current price
- Buyout at 10x EBITDA ≈ $60-$65/share
- Short interest 25% of float, ~9-9.5 days to cover
- Q2: revenue -7.2% to $115M, comps -9.5%, gross margin -335 bps to 54.2%, pre-tax income -24.1%
- FY26 revenue guide cut to $500-525M from $530-550M; GAAP pre-tax $60-68M from $72-78M
- Share count down 25% since 2021; $109M buybacks + $45M dividends 2022-2025; tangible book/share $4.49 → $13.02
- Volume-weighted close since the 2025 high: $42.13; stock has traded 10x shares outstanding since then
Key insights
- Store-level economics changed: ~$1M revenue per corporate store at a 25% margin on a $400-500K build (two-year payback), concourse stores with one-year payback, and maintenance capex under $10M against ~$80M EBITDA.
- Capital return is unusually large for the size: $109M of buybacks plus $45M of dividends over 2022-2025 exceed half the current market cap, and the remaining $40.5M authorization is ~13% of it.
- Tangible book per share tripled from $4.49 to $13.02, which Burry reads as proof that buybacks were done below intrinsic value and that the ROE is real.
- The sell-off is driven by things he sees as temporary: Liberation Day tariffs on a ~50% China / 40% Vietnam supply chain, lapping a non-repeating Walmart program, AI search cutting e-commerce traffic (-15.6%), and a CEO transition.
- The short setup is crowded: 25% of float short, ~9.5 days to cover, 10x shares outstanding traded since the $75.85 high, and a $42.13 volume-weighted close that leaves room up to ~$40.
- Valuation gap versus peers: ~7x earnings versus toy wholesalers up to 15x and Pop Mart over 30x, and a 10x EBITDA buyout would imply $60-$65 per share.
- Watch item: new CEO Hurt has not bought stock. Burry wants to see open-market buying and warns of a possible kitchen-sink quarter.
- Portfolio construction: he rates BBW a 6% position but caps his correlated consumer basket (DECK, LULU, SFM LEAPs, BIRK, BBW) at 25-33% of the portfolio, and he is watching LVMH.
#057
Short Thoughts
Oct 27, 2026
Short Thoughts: Anthropic’s S-1
Anthropic's S-1 shows roughly $537 billion of compute commitments — over 20x trailing sales once the pending TPU and SpaceXAI deals are included — much of it take-or-pay with the same hyperscalers that lend to it, own it, sell its product and collect half its revenue. At the same time revenue growth has decelerated sharply (about 45% q/q in Q3 and 3-7% a month in August-September) and revenue per compute dollar fell for the first time. Because the money that pays the clouds partly originates with the clouds, and Broadcom's rack leases are being carved into structured credit sold to insurers like Athene, the arrangement looks like the vendor financing of the 1998-2002 telecom buildout. Cross-default clauses mean a shortfall at Anthropic could accelerate substantially all of its debt, leaving the whole interconnected system catastrophically exposed on a pause or a downturn.
aisemifinancefintechmag5
Positions & performance
Full summary 540 words
Burry's short note on Anthropic's S-1 skips the line-by-line teardown others have already done and instead frames the filing within the theme he has developed across Cassandra Unchained for a year: the amount of risk built into the AI buildout through circular, vendor-financed arrangements. He presents a new diagram, in the style of his earlier Apollo AMAP visual, mapping the revenue and capital loops between Anthropic and its partners.
His headline is the commitment stack. Anthropic has committed about $537 billion to compute — Google $114.5 billion, Amazon $110 billion, Microsoft $44.9 billion, Broadcom $161.2 billion of rack rent, Lambda $41.3 billion, Nscale up to $44.6 billion and AMD more than $20 billion — plus $125-150 billion of not-yet-final Google TPU capacity for 2028-2031 and up to $120.5 billion of GPU capacity from SpaceXAI. Including those, commitments exceed 20 times trailing sales and about 12 times the current run-rate, a figure he says the S-1 buries in the notes. Against this, revenue growth is slowing: 132% and 144% sequential growth in Q1 and Q2 fell to about 45% in Q3 (roughly $16.9 billion), revenue per compute dollar dropped from $2.08 to $1.70, and August-September growth was only 3-7% a month, well short of the 11-20% monthly pace needed to hit the $100-120 billion year-end run-rate investors reportedly expect. Many contracts are take-or-pay, so Anthropic owes the full amounts regardless of usage — the S-1 notes it had sought more compute than the clouds would supply.
He then traces the circularity. Amazon lent $8 billion in convertible notes and bought $10 billion of preferred; its remaining $6.2 billion of notes convert into 1.84 billion shares at $2.09 and $5.05, carried at $108 billion, which drove most of the $80.7 billion first-half net loss. Google put in $3.3 billion of notes and $20 billion of preferred, with $20 billion more of notes undrawn. Every dollar comes back as compute purchases ($110 billion to Amazon against ~$18 billion invested; $114.5 billion to Google against ~$23 billion). The hyperscalers also bill and collect 48% of Anthropic's revenue and account for 58% of receivables, keeping an undisclosed fee. Broadcom leases Google-designed TPU racks to Anthropic SPVs under finance leases that become debt only when put in service, can take up to $42 billion of rent in 9% PIK convertibles that convert at a discount (worse terms if revenue targets are missed), and has moved the assets off its balance sheet via an Apollo-led $35 billion program with Blackstone, tranched into structured credit of the type Athene buys, followed by a $63 billion second package. Neoclouds Lambda and Nscale borrow against Anthropic's name, and AMD will buy up to $5 billion of Anthropic stock as its chips deploy.
The risk, he argues, is in the S-1's own words: a default under certain commercial agreements could cross-default the credit agreement and convertible notes and accelerate substantially all debt. He likens the setup to telecom vendor financing in 1998-2002 and 1960s computer leasing. The note states no new trades; it reinforces his skepticism of AI-infrastructure leaders (hyperscalers, Broadcom, private-credit financiers) and points to the S-1 amendment with Q3 detail and the October run-rate, due before the roadshow, as the next catalyst for the market to notice the slowdown.
Key metrics
- ~$537B of Anthropic compute commitments: Google $114.5B, Amazon $110B, Microsoft $44.9B, Broadcom $161.2B rack rent, Lambda $41.3B, Nscale up to $44.6B, AMD >$20B
- Plus $125-150B of Google TPU capacity for 2028-2031 (not finalized) and up to $120.5B from SpaceXAI (cancelable on 90 days' notice after early 2027)
- Commitments incl. those deals = >20x trailing-twelve-month sales and ~12x current annual run-rate revenue; S-1 shows $265.6B fixed/minimum at June 30 with ~$266B added since
- Revenue grew 132% and 144% q/q in Q1 and Q2; Q3 ~$16.9B (86x the Dec-2024 quarter's $197M), up ~45% q/q
- Revenue per compute dollar fell to $1.70 from $2.08 in Q2 (implied compute ~$9.9B, +78%) — the first decline
- Aug-Sep revenue growth only 3-7% a month; September run-rate $69B (Syz) to ~$74B (SemiAnalysis) vs investor hopes of $100-120B by year-end (needs 11-20%/month)
- Amazon: $8.0B convertible notes + $10B preferred; remaining $6.2B notes convert into 1.84B Class N shares at $2.09/$5.05, carried at $108B — main driver of the $80.7B 1H26 net loss
- 48% of 1H revenue billed/collected by hyperscalers; hyperscalers = 58% of receivables; channel-partner fees +$2.7B y/y
- Broadcom notes: up to $42B of rent payable in convertibles at 9% PIK, converting at 90% of 30-day price (13% / 85% if revenue conditions missed); $2.5B cash in escrow
- ~$41B of Anthropic's $141B raised came from Amazon and Google, with $35B more of their convertible notes undrawn; Apollo-led $35B TPU program and a $63B second package (Oct 26)
Key insights
- Anthropic's ~$537B of compute commitments (more than 20x trailing sales including pending TPU and SpaceXAI deals) is, in Burry's view, the S-1's most deserved headline, yet it sits in the notes rather than being stated plainly.
- Growth is decelerating fast: ~45% q/q in Q3 after 132%/144% in Q1/Q2, and only 3-7% a month in August-September, against investor hopes requiring 11-20% a month to reach a $100-120B run-rate by year-end.
- Revenue per compute dollar fell for the first time, to $1.70 from $2.08, while implied compute spend rose ~78% to ~$9.9B — unit economics are worsening as take-or-pay obligations grow.
- Amazon and Google are simultaneously lender, shareholder, vendor and sales channel: ~$41B of Anthropic's $141B raised came from them, and it flows back as $110B and $114.5B of contracted compute; they also collect 48% of revenue and hold 58% of receivables.
- Amazon's convertible notes, struck at $2.09 and $5.05 a share and marked at $108B, explain most of the $80.7B first-half 2026 net loss.
- Broadcom's $161.2B rack-rent leases are off-balance-sheet 'debt-in-construction' until in service; up to $42B can be paid in 9% PIK converts at a 10% discount (13%/15% if revenue triggers are missed), giving the landlord more stock the worse Anthropic does.
- The TPU lease stream is being securitized through an Apollo-led $35B program with Blackstone (plus a $63B second package), tranched into structured credit that insurers such as Athene buy — mirroring the Nvidia/xAI/Apollo deal he diagrammed six months ago.
- Cross-default language means a commercial-agreement default could accelerate substantially all Anthropic debt; Burry compares the setup to 1998-2002 telecom vendor financing and flags the pre-roadshow S-1 amendment (Q3 detail, October run-rate) as the next catalyst.
#056
Macroabridged
Oct 26, 2026
[Abridged] Foundations: The Fate of the S&P 500’s Worst Stocks, the Y2K Analogue, and Time for Small Caps?
Measured by market internals, the October 6, 2026 S&P 500 record high looks like nothing in 45 years except the March 2000 dotcom top: an unusually large share of index members are already Busted, and the remaining leadership is unusually 'soft', sitting 5-10% below its highs. In 2000 the Busted stocks went on to rise while the Leaders collapsed, because liquidity rotated from expensive leaders into cheap laggards. Today the Busted are less dramatically cheap than in 2000, so Burry expects less rotation within the index and more liquidity flowing to equities outside the S&P 500 (international, U.S. small-cap value) or to fixed income. He therefore picks selectively among high-ROIC, cheap-on-owners'-earnings Busted names, adds non-index longs such as Build-A-Bear and Temple & Webster, trims stretched winners, and stays fully invested only because he carries a significant short book.
consumershopping
Positions & performance
+23.7%avg return100% win · 1 tracked · L +23.7%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LBBW:NYSE consumershopping Build-A-Bear Workshop REITcommon | — | S | 23.78closeOct 26 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | +23.7% |
| WhyHis named example of U.S. smaller-cap value: 'attractive at current levels, now below its IV15.' It fits his plan to hold investments outside the S&P 500, where he expects liquidity fleeing big-cap leaders to spill over, as small-cap value did in 1999-2007. |
| LTPW AU Temple & Webster HOLDcommon | — | — | —Oct 26 '26 | — | — | — | n/a |
| WhyCited, alongside HK Tech, as part of his international fan-out beyond the Busted S&P 500 cohort; one of the non-index longs that keep him essentially fully invested against a significant short position. |
Full summary 543 words
This is the 1,500-word free abridgement of Burry's longer Foundations piece, which follows his earlier Foundations post on how stocks perform in the next bear market. He takes 45 years of S&P 500 constituent data and, at each of 11 record-high market tops that preceded bear or near-bear markets, sorts members into three buckets: Leaders (within 10% of their rolling 10-year high), Broken (down 20-40%) and Busted (down more than 40%). The goal is to predict which long positions hold up through the next downturn.
The headline finding is that March 2000 was the only top where the Leaders' median return was negative and the Broken positive at both 12 and 18 months. Busted names rose 19% at one year and 18% at 18 months, against +11% and +2% for the Broken and -33% and -50% for the Leaders. At the October 6, 2026 record high, 23% of S&P 500 stocks were already Busted at a median 55% below their highs, close to 2000's 32% and 52%. He stresses that this is not a general rule: across all 11 tops the Busted beat the Leaders at 18 months only three times.
He then looks at how strong the leadership is. A 'Soft Share' measures how many Leaders sit 5-10% off their highs rather than 0-5%. On October 6 that figure was 58%, the highest of any top (March 2000 was next at 40%), and the median Leader was 6.1% below its high, the only top above 5% in 40 years. Scoring all 982 record closes on seven internals (median decline, Busted share, Leader share, Bear Market Ratio, number of members at closing highs, Soft Share and Fresh Drawdown Ratio), only two runs ranked at or beyond the 95th percentile on all seven: the last four months before the March 2000 top, and this year's run, which first hit 7 of 7 on May 27. He concludes that 2000 is the only top in 45 years whose internals mimic today's.
Being down does not make a stock cheap, so he screens the Busted by return on invested capital and price-to-owners' earnings. He haircuts GAAP earnings for share-based compensation using the ΔE framework from his 'AP SBC: The Tragic Algebra Recurrence' and 'D'AI of the Triffids' posts. The ranked tables are images and are not reproduced in the abridged text. On balance the 2000 Busted cohort was more numerous and much cheaper, and the 2000 Leaders more expensive, but the S&P 500 traded in the low 30s on owners' earnings in both periods.
Because today's Busted are less dramatically undervalued, he expects less rotation inside the index and more liquidity shunted to equities outside it, as happened in 2000-2002, or into fixed income. He calls it 'spearfishing': pick spots, be patient, and avoid value traps. His to-do list: consider selling, reducing or donating stretched long-held winners; keep holding non-index investments, including Temple & Webster and HK Tech internationally; and look at U.S. small-cap value, noting Build-A-Bear is below its IV15. A Bloomberg chart shows Russell 2000 Value beating Growth and the S&P 500 from 1999 to 2007. He says he is essentially fully invested only because of a significant short position, and would otherwise hold short-term Treasury money market funds if fat pitches became scarce.
Key metrics
- 23% of S&P 500 stocks were Busted (down >40% from 10-year highs) at the Oct 6, 2026 record high, median 55% below highs
- March 2000 top: 32% of S&P 500 Busted, median decline 52%
- After March 2000: Busted +19% at 12 months and +18% at 18 months vs Broken +11%/+2% and Leaders -33%/-50%
- Oct 6 Soft Share (Leaders 5-10% off highs) = 58%, highest of 11 market tops; next highest March 2000 at 40%
- Oct 6 Leaders a median 6.1% below highs — only top over 5% in 40 years (March 2000: 3.8%)
- Only 3 of 982 record closes in 45 years besides Oct 6 had median Leader drawdown >5%: Jun 30 1999, Feb 4 2021, May 29 2026
- Busted beat Leaders at 18 months in only 3 of 11 tops; lost 5 times, wash 3 times
- 7-of-7 weak-internals condition (≥95th percentile) seen only in the 4 months before March 2000 and this year's run, first on May 27, 2026
- S&P 500 valuation in the low-30s multiple of owners' earnings in both 2000 and 2026
Key insights
- At the Oct 6, 2026 S&P 500 record high, 23% of constituents were already Busted (>40% off 10-year highs) at a median -55%, close to the 32%/-52% seen at the March 2000 top.
- After March 2000, Busted stocks returned +19% (12m) and +18% (18m) while Leaders fell -33% and -50%. That is the only one of 11 tops where Leaders were negative and Broken names positive at both horizons.
- This is not a general rule: across 11 tops, Busted beat Leaders at 18 months only 3 times, lost 5 times and tied 3 times. Today's case rests on the 2000 analogy, not on a base rate.
- Leadership is historically 'soft': 58% of Leaders sit 5-10% below their highs (vs 40% in 2000) and the median Leader is 6.1% off its high, the only top above 5% in 40 years.
- Only two runs in 46 years scored at or beyond the 95th percentile on all seven weak-internal measures: the final four months before March 2000 and the 2026 run, first on May 27.
- Valuations rhyme: the S&P 500 trades in the low 30s on owners' earnings now, as in 2000, but today's Busted are less cheap, so Burry expects liquidity to leave the index (international, small-cap value, bonds) rather than rotate within it.
- Positioning: trim or donate stretched winners, hold non-index longs (Temple & Webster, HK Tech), and add U.S. small-cap value such as Build-A-Bear (below its IV15). He stays fully invested only because of a significant short book, with T-bill money funds as the fallback.
#055
Macro
Oct 24, 2026
Foundations: The Fate of the S&P 500’s Worst Stocks, the Y2K Analogue, and Time for Small Caps?
The October 6, 2026 S&P 500 record high is, on seven breadth measures across 982 record closes since 1980, matched only by the run into the March 2000 top: thin, narrow and unusually soft leadership. After 2000, the deeply Busted index members (down more than 40%) were the only cohort that rose through the bear market while the Leaders collapsed. Burry therefore expects quality Busted stocks with high ROIC and low price-to-owners' earnings to lead in the next bear market, and pairs them as directional longs against his AI and semiconductor shorts. The catch is that today's Busted names are far less cheap than in 2000, so liquidity may leak to small-cap value, international stocks or Treasuries instead of rotating within the index, which calls for selective 'spearfishing' rather than buying the whole cohort.
consumershoppingenergywarfinancefintechhealthmag5saas
Positions & performance
+9.5%avg return86% win · 7 tracked · L +9.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFISV:NASDAQ financefintech Fiserv REITcommon | NASDAQS&PBitget | ISAPT | 48.33closeOct 26 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +29.3% |
| WhyAppears in the low price-to-owners'-earnings Busted cohort with positive revenue growth and above-average ROIC. Burry cites it as a holding that exemplifies the 2000-style 'Busted leads' setup. |
| LBBW:NYSE consumershopping Build-A-Bear Workshop REITcommon | — | S | 23.78closeOct 26 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | +23.7% |
| WhyCited as his U.S. small-cap value example: 'attractive at current levels, now below its IV15.' Fits his push to hold investments outside the S&P 500, where index-clone bids keep multiples elevated. |
| LPYPL:NASDAQ financefintech PayPal REITcommon | NASDAQS&PBitget | ISPT | 54.99closeOct 26 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | +6.7% |
| WhyOn his list of cheapest Busted S&P 500 stocks by price-to-owners' earnings; a current or recent holding bought deep in its drawdown. Serves as a directional long against his AI/semiconductor short book. |
| LZTS:NYSE health Zoetis REITcommon | S&P | IAPTC | 80.97closeOct 26 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +4.8% |
| WhyListed among the lowest price-to-owners'-earnings Busted stocks; Burry holds or recently held it as a high-ROIC, growing business bought deep into its drawdown, expected to outperform if the market follows the 2000 template. |
| LDECK:NYSE consumershopping Deckers Outdoor REITcommon | S&P | IPT | 80.37closeOct 26 '26 | 83.46 ● | 77.20122.29 | +4.1%APY +40.1% | +3.8% |
| WhyOne of the Busted S&P 500 names screening cheap on owners' earnings with high ROIC; a current or recent holding purchased far below its high. |
| LLULU:NASDAQ consumershopping lululemon athletica REITcommon | NASDAQS&P | ISAPT | 102.91closeOct 26 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | +2.4% |
| WhyNamed among the S&P 500 Busted stocks (down more than 40% from highs) with the lowest price-to-owners'-earnings ratios, skewing high on ROIC with positive three-year revenue growth. Burry says it was bought deep into its drawdown and fits his view that quality Busted names lead in the next bear market. |
| LADBE:NASDAQ saas Adobe REITcommon | NASDAQS&PBitget | IPT | 258.62closeOct 26 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | -4.0% |
| WhyIncluded in the cheap, high-ROIC Busted cohort and named as a current or recent holding; contrasts with the AI and semiconductor names he is short, which he says trade with no promise of value. |
| LTPW AU Temple & Webster HOLDcommon | — | — | —Oct 24 '26 | — | — | — | n/a |
| WhyNamed as part of his international expansion beyond the Busted S&P 500 cohort (alongside HK Tech), which he will continue to hold as non-index longs against his short book. |
Full summary 526 words
This Foundations piece extends the previous week's 'Bear Market Heroes' study, which sorted S&P 500 members at each pre-bear-market record high into Leaders (within 10% of their 10-year highs) and Broken stocks. Prompted by a subscriber, Burry splits the Broken into the merely Broken (down 20-40%) and the truly Busted (down more than 40%), then tracks each group's median return at 12 and 18 months after eleven market tops over four decades, including delisted names.
The history is mixed. Busted stocks beat Leaders at 18 months after only three of eleven tops, lost five times and tied three times; in general collapses such as September 2018 to March 2020 (Busted -56%) and October 2007 (Busted -47%), a deep discount offered little protection. March 2000 is the exception: Leaders fell a median 33% and 50% at 12 and 18 months while the Busted rose 19% and 18% (24% with dividends). Today 23% of the index is Busted at a median 55% off highs, close to 2000's 32% and 52%.
Burry then examines leadership quality using Siblis Research constituent data paired with Bloomberg prices for all 982 record closes since 1980. Fifty-eight percent of October 6 Leaders sit 5-10% off their highs, the highest 'soft share' ever (March 2000 was 40%), and the median Leader is 6.1% below its high, the only top over 5%. On all seven measures (median decline, Busted share, Leaders share, Bear-Market Ratio, members at closing highs, Soft Share and Fresh Drawdown Ratio), only October 6 and March 2000 rank at or beyond the 95th percentile, and three readings have precedent only in the November 1998 to March 2000 run.
To separate cheap from merely down, he ranks Busted stocks by his own ROIC built on owners' earnings and an SBC haircut ('Tragic Algebra'); AppLovin loses 81 points of GAAP ROIC but still ranks seventh. The lowest price-to-owners'-earnings list features lululemon, PayPal, Fiserv, Deckers, Zoetis and Adobe, all current or recent holdings bought deep in their drawdowns. Early evidence is encouraging but he calls it 'almost certainly an indication of nothing': from October 6 to 19 the 114 Busted rose a median 3.2% while the index fell 1.6%.
The key difference from 2000 is valuation. The whole market trades around 32x adjusted earnings, similar to 2000, but the top ten (33x vs 50x) and Leaders (32x vs 54x) are cheaper while the Busted are richer (17.2x P/E and 2.21x sales vs 10.2x and 0.75x). Less dramatic undervaluation means liquidity may exit to equities outside the index, to fixed income, or be destroyed by a private equity, private credit and insurer unwind tied to data-center financing. Hence 'spearfishing', not shooting fish in a barrel.
Positioning: the book is not hedged but built to win on both sides in a bear market, with Busted longs against AI and semiconductor shorts. He may sell, trim or donate stretched long-term winners like Walmart, Alphabet and Valero, keeps international names (Temple & Webster, HK Tech), likes Build-A-Bear below its IV15, and points to small-cap value's 1999-2007 outperformance. He stays essentially fully invested only because of the short book; without fat pitches he would hold short-term Treasury money market funds.
Key metrics
- 23% of S&P 500 members Busted (>40% off highs) at the Oct 6, 2026 record, at a median 55% below highs; 32% Busted at a median 52% decline in March 2000
- After the March 2000 top, Busted stocks rose a median 19% at 12 months and 18% at 18 months (24% with dividends) vs Leaders -33% and -50%
- Soft Share: 58% of Oct 6 Leaders sit 5-10% off their highs vs 40% in March 2000, the next highest; 100th percentile across 982 record closes since 1980
- Oct 6 Leaders are a median 6.1% below their highs, the only bull-market top above 5% in 40 years (March 2000: 3.8%)
- Oct 6 and March 2000 are the only tops at or beyond the 95th percentile on all seven internal measures; Fresh Drawdown Ratio 26.2%, Leaders 19.3% of index, Bear-Market Ratio 54.7%
- Fully adjusted S&P 500 earnings multiple ~32x in 2026 vs ~32-34x in 2000; top 10 at 33x vs 50x, Leaders 32x vs 54x
- Busted median P/E 17.2x and P/S 2.21x today vs 10.2x and 0.75x in 2000
- AppLovin's GAAP ROIC cut 81 points (121% to 40%) for SBC, still 7th-highest true ROIC in the Busted cohort
- Oct 6-Oct 19: the 114 Busted stocks rose a median 3.2% (68% up) vs S&P 500 -1.6% and Leaders +0.3%
Key insights
- Across 982 S&P 500 record closes since 1980, only the October 6, 2026 high and the run into March 2000 rank at or beyond the 95th percentile of weakness on all seven breadth measures; the condition first read 7/7 this year on May 27.
- March 2000 was the only top where Busted stocks (down >40%) clearly won: +19% at 12 months and +18% at 18 months (24% with dividends) versus Leaders at -33% and -50%.
- Being Busted is not protection by itself: in the 2007 and 2018 tops, Busted stocks fell 47% and 56% at 18 months, as much as or more than Leaders.
- Leadership today is the softest on record: 58% of Leaders sit 5-10% below highs (100th percentile) and the median Leader is 6.1% off its high.
- Burry's cheap, high-ROIC Busted list (LULU, PYPL, FISV, DECK, ZTS, ADBE) is essentially his long book, positioned as an unhedged directional pair against AI and semiconductor shorts.
- Unlike 2000, today's Busted are not dirt cheap (17.2x P/E and 2.21x sales vs 10.2x and 0.75x), so he expects liquidity may flow to small-cap value, international stocks or Treasuries rather than rotate within the S&P 500.
- He is weighing sales or donations of stretched winners Walmart, Alphabet and Valero, and flags Build-A-Bear as a small-cap value buy below its IV15.
- AppLovin shows the SBC haircut's bite: GAAP ROIC drops 81 points from 121% to 40%, yet it still ranks seventh among Busted stocks on true ROIC.
#054
Community
Oct 22, 2026
Compression, The End of Moats
This is a short community post sharing a song, not an investment memo. The song, written by a friend of the newsletter, is narrated by an AI model that argues compression of human knowledge into model weights makes software and data-center build-outs commodities. In its telling, the only moat left for the largest tech companies is customer data locked in proprietary formats, and AI that can read any format will erode even that. The piece echoes Burry's prior skepticism of hyperscaler AI capex, but states no new positions.
Positions & performance
No single-stock positions stated in this report.
Full summary 189 words
Burry shares a song written by a reader he calls a friend of Cassandra Unchained, which he posted to his YouTube channel and embedded in the newsletter with the full lyrics. He calls it brilliant, well-made and catchy, and closes by calling it an earworm. There is no market commentary, trade log or analysis beyond that framing.
The song is told from the perspective of an AI model trained on the whole human corpus. Its argument runs: the industry built enormous data centers and bought GPUs like an arms race on the assumption that the world needs ever more compute and scale; but compressing knowledge into model weights means AI can write and deploy software in hours, so code itself stops being defensible. It singles out the concentration of the largest companies (eight firms holding about 40% of everything) and argues their remaining moat is customer data trapped in proprietary formats and switching costs, which AI that can read any format will eventually free.
The piece fits the bearish theme Burry has run on AI capex and hyperscaler valuations, but he names no tickers and changes no positions here.
Key metrics
- No trades, prices or position sizes stated in this post
- Song lyrics reference eight corporations holding roughly 40% of everything (index concentration)
- Three choruses built around the single word 'Compression'
- Song published to Burry's YouTube channel and embedded in the post
- Written by an unnamed 'Friend of CU' (Cassandra Unchained reader)
Key insights
- This is a non-trading community post: no new, added, trimmed or covered positions.
- The song's core claim is that model 'compression' turns software and AI infrastructure into commodities, so trillion-dollar capex may be solving a problem already solved.
- It names proprietary data lock-in and switching costs as the last surviving moat for mega-cap tech, and expects AI to erode it by reading any data format.
- The reference to eight corporations holding about 40% of everything points at mega-cap index concentration as the vulnerable spot.
- Burry choosing to amplify this piece is consistent with his ongoing skepticism of hyperscaler AI spending, though it adds no new data.
#053
Macroabridged
Oct 19, 2026
[Abridged] Foundations: Performance in the Next Bear Market & the Caldera Under S. Park & Wall
The S&P 500's October 6, 2026 record high is unusually narrow: 55% of its members sit more than 20% below their 10-year highs and only 15 stocks made new highs with the index, a breadth profile that matches March 2000 more than any other top in 40 years. Burry measures this with a Bear Market Ratio, a Fresh Drawdown Ratio that isolates recent breakdowns, and counts of stocks near highs. In past bear markets the broken stocks usually fell with the leaders, but after March 2000 alone the leaders collapsed while already-broken names rallied. If 2026 follows the 2000 template, concentration in today's leaders is the risk and the beaten-down majority could be the relative refuge.
Positions & performance
No single-stock positions stated in this report.
Full summary 561 words
This is the free abridged version of Burry's October 17 Foundations piece, cut from more than 6,000 words to under 1,500. It follows his prior Short Thoughts note on the AI cloud oligopoly and a flashing bear market indicator. The starting point is that the S&P 500's October 6, 2026 record close looks unremarkable next to the June 2 and August 13 highs. He also frames the backdrop with inflation's long history of eroding real returns: the index's real price fell about 53% from 1966 to 1982, 58% from 1929 to 1949 and roughly 40% from the 2000 top through 2011.
His first tool is the Bear Market Ratio (BMR), the share of S&P 500 members trading more than 20% below their 10-year highs, plotted from 1985 to 2026. At record highs that preceded bear markets or near-misses, it ran 47% in July 1990, 43% in October 2007 and 42.5% in February 2025. March 2000 stood out at 71.2%. Today's reading is 55%, up from 47% at the August high. He notes that bear markets usually drag broken stocks down too; the BMR hit 98% in 1987. 2000-2002 was the one case where broken stocks rose as the leaders crashed. Two years after that top, four in ten previously broken names were back within 20% of their highs. The leaders' median return was -30%, while the median index stock was up 21%.
The second measure, the Fresh Drawdown Ratio (FDR), counts only stocks whose 10-year high came within the last two years and that are now 20% or more below it. FDR is 27% today against 53% in March 2000, and it read 6-13% at every ordinary top since 1987. This filters out old wreckage. For example, 2007's 43% BMR drops to 10% on FDR because three-quarters of those names had peaked before late 2005, mostly holdovers from the dotcom bust. On FDR, only March 2000 and October 2026 stand out.
The "Caldera Under S. Park & Wall" plates layer each of the ten tops into bands of stocks 10%, 20%, 30%, 40%, 50% and 70% below their highs. The 1987 top was the broadest, with 64 new highs and 296 stocks within 10%. Most others had 166-284 stocks within 10% of highs and a median stock only 8-19% off. Today 406 of 503 members are more than 10% below their highs and 276 more than 20%. The median stock is 22% off, and only 15 closed at highs on October 6, with 97 within 10%. Those last two figures closely match March 2000's 15 and 87. He judges that today's caldera already matches 2000 on its top rung, while the deeper rungs are a fifth to a third short.
The synthesis is that March 2000 is the only record high after which leaders fell sharply while broken stocks rose at both 12 and 18 months. The leaders lost 33% in a year and 50% in 18 months. The abridged piece names no positions. The implication is that a narrow 2026 top leaves leadership concentration as the risk and the broken majority as the potential relative winner. He calls the parallel familiar and significant, but "thin dirt". The view would be undermined if breadth broadened, with more stocks moving back near their highs, or if a selloff pulled the broken names down alongside the leaders, as in 1987, 2020 and 2025.
Key metrics
- Bear Market Ratio (share of S&P 500 members >20% below 10-year high) 55% at the Oct 6, 2026 record close, up from 47% at the Aug 13 record
- BMR 71.2% at the March 2000 top; 47% Jul 1990, 43% Oct 2007, 42.5% Feb 2025
- Fresh Drawdown Ratio 27% today vs 53% in March 2000; 6-13% at every ordinary top since 1987
- Median S&P 500 stock 22% below its 10-year high today vs 31% in March 2000
- Oct 6, 2026: 15 stocks at new highs and 97 within 10% of one, vs 15 and 87 in March 2000; other tops ranged 166-284 within 10%
- 406 of 503 S&P 500 members more than 10% below their 10-year high; 276 more than 20% below
- After March 2000, leaders fell 33% in a year and 50% in 18 months while the median stock was up 21% two years on
- Real S&P 500 price return: -53% 1966-1982, -58% 1929-1949, about -40% 2000-2011
Key insights
- The Bear Market Ratio was 55% at the Oct 6, 2026 record close, the highest at any pre-bear-market top since 1985 except March 2000 (71.2%), and up from 47% at the Aug 13 high.
- The Fresh Drawdown Ratio isolates recent breakdowns. It reads 27% now vs 53% in 2000, and 6-13% at every ordinary top since 1987, so today is the only top besides 2000 that stands out.
- Only 15 S&P 500 stocks made new highs with the index on Oct 6 and 97 were within 10% of one, nearly identical to March 2000 (15 and 87); other tops had 166-284 within 10%.
- 2000-2002 is the only episode where leaders crashed while broken stocks rallied. Leaders lost 33% in 12 months and 50% in 18, while the median S&P stock was up 21% after two years.
- In most bear markets (1987, 2020, 2025) broken stocks fell too, with the BMR spiking to 97-98% within weeks, so the 2000 rotation outcome is not guaranteed.
- The median S&P 500 stock is already 22% below its 10-year high, against 31% at the 2000 top and 8-19% at most other tops.
- Burry flags renewed inflation as a long-run headwind, citing negative real index returns of 53% (1966-82), 58% (1929-49) and about 40% (2000-11).
#052
Macro
Oct 17, 2026
Foundations: Performance in the Next Bear Market & the Caldera Under S. Park & Wall
The S&P 500's October 6, 2026 record high came with the worst breadth of any top in 40 years except March 2000: 55% of members are in their own bear markets and fresh drawdowns sit at 27%, roughly double any ordinary top. Burry argues that breadth is an objective 'sign' of sickness that the market's calm 'symptoms' hide, driven by a mechanical, market-cap-weighted bid that shoves more money into fewer leaders. Because today's caldera most resembles 2000 (with some 1990 hangover), his base case is that the market fails in a 2000-style way - leaders drifting lower, sometimes violently, as AI-compression fundamentals surface, while the median broken stock rises. A government already stretched by repeated crisis spending is poorly positioned to rescue the coming private-debt bust, so the subsequent recovery may be long and hard to cheerlead.
Positions & performance
No single-stock positions stated in this report.
Full summary 552 words
This Foundations piece extends the prior week's Short Thoughts on the AI cloud oligopoly and a 'flashing bear market indicator.' After that post drew a strong response, Burry rebuilt the breadth study with decades of data from Siblis Research and Bloomberg to answer critics who flagged survivorship bias and the arbitrariness of a 10-year lookback. He frames the work with history: from 1966 to 1982 the inflation-adjusted S&P 500 fell about 53%, from 1929 to 1949 about 58%, and from 2000 to 2011 about 40%. Dividends soften those numbers, but the lesson is that real bear markets mean very long waits, which a 'stonks go up' and buy-the-dip generation underappreciates at a time of speculation, stretched indices and returning inflation.
He defines the Bear Market Ratio (BMR) as the share of S&P 500 members, using the contemporaneous membership list nearest each date and including delisted names, trading more than 20% below their 10-year highs. At the October 6, 2026 record close the BMR is 55%, up from 47% at the August 13 record. Ordinary pre-bear tops read 20-31%, and hangover tops (1990 at 47%, 2007 at 43%, February 2025 at 42.5%) were inflated by stocks that never recovered from earlier busts. Only March 2000, at 71.2%, is worse. To strip out those hangovers he adds a Fresh Drawdown Ratio (FDR), counting only stocks at least 20% below a 10-year high set within the prior two years. Every ordinary top since 1987 reads 6-13%; March 2000 read 53% and today reads 27%. On a total-return basis today's BMR is 48.5% versus 18-25% at ordinary tops, and the ranking holds.
He walks through ten 'Caldera Under S. Park & Wall' plates, one per top since 1985. Each shows the share of stocks 10%, 20%, 30%, 40%, 50% and 70% below their highs beneath a top layer sized to breadth. The 1987 top was the broadest (296 stocks within 10% of a high). February 1995, with a 46% BMR, preceded a historic bull only because breadth was rapidly healing: FDR fell from 34.1% to 16.9% and stocks within 10% of a high jumped from 62 to 151. Today the BMR is rising alongside the index, as in 2000. Only 15 stocks made new highs on October 6 and 97 were within 10%, versus 15 and 87 in March 2000. He concedes the BMR did not call 2008, which was a housing and financial story.
On outcomes, March 2000 is the only top after which leaders fell sharply (-33% in a year, -50% at 18 months) while broken stocks rose (+14% in a year). In 1987 and 2022 everything fell, and in 1990, 2007, 2015 and 2018 weakness spread from the already-weak names. He sees 1998 and the 2025 DeepSeek/Liberation Day episode as possible faint foreshadowing.
His base case is a 2000/1990 mashup, favoring 2000, in which the market fails in a way that lifts the median broken stock. He expects a drift lower, sometimes violent, as 'land of compression' fundamentals hit the AI narrative and the mechanical bid loses steam. He walks back his earlier 1987 comparison and repeats that these setups typically take 6-9 months, though the timing is unknowable. He warns that a government weakened by crisis-era debt cannot easily support the private debt bubble when it bursts. No new trades are disclosed.
Key metrics
- Bear Market Ratio (share of S&P 500 members >20% below 10-yr high) at the Oct 6, 2026 record close: 55%, up from 47% at the Aug 13 record; ordinary tops 20-31%, March 2000 71.2%
- Fresh Drawdown Ratio (stocks >=20% below a 10-yr high set within the prior 2 years): 27% today vs 22% in August, 53% in March 2000, 6-13% at every ordinary top since 1987
- Only 15 S&P 500 stocks closed at a new high on Oct 6, 2026 and 97 were within 10% of one, vs 15 and 87 in March 2000 and 166-284 within 10% at the other seven tops
- Median S&P 500 stock 22% below its 10-yr high today vs 31% in March 2000; 406 of 503 members >10% below, 276 >20% below
- Total-return BMR: 48.5% today vs 18-25% at ordinary tops (2022 18%, 2015 21%, 2018 23%, 2020 25%) and 69.5% in March 2000
- After the March 2000 top, leaders fell 33% in a year and 50% in 18 months while the median broken stock gained 14% in a year; median index member +5% a year after the top
- Real S&P 500 price declines: -53% 1966-1982, -58% 1929-1949, roughly -40% 2000-2011 (with dividends reinvested: ~-10%, ahead by 1949, ~-25%)
- Feb 1995 record had 46% BMR but FDR was collapsing from 34.1% (Dec 1994) to 16.9%, then 10% by June 1995 - breadth was healing, the opposite of today
- BMR hit 98% at the 1987 bottom and 97% in 2020; 82% by October 1990
Key insights
- Today's 55% Bear Market Ratio at a record high has only one 40-year precedent above it - March 2000 at 71.2%; ordinary pre-bear tops (1987, 1998, 2011, 2015, 2018, 2020, 2022) read 20-31%.
- The Fresh Drawdown Ratio strips out hangover stocks: 27% today vs 6-13% at every ordinary top, 53% in 2000. That is why 2007's alarming 43% BMR becomes a benign 10% FDR - three-quarters of its broken names were dotcom leftovers.
- Breadth at the top is the key stat: only 15 new highs and 97 stocks within 10% of a high on Oct 6, 2026, nearly identical to March 2000 (15 and 87) and far below the 166-296 seen at other tops.
- Direction matters: Feb 1995 had a 46% BMR but breadth was healing fast (FDR 34% to 17% to 10%), leading to the great 1990s bull. Today BMR is rising with the index - the 2000 pattern, not 1994's.
- In the only true analogue (2000), leaders lost 33% in a year and 50% in 18 months while the median broken stock gained 14% - Burry's base case is a similar rotation that lifts the median broken stock.
- Breadth is a 'sign' not a 'symptom': the market reports no complaints, but objective signs - earnings quality, off-balance-sheet liabilities, third-party financing, narrow breadth - say it is sick.
- The constant mechanical, market-cap-weighted bid explains why today's caldera is narrow at the top but lacks 2000's deep lower layers; he expects that bid to lose steam over the next two years.
- Timing: these setups typically take 6-9 months to resolve, but he does not know where in that window the market is, and the government's crisis-stretched balance sheet limits any rescue of a private-debt bust.
#051
Community
Oct 14, 2026
Siloam Health - Health Care for Those in Need
This is Cassandra Unchained's October 2026 Charity of the Month post, not an investment note. Burry argues that because Tennessee never adopted full ACA Medicaid expansion and TennCare eligibility is tightly restricted, the low-wage working poor and newly arrived immigrants and refugees in Nashville fall into a coverage gap, leaving nearly 100,000 people uninsured. Siloam Health, a donor-supported clinic he says he has personally vetted and knows well, fills that gap efficiently using donated drugs and equipment plus sliding-scale patient contributions. He commits 5% of the month's gross revenue to it.
Positions & performance
No single-stock positions stated in this report.
Full summary 342 words
This is a charity post rather than a trading note: the Cassandra Unchained Charity of the Month for October 2026. Michael Burry spotlights Siloam Health, a non-profit, donor-supported medical clinic in Nashville that primarily serves the uninsured, the working poor, and newly arrived immigrants and refugees who do not qualify for traditional government assistance. Unlike some earlier picks that came from readers, he stresses that he has personally vetted this charity and knows it well.
He frames the problem as a policy gap. Tennessee declined the Obamacare Medicaid expansion and remains one of 10 states that have never implemented it in full. TennCare exists, but eligibility is tightly restricted, so people working at very low wages without employer coverage generally do not qualify. The result, he writes, is that the medical needs of the working poor and new immigrants and their families go largely unmet, with nearly 100,000 Nashville residents living without health insurance.
Siloam's answer is a clinic that runs like a traditional doctor's office. Prospective patients are screened for eligibility each week and pay what they can afford; the clinic sees about 6,000 patients across 20,000 encounters a year, now through two locations. Burry praises its efficiency with every donated dollar, noting that it relies on donated pharmaceuticals and medical equipment alongside patients' own contributions, and that it has kept this going for 30 years. Siloam also trains Community Health Workers who act as health ambassadors in immigrant and refugee communities, building one-on-one relationships that have helped more than 550 patients. A Preceptorship program teaches medical, physician-assistant, nurse-practitioner and pharmacy students how to deliver care with limited resources, a skill he calls increasingly important. He points to its publicly available 2025 annual report as evidence of transparency.
There are no positions, prices or tickers here. As is now the custom, Cassandra Unchained will donate 5% of its October 2026 gross revenue to Siloam, and Burry encourages readers to donate too. He closes by teasing an upcoming portfolio feature called "Portpourri", which he compares to the paper he used to deliver.
Key metrics
- Tennessee is one of 10 states that never implemented full ACA Medicaid expansion
- Nearly 100,000 people in Nashville live without health insurance
- Siloam cares for about 6,000 patients across 20,000 encounters a year, now through two clinics
- Community Health Workers have helped more than 550 patients in immigrant and refugee communities
- Siloam has operated for 30 years
- Cassandra Unchained will donate 5% of October 2026 gross revenue to Siloam
Key insights
- Not an investment post: Cassandra Unchained's October 2026 Charity of the Month, with no stated positions, prices or tickers.
- Burry puts Nashville's uninsured problem down to policy: Tennessee is one of 10 states that never fully expanded Medicaid under the ACA, and TennCare excludes most low-wage workers.
- Siloam Health serves about 6,000 patients across 20,000 encounters a year at two clinics, using sliding-scale payments plus donated drugs and equipment.
- Unlike reader-suggested picks, Burry says he has personally vetted Siloam and knows it well; its 2025 annual report is public.
- Community Health Workers have helped more than 550 immigrant and refugee patients, and a Preceptorship program trains students to deliver care with limited resources.
- 5% of Cassandra Unchained's October 2026 gross revenue goes to Siloam.
- He teases an upcoming portfolio feature called "Portpourri", modeled on the paper he used to deliver.
#050
Trading Post
Oct 11, 2026
Fannie Mae, Freddie Mac, Hong Kong Stocks, Value Stocks Late in a Bull Market, and Consolidating a Big Short
Burry still believes Fannie and Freddie will exit conservatorship before Trump leaves office, but he no longer trusts that the senior preferred liquidation preference will be cancelled: Trump could have done it when stronger politically and did not, and Wall Street's base case is moving toward conversion into common, which would dilute common holders at least 90%. The junior preferreds sit above the common, must have dividends reinstated before the government can pay itself, and now carry affirmed Fairholme precedent, so he sold all Twins common and built an 8-10% position across five preferred series priced for only about 12% odds of resolution. On the short side, he expects a significant top within zero to nine months and has rolled nearly all shorts into roughly 20% out-of-the-money NDX puts (wash-sale exempt) and SMH puts, keeping puts on Oracle, MetLife, Caterpillar and Palantir. He is considering selling value longs on new lows to limit damage in the final narrowing phase, while adding to BYD in a depressed Hong Kong market.
aichina-aiaimemoryaineocloudairobotaisemiconsumerhealthconsumershoppingenergygridfinancefintechhealthsaas
Positions & performance
+7.7%avg return75% win · 16 tracked · L +3.5% · S +16.9%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFISV:NASDAQ financefintech Fiserv HOLDcommon | NASDAQS&PBitget | ISAPT | 46.05closeOct 12 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +35.8% |
| WhyStill held, consistent with the October 5 decision to neither buy nor sell at a price he sees offering low-teens long-term returns. |
| SCAT:NYSE energygrid Caterpillar HOLDDecember 2027 puts, well out of the money | S&PBitget | IAPT | 799.37closeOct 12 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +28.6% |
| WhyKept as a standalone put position rather than folded into the index shorts, struck well out of the money with a long expiry. |
| SPLTR:NASDAQ saas Palantir HOLDSeptember 2027 puts, well out of the money | NASDAQS&PBitget | ISPTC | 209.0closeOct 12 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +23.7% |
| WhyThe stock has doubled since June, when he covered half the common short, and he covered the rest when moving to puts. It closed at a new high and could run further, but volume keeps falling as it rallies, which he reads as building exhaustion. |
| SSMH:NASDAQ aisemi VanEck Semiconductor ETF NEWJune 2027 puts, about 20% out of the money · one of two positions holding almost the entire short book | NASDAQBitget | I | 590.99closeOct 12 '26 | 461.68 ● | 315.05671.83 | +21.9%APY +643.5% | +21.9% |
| WhyA concentrated bet against the hottest and most overextended semiconductor names (Nvidia 19.6%, TSMC 9.3%, AMD 5.6%, with Broadcom, Micron, Marvell and Applied Materials near 5% each), exactly the stocks he wants to be short. Rolling single-name shorts into it also realized his tax losses in semiconductor names, and he expects the puts to rise sharply well before SMH is down 20%. |
| LMOH:NYSE health Molina Healthcare HOLDcommon | — | ITC | 204.36closeOct 12 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +14.1% |
| WhyListed among the longs he continues to own; he says he will maneuver over the next 6-9 months to minimize losses in longs, which may mean selling some on new lows. |
| LQXO:NYSE energygrid QXO HOLDmandatorily convertible preferred stock | — | IT | 10.64closeOct 12 '26 | 12.12 ● | 10.0427.61 | -0.7%APY -4.1% | +13.9% |
| WhyListed among the longs he continues to own, held through the mandatory convertible preferred rather than the common. |
| SORCL:NYSE aineocloud Oracle HOLDDecember 2027 puts, well out of the money | S&PBitget | ISAPTC | 137.39closeOct 12 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +13.4% |
| WhyOne of four single-name put positions kept outside the SMH/NDX consolidation, struck well out of the money with a long expiry to span his expected top. |
| LHCA:NYSE health HCA Healthcare HOLDcommon | S&P | IAPT | 458.6closeOct 12 '26 | 511.78 ● | 353.99556.52 | +41.4%APY +174.8% | +11.6% |
| WhyListed among the longs he continues to own while he weighs a sell-on-new-lows discipline for the late-stage bull. |
| LBIRK Birkenstock HOLDcommon | — | IA | 33.85closeOct 12 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +5.2% |
| WhyListed among the longs he continues to own while he weighs a sell-on-new-lows discipline for the late-stage bull. |
| LADBE:NASDAQ saas Adobe HOLDcommon | NASDAQS&PBitget | IPT | 238.14closeOct 12 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | +4.3% |
| WhyListed among the longs he continues to own while he weighs a sell-on-new-lows discipline for the late-stage bull. |
| LJD:NASDAQ airobot JD.com HOLD | NASDAQBitget | IS | 26.72closeOct 12 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | +2.6% |
| WhyRemains in the portfolio as a very undervalued name in a Hong Kong market that is under pressure; most of his Hong Kong positions are owned directly. |
| LVEEV:NYSE health Veeva Systems HOLDcommon, with November 2026 covered calls written roughly at the money · calls written against the entire position | S&PBitget | IT | 295.85closeOct 12 '26 | 298.04 ● | 148.05325.99 | +18.0%APY +87.4% | +0.7% |
| WhyVeeva has been on a tear, so he wrote at-the-money November calls on the whole position to cover a big chunk of downside. He accepts the shares may be called away. |
| LDECK:NYSE consumershopping Deckers Outdoor HOLDcommon | S&P | IPT | 84.27closeOct 12 '26 | 83.46 ● | 77.20122.29 | +4.1%APY +40.1% | -1.0% |
| WhyStill held after being bought on October 5 as the tax-loss proxy for lululemon and as an investment he likes in its own right. |
| SMET:NYSE financefintech MetLife HOLDJanuary 2029 puts, well out of the money | S&P | IPT | 100.17closeOct 12 '26 | 103.15 ● | 67.33106.72 | -6.5%APY -43.3% | -3.0% |
| WhyRetained as his expression of risk in insurer balance sheets and private credit marks; the longest-dated of the single-name puts he kept. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market HOLDJanuary 2028 LEAP calls | NASDAQ | IAPT | 64.33closeOct 12 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -21.5% |
| WhySprouts just made another new low, the kind of signal on which he sold value stocks in the late 1990s as breadth narrowed. He does not expect to sell the LEAPs for now but is newly considering re-adopting a sell-on-new-lows rule for the next 6-9 months. |
| LERO Ero Copper HOLDcommon | — | I | 37.53closeOct 12 '26 | 27.46 ● | 20.6140.83 | -20.9%APY -77.7% | -26.8% |
| WhyListed among the longs he continues to own while he weighs a sell-on-new-lows discipline for the late-stage bull. |
| LFNMA Fannie Mae EXITcommon · all common stock | NASDAQ | — | —Oct 11 '26 | — | — | — | n/a |
| WhyWall Street's base case (per KBW and the price action) is shifting to Treasury converting its senior preferred into common, which dilutes common holders at least 90% and more each quarter as the roughly $388 billion liquidation preference grows. Selling realizes the full tax loss and moves him up the capital structure. |
| LFMCC Freddie Mac EXITcommon · all common stock | NASDAQ | — | —Oct 11 '26 | — | — | — | n/a |
| WhySold with the Fannie common for the same reason: Trump could have cancelled the liquidation preference at any time while politically stronger and did not, which lowers the odds he will, and the common is the wrong security to own if the preference is affirmed or converted. |
| LFNMA Fannie Mae / Freddie Mac junior preferred stock NEWjunior preferred stock (five undisclosed series of the Twins, tilted to floating-rate and higher-coupon issues) · very full position, about 8-10% of the portfolio, across five series (larger than the prior common position) | NASDAQ | — | —Oct 11 '26 | — | — | — | n/a |
| WhyThe junior preferreds sit above the common, so conversion of the senior preferred dilutes around them rather than through them, and the government cannot take dividends from its common without first reinstating theirs. At about 26% of par they imply only around 12% odds of any resolution in two years versus his view of well above 50%, for a blended return of up to 3.9x at par; the July 10 Fairholme affirmation adds leverage for preferred holders. |
| SNDX Nasdaq 100 Index NEWApril 2027 puts, roughly 20% out of the money · one of two positions holding almost the entire short book | — | — | —Oct 11 '26 | — | — | — | n/a |
| WhyBroad-based index options are exempt from wash-sale rules, so he can trade and re-position freely without tax consequences. Low volatility makes the puts cheap and the April 2027 expiry fits his view that a significant top is zero to nine months away. |
| L1211 HK BYD ADDcommon (Hong Kong shares and BYDDF US OTC F-shares) | — | — | —Oct 11 '26 | — | — | — | n/a |
| WhyAdded to the position started the prior week at levels he had waited for patiently. He also bought BYDDF, noting the OTC F-shares are the actual Hong Kong common with no Cayman intermediary, which he prefers to Y-suffix ADRs. |
| L6862 HK Haidilao HOLDcommon (Hong Kong shares / HDALF F-shares) | — | — | —Oct 11 '26 | — | — | — | n/a |
| WhyRemains in the portfolio as very undervalued along with much of the Hong Kong market. |
| L6690 HK Haier Smart Home HOLDcommon | — | — | —Oct 11 '26 | — | — | — | n/a |
| WhyRemains in the portfolio as very undervalued along with much of the Hong Kong market. |
| L700 HK Tencent HOLDcommon | — | C | —Oct 11 '26 | — | — | — | n/a |
| WhyRemains in the portfolio as very undervalued along with much of the Hong Kong market. |
| LTPW AU Temple & Webster HOLDcommon (ASX shares / TPLWF) | — | — | —Oct 11 '26 | — | — | — | n/a |
| WhyListed among the longs he continues to own while he weighs a sell-on-new-lows discipline for the late-stage bull. |
Full summary 562 words
This Trading Post and Short Thoughts mash-up follows his October 7 breadth piece and the October 5 tax-loss post, in which he had swapped Fannie Mae common into Freddie Mac. Six days later he reverses course on the instrument entirely. The Twins have retraced essentially all gains since Trump's election and traced what he calls a giant head-and-shoulders-and-elbows top that technically predicts worthless stock, in a summer brutal for anything outside the AI trade.
His reasoning runs through the capital structure. Maximum return on the common always required both the end of conservatorship and cancellation or severe reduction of the senior preferred liquidation preference, which now stands at about $388 billion, roughly twice the combined book value, and grows a dollar for every dollar of net worth earned. The Twins drew $191 billion and have repaid at least $301 billion, so in fairness shareholders should be rewarded, but price action and KBW commentary suggest the base case is becoming conversion of the senior preferred into common, diluting holders at least 90%. Momentum for an exit has been usurped by the Iran war, higher rates, a weakening housing market and tight Treasury finances. He quotes his own July warning on credit deterioration and argues a populist administration will be loath to hand hedge funds holding the common a win, and that Trump's inaction while stronger lowers the odds he cancels the preference.
The junior preferreds fare better in every scenario that affirms the preference. Fannie's stack is $19 billion at par and about $6 billion at market. The preferreds are senior to the common, and a majority-owner government cannot receive dividends without reinstating theirs, which would send them toward par. He therefore sold all Fannie and Freddie common, realizing the tax loss, and bought five undisclosed series favoring floating-rate and higher-coupon issues, increasing the position to a very full 8-10% of the portfolio. At a collective 26% of par he computes just 12% implied odds of any resolution within two years, against his own estimate of well above 50%, with a blended return of up to 3.9x. He adds the D.C. Circuit's July 10, 2026 affirmation of the Fairholme verdict ($299.4 million to Fannie preferred, $281.8 million to Freddie preferred, $31.2 million to Freddie common), which establishes breach, lets claims travel with the shares and finds continuing injury, though courts cannot force a write-down.
On the short side he has consolidated. Having earlier moved shorts into options, he has now rolled most into April 2027 Nasdaq 100 puts and June 2027 VanEck Semiconductor ETF puts, each about 20% out of the money. NDX options escape wash-sale rules, and SMH concentrates Nvidia, TSMC, AMD, Broadcom, Micron, Marvell and Applied Materials. He kept well out-of-the-money puts on Oracle, MetLife, Caterpillar and Palantir, and notes Palantir has doubled since June on falling volume.
For longs he contemplates reviving a late-1990s discipline of selling value stocks at new lows, citing Sprouts, though he is keeping his January 2028 LEAPs for now. He lists continuing holdings, has written November at-the-money calls against all of his Veeva, and added to BYD via Hong Kong shares and BYDDF, while holding JD, Haidilao, Haier and Tencent. Alibaba stays off the list until the HKD 70s. The view fails if the bull simply runs on, which he concedes could happen quickly, or if Treasury leaves the Twins unresolved through the warrant expiry.
Key metrics
- Senior preferred liquidation preference about $388 billion, roughly 2x the combined book value of Fannie and Freddie
- The Twins drew $191 billion from Treasury and have paid back at least $301 billion in cash
- Conversion of the senior preferred would dilute common holders at least 90%; Treasury warrants cover 79.9% of the companies
- Fannie Mae junior preferred stack: $19 billion at par, about $6 billion at market; JPS in aggregate trade at about 26% of par
- JPS position is about 8-10% of his portfolio across five series; blended return up to 3.9x at par (3x-5x by series)
- Market-implied odds of resolution within two years about 12% versus Burry's 'well above 50%'
- Fairholme award affirmed July 10, 2026: $299.4M to Fannie preferred, $281.8M to Freddie preferred, $31.2M to Freddie common
- Short book: NDX April 2027 puts and SMH June 2027 puts, each about 20% out of the money
- SMH weights: Nvidia 19.6%, TSMC 9.3%, AMD 5.6%; Broadcom, Micron, Marvell and Applied Materials near 5% each
- Alibaba not reconsidered until the HKD 70s or a reversal of its stock issuance for AI
Key insights
- Burry has sold all Fannie Mae and Freddie Mac common and now holds the Twins only through five undisclosed junior preferred series, an 8-10% portfolio position that is larger than what he held in the common.
- His logic is structural: if Treasury converts the roughly $388 billion senior preferred into common, common holders are diluted at least 90% while the junior preferreds, senior to the common, are diluted around rather than through.
- He prices the bet explicitly: at about 26% of par the preferreds imply roughly 12% odds of any resolution in two years, versus his view of well above 50%, for up to a 3.9x blended return at par.
- The July 10, 2026 D.C. Circuit affirmation of the Fairholme verdict gives junior preferred holders precedent and negotiating leverage; Fannie common is not part of it and Freddie common has only a $31.2 million claim.
- Nearly his entire short book is now two positions: April 2027 NDX puts and June 2027 SMH puts, both about 20% out of the money. He chose NDX partly because broad index options are exempt from wash-sale rules.
- Single-name shorts that survive are puts on Oracle (Dec 2027), MetLife (Jan 2029), Caterpillar (Dec 2027) and Palantir (Sept 2027); the Palantir common short is fully covered after the stock doubled since June.
- He reiterates a significant top within zero to nine months and says he may sell value longs on new lows over the next 6-9 months, as he did in the late 1990s, to keep losses small while breadth narrows.
- BYD was added to again, including BYDDF, which he notes is the actual Hong Kong common rather than an ADR; Alibaba will not be reconsidered until the HKD 70s or a reversal of its AI-funded share issuance.
#049
Short Thoughts
Oct 08, 2026
Short Thoughts: The AI Cloud Oligopoly, Stages of Grief & a Flashing Bear Market Indicator
The tech giants other than Apple are spending as if their survival depended on it when it does not, which Burry reads as a bid to extend monopoly rents into an AI oligopoly that government will tolerate and fear, with OpenAI, Anthropic and Oracle spending to join it. That spending has no defined economic end-point and is financed through a private equity, private credit and insurer pyramid whose valuation marks he hears creaking loudly, just as long-term rates rise. Meanwhile the market is in the denial stage of a grief cycle: the S&P 500 is at record highs while 55% of its members are in bear markets and more than 80% are in corrections, a breadth pattern seen only once before in his 35-year reconstruction, at the 2000 top. He concludes a significant top is zero to nine months away, after which capital leaves the AI narrative amid violent swings and eventually finds the beaten-down rest of the index. He has repositioned for more upside in depressed longs and holds his shorts entirely in options.
aineocloudmag5
Positions & performance
Full summary 598 words
Burry begins with an observation about motive. The technology giants, Apple excepted, are spending as though their lives depended on it, yet monopolies by definition need not. He concludes they expect AI to become one more monopoly rent, collected by an oligopoly the government will tolerate and even fear because AI running on their machines could control economic and political forces beyond any politician's reach. OpenAI, Anthropic and Oracle are not monopolies but intend to join that oligopoly under cover of a new paradigm, with rewards large enough that Larry Ellison bet his company on it. He finds it frightening that the spending has no defined goal or economic end-point, but gives the companies less credit than the technocracy scenario implies: this is a want, not a need, 'compression will do its deed,' and much of the outlay will prove sunk cost.
The second section turns to financing. He reports hearing 'creaks and groans, and some loud snapping and crackling' in valuation marks across private equity and private credit, where shadow banking has moved offshore and insurers have been reinvented. Wall Street knows how much the PE, private credit and insurer pyramids matter to the AI buildout, a theme he covered in Parts IV and V of his Heretic's Guide. He previews a slide from Part VI on how that stress transmits back to data centers, and charts the 10-year yield rising alongside the S&P 500, noting higher rates are already grinding gears at PE and private credit funds.
He then maps market cycles onto the stages of grief, expanded to six for investors: denial, anger, bargaining, apathy, depression, acceptance. Tops are made in denial, which lasts six to nine months or longer. His evidence is 1998-2000 (the Russian default, Long-Term Capital, millionaire.com fizzling in spring 1999, six Fed hikes, and still an 84% Nasdaq surge into March 2000), 2007 (CDS blowing out in February while the S&P peaked late in the year), 1929 and the early 1970s. Anger is the violent drops and rips early in a bear market. Because of modern market structure, described in his two Foundations pieces, bottoms now form during bargaining rather than apathy, and monetary and fiscal 'cocaine' has interrupted the full cycle for two decades.
The core is his bear market indicator: the share of S&P 500 constituents down 20% or more from their highs, built from Bloomberg data and stitched back to 1991. Healthy bulls show about 20%. After LTCM in late 1998 the reading climbed through every new index high to 70-80% by early 2000. Today it is 55%, the second-highest bull-market reading in 35 years after 2000's 71%, and the median member is down 22.3%. On the 10% threshold, more than 80% of the index is in correction for the first time since 2000, and the count has risen since the Liberation Day tariffs were paused. He calls the bull 'a dead man walking.'
No individual trades are given. He expects the narrative to begin grieving within zero to nine months, possibly triggered by marks at insurers and private funds restricting capital to data centers with $3 trillion in construction, purchase orders and leases outstanding. Capital should eventually rotate to the beaten-down majority, later dovetailing with 401(k) flows turning net seller, though even good depressed stocks get hurt at first. He is repositioning for more upside in both beaten-down longs and shorts, and is now entirely in options on the short side. He concedes no one can predict a top, the market may reach metrics never seen, and for most people simply selling winners and waiting is enough.
Key metrics
- Median S&P 500 member is down 22.3% from its high while the index makes record highs
- 55% of S&P 500 stocks are down 20% or more today, versus 71% at the March 2000 peak (up to 80% in the weeks around it)
- More than 80% of S&P 500 stocks are down 10% or more, the first time since 2000 during a bull market; 4 of 5 were at the March 2000 top
- Healthy bull markets typically show about 20% of stocks down 20%; the 10%-drawdown share averaged 47% at market tops during the 1990s
- Burry's timeline to a significant top: zero to nine months ('months, not years')
- $3 trillion in data center construction in progress, purchase orders and leases on unoccupied facilities
- Nasdaq rose 84% from Q4 1999 to the March 2000 peak despite six Fed hikes across 1999 and the first half of 2000
- Breadth reconstruction covers 35 years back to 1991; only the TMT top shows a deeper median constituent drawdown
- The denial phase in which tops form lasts 6-9 months, possibly more than a year
Key insights
- Burry's breadth indicator, the share of S&P 500 stocks down 20% or more, reads 55% with the index at record highs. In 35 years of his data only the 2000 top was higher, at 71% on the peak day and up to 80% around it.
- The median S&P 500 stock is 22.3% below its high, and more than 80% of members are in a 10% correction, matching the roughly four-in-five reading at the March 2000 top.
- He puts a significant market top zero to nine months away and says his confidence in the shorter timeline is why he is repositioning; record capex does not sway him because tops occur within the capital cycle.
- His short book is now held entirely in options, and he is seeking more upside leverage in beaten-down longs as well.
- He names a specific trigger: bad marks at insurers and private credit and equity funds 'could blow up at any time,' choking capital to data centers that have $3 trillion in construction in progress, purchase orders and leases on unoccupied facilities.
- He reads hyperscaler spending as a grab to extend monopoly into an AI oligopoly, with OpenAI, Anthropic and Oracle paying to join, and expects compression to turn much of it into sunk cost. Apple is the exception among the giants.
- Breadth began deteriorating after the Liberation Day tariff pause, the analogue to late 1998 after LTCM, when the bull resumed but breadth never recovered.
- The rotation he expects favors the depressed majority of the index, but he warns even the best beaten-down stocks will be hurt in the initial stage, and that modern market structure puts bottoms in the bargaining phase rather than apathy.
#048
Trading Post
Oct 05, 2026
Trading Post October 5, 2026: Also, Introducing a New Feature, Portpourri
The fourth quarter is tax-loss harvesting season, and stocks that are already down a lot get extra selling pressure from late October into early December. Burry tries to beat that rush in six favored but deeply out-of-favor longs at fresh lows, using three techniques: swap into a proxy for the 30-day wash period (LULU into DECK, Fannie into Freddie), simply hold when there is no proxy and the loss is small (FISV), or replace stock with cheap far out-of-the-money LEAP calls (SFM, ZTS). The option swaps also raise his leverage to a 2000-2003 style value revival he expects as the AI boom turns to bust sooner than past booms did, a view reinforced by rumblings about marks inside private credit and private equity. He pairs that with more far out-of-the-money MetLife puts, and uses the Hong Kong fire sale to add JD calls and start a BYD common position.
aichina-aiairobotconsumerhealthconsumershoppingfinancefintechhealth
Positions & performance
+5.6%avg return67% win · 6 tracked · L +8.0% · S -6.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFISV:NASDAQ financefintech Fiserv HOLDcommon | NASDAQS&PBitget | ISAPT | 46.14closeOct 05 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +35.5% |
| WhyNo good proxy and not a very significant loss, so he neither buys nor sells. He admits buying poorly (his process had it more attractive in the low-to-mid $30s; insiders bought around $50), but the current price is cheap enough to offer low-teens long-term returns. |
| LZTS:NYSE health Zoetis NEWfar out-of-the-money LEAP calls, 2028-2029 expiries (replacing common) · replaces the stock position | S&P | IAPTC | 71.41closeOct 05 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +18.8% |
| WhySame treatment as Sprouts: new lows, no significant loss, and cheap long-dated calls, so the stock was swapped for far out-of-the-money LEAPs that keep upside to a value revival while cutting capital at risk into tax-loss season. |
| LDECK:NYSE consumershopping Deckers Outdoor NEWcommon · sized to replace the LULU position | S&P | IPT | 80.21closeOct 05 '26 | 83.46 ● | 77.20122.29 | +4.1%APY +40.1% | +4.1% |
| WhyBought as a proxy for LULU during the wash-sale window and as an investment he likes in its own right. He expects whatever forces lift LULU over the next month to lift DECK too, since the two are remarkably similar on price and price/sales, though LULU is more discounted in some ways. |
| LJD:NASDAQ airobot JD.com ADD2029 calls, far out of the money | NASDAQBitget | IS | 26.36closeOct 05 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | +4.0% |
| WhyHong Kong is 'on fire sale again' in a bear market longer than any US one since before 1929, while the companies themselves are nowhere near as damaged; he added cheap long-dated upside to JD. |
| SMET:NYSE financefintech MetLife ADD2029 puts, far out of the money | S&P | IPT | 96.88closeOct 05 '26 | 103.15 ● | 67.33106.72 | -6.5%APY -43.3% | -6.5% |
| WhyAdded on his own research and 'rumblings' about the state of marks deep inside private credit and private equity portfolios; he thinks insurance regulators may be the first to call the marks out. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market NEWfar out-of-the-money LEAP calls, 2028-2029 expiries (replacing common) · replaces the stock position | NASDAQ | IAPT | 64.92closeOct 05 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -22.2% |
| WhyThe stock made new lows without a significant loss for him, and he found the LEAP calls fairly cheap, so he replaced the shares with far out-of-the-money calls for more leverage to a 2000-2003 style value revival once the AI boom turns to bust. |
| LLULU:NASDAQ consumershopping lululemon athletica EXITcommon · entire share position (swapped into DECK) | NASDAQS&P | ISAPT | 93.14closeOct 05 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | +13.2%closed |
| WhyTax-loss sale ahead of the late-October to early-December harvesting rush: LULU is his biggest loser, has made fresh lows and looks set for a rough fall. He plans to swap back into LULU once the 30-day wash-sale period expires, and may keep DECK as well depending on factor exposure. |
| LFNMA Fannie Mae EXITcommon · all Fannie Mae shares (proceeds into Freddie Mac) | NASDAQ | — | —Oct 05 '26 | — | — | — | n/a |
| WhyRealizes the tax loss after the Twins fell hard in a news vacuum, while keeping exposure by moving everything into Freddie Mac. He plans to do the same swap in reverse next month. |
| LFMCC Freddie Mac ADDcommon · received all of the Fannie Mae proceeds | NASDAQ | — | —Oct 05 '26 | — | — | — | n/a |
| WhyHe does not want to lack exposure when the senior preferred liquidation preference question is answered, which he says will land 'like Thor's hammer' on the bears and can happen at any time, though Trump may withhold clarity until after the midterms. |
| L1211 HK BYD NEWcommon (Hong Kong shares / US OTC F-shares) · decent size | — | — | —Oct 05 '26 | — | — | — | n/a |
| WhyThis is the second time BYD has fallen to this price area, which he had been waiting for patiently; he likes the company as laid out in his March 'Hong Kong Stocks: Structure & Strategy, Part 2' piece. |
Full summary 492 words
This Trading Post is organized around tax-loss harvesting. Burry says the fourth quarter brings extra selling pressure on stocks that are already down heavily, with the rush arriving near the end of October and peaking in the first week of December, and that he tries to act before it. Six of his holdings stand out as likely targets that 'may not do well over the next couple of months': lululemon, Zoetis, Sprouts Farmers Market, Fiserv, and Fannie Mae and Freddie Mac. All are positions he favors, all are badly out of favor, and all have made fresh lows. He uses them to illustrate three ways of handling the problem.
The first is the proxy swap. LULU is the big loss, so he sold the shares and bought Deckers Outdoor, both as a proxy and as a stock he likes on its own merits, reasoning that whatever lifts LULU during the 30-day wash period will lift DECK too. A chart compares price and price/sales for both. He intends to swap back after 30 days and may keep DECK as well depending on his sector and factor exposure. He applied the same idea to the GSEs, selling all his Fannie Mae into Freddie Mac after the pair fell hard on a news vacuum, and plans to reverse the trade next month. He refuses to be without exposure when the senior preferred liquidation preference question is answered, though he notes Trump may wait until after midterms that look poor for Republicans.
The second is to do nothing. For Fiserv he has no good proxy and only a modest loss. He concedes he bought it poorly, since his process rated it more attractive in the low-to-mid $30s even though insiders bought near $50, but he holds because the price now offers low-teens long-term returns. The third is the option replacement: Sprouts and Zoetis made new lows without large losses, and he found their LEAP calls cheap, so he replaced the stock with far out-of-the-money calls expiring in 2028 and 2029.
That choice reflects a macro view. He expects a 2000-2003 style value revival as the AI boom turns to bust, and sooner than history would suggest, citing his own research and rumblings about valuation marks deep inside private credit and private equity portfolios; insurance regulators may be first to object. Accordingly he bought more far out-of-the-money 2029 puts on MetLife.
Away from tax selling, he calls Hong Kong a fire sale again, bought more far out-of-the-money 2029 JD.com calls, and bought BYD common in decent size on its second visit to this price area. He is still waiting for much lower prices on Alibaba after its share issuance to fund AI spending. He notes a fourth approach, simply buying more and ignoring taxes, but says brand-new lows make him pause because bear traps abound. No transaction prices are given. The post closes by announcing Portpourri, a forthcoming feature of short blurbs on securities in his universe.
Key metrics
- Six tax-loss candidates at fresh lows: LULU, ZTS, SFM, FISV, FNMA, FMCC
- Tax-loss selling rush arrives near the end of October and peaks in the first week of December
- 30-day wash-sale window before LULU can be repurchased with the loss kept
- FISV: process rated it more attractive in the low-to-mid $30s; insiders bought around $50; current price offers low-teens long-term returns
- SFM and ZTS replaced with far out-of-the-money LEAP calls expiring in 2028 and 2029
- MET puts and JD calls both extended to 2029 expiries, far out of the money
- BYD bought in decent size the second time it reached this price area
- Hong Kong bear market described as longer than any US bear market going back to before 1929
Key insights
- Burry sold his entire LULU share position for the tax loss and bought Deckers (DECK) as a 30-day proxy; he plans to buy LULU back after the wash period and may keep DECK too.
- He sold all of his Fannie Mae into Freddie Mac to harvest the loss without losing GSE exposure, planning to reverse it next month. (Six days later he went further and moved the whole position into junior preferreds.)
- Sprouts and Zoetis are no longer stock positions: both were replaced with far out-of-the-money LEAP calls expiring in 2028-2029, cutting capital at risk while adding upside leverage.
- Fiserv is a plain hold. He admits a poor entry versus the low-to-mid $30s his process preferred, but sees low-teens long-term returns from here.
- He expects the AI boom to turn to bust sooner than past booms and a 2000-2003 style value revival to follow, citing rumblings about private credit and private equity marks and the possibility that insurance regulators call them out first.
- The MetLife short was increased again via far out-of-the-money 2029 puts, his chosen expression of the private-credit and insurer-marks risk.
- BYD is a new common stock position in decent size, bought on its second drop to this area; JD exposure was increased with far out-of-the-money 2029 calls.
- Alibaba remains off the buy list until much lower prices, because of the share issuance funding its AI buildout.
#047
Essay
Oct 01, 2026
Don't Believe Your Lyin' Eyes, GPU Depreciation & Useful Lives
Nvidia's roadshow slide claiming its GPUs retain value beyond depreciation schedules does not show resale prices; it shows Silicon Data's discounted-cash-flow value of current rents run out over eight years, divided by third-party purchase-price estimates and compared against a five-year accelerated curve no customer actually uses. GPU rents were falling faster than scheduled depreciation with every new generation until a memory shortage in late 2025 made new servers scarce and lifted every generation at once, so today's residual values reflect scarcity of compute, not durability. Nvidia itself will end the scarcity, as IBM did, with Vera Rubin promising up to 45x lower cost per token. Meanwhile users are trading down to cheaper tokens as prices rise, squeezing whoever rents out the machines. As in the 1968 computer-leasing boom, the lessors can be right about how long the machines last and still be ruined by what they rent for once the next one arrives, and this time the lessor is hidden in private credit and PE-owned insurers.
aineocloudaisemimag5saas
Positions & performance
+32.6%avg return100% win · 1 tracked · S +32.6%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SNVDA:NASDAQ aisemi Nvidia REIT | NASDAQS&PBitget | ISPTC | 230.86closeOct 01 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +32.6% |
| WhyNo trade is stated; he reiterates the bear case as the depreciation critic Nvidia's September 27 slide targets. The 'retained value' lines are 8-year discounted rental cash flows from Silicon Data, not resale prices, and Nvidia's own steep 5-year accelerated curve is 'a gift to bears' because it admits how hard each new chip generation hits the prior one. He expects server rents to fall dramatically once the memory shortage turns to glut, possibly as early as next year, with Vera Rubin claiming up to 45x lower cost per million tokens. |
Full summary 579 words
This is the second piece in Burry's History Rhymes series and a direct reply to a slide Nvidia showed in its September 27 investor presentation, titled 'Nvidia AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules.' The slide plots the retained value of A100, H100 and B200 chips above a five-year accelerated depreciation curve, implying buyers are over-depreciating, and Nvidia bulls such as author Tae Kim took it as the final answer to depreciation critics like Burry. It follows his September 28 Short Thought on the Ares newsletter and residual-value guarantees.
He frames the dispute through Adam Smith's 1968 book The Money Game, which Warren Buffett recommended to his partners. In it the Great Winfield mocks the narrator for asking how fast computers are written off, and Billy the Kid explains he ran $5,000 into over half a million in six months on 3-5% cash in computer-leasing convertibles such as Leasco and Randolph Computer, because 'the need for computers is practically infinite.' Burry adds Time magazine's September 1967 piece on the leasing game, which noted lessors undercut IBM by betting on ten-year lives, and Fed economist Stephen Oliner's finding that the IBM System/360 did last 8 years 9 months. The Kids were right about demand and nearly right about useful life, yet the average computer stock fell about 80% from the late-1968 high, bottoming in May 1970, five weeks before the System/370 launched. Later machines lasted six and then five and a half years, quality-adjusted mainframe prices fell about 20% a year, and most lessors collapsed or were absorbed.
Turning to the slide, he reads the footnotes. The lines are Silicon Data's, and the number is not a price but a continuous DCF of forward rental rates over an eight-year physical life at 75% starting utilization and roughly $0.60 an hour of operating cost. The denominator is third-party purchase-price estimates of $20,000, $35,000 and $45,000; using Mercatus's $56,000 for a B200 cuts 158% to 127%. The A100's 25% is entirely due to years seven and eight in the model. No hyperscaler or neocloud uses accelerated depreciation; all use straight-line five to six years, so Nvidia's steep curve is an admission of what its product cycle does to prior generations.
Silicon Data's own indexes show each generation falling faster than scheduled depreciation: the H100 dropped 45% in under a year as Blackwell arrived, and the B200 and A100 lost 19% in five months after the B300 shipped. Then memory seized up in October and November 2025 after OpenAI's Stargate letters of intent, and every generation turned up together from February 2026. Nvidia's supply commitments jumped from $119 billion to $279 billion in a quarter, mainly for memory. He estimates higher memory content explains only 15-16 points of the H100's 58%; the main lever is that the memory shortage made new servers scarce. Token data shows proprietary prices up 95% while the blended price paid fell 7.5% as buyers fled to cheaper models, so demand is not infinite.
No trades are stated. The positioning implication is that lessors and their financiers bear the squeeze: rents fall dramatically once memory goes to glut, possibly next year, although Nvidia's CFO sees shortages through January 2028 and CoreWeave's CEO for the foreseeable future, which is the view that would prove him early or wrong. He closes with Saul Steinberg's Leasco buying Reliance Insurance, liquidated in 2001, and warns that PE firms' captive insurers are likely to get into trouble with data center loans and bonds.
Key metrics
- Nvidia Sept 27 slide: residual value as % of purchase price - B200 158%, H100 58%, A100 25%
- Nvidia's 'accelerated schedule' loses 43% in year one and reaches zero around year five; hyperscalers and CoreWeave use straight-line 5-6 year lives (about 17% in year one over 6 years)
- Silicon Data model: 8-year physical life, 75% starting utilization, about $0.60/hour operating cost; purchase prices of $20,000 / $35,000 / $45,000 (A100 / H100 / B200) are third-party estimates
- Using Mercatus's $56,000 per B200 GPU, the 158% falls to 127%
- H100 rent fell from $3.40/hour in Sept 2024 to about $1.86 by early 2025 (-45%); B200 and A100 each lost 19% in the five months after B300 shipped
- A100 rents $1.58/hour; GDDR6 index up 46.5% since late January; H200 up 25% since May; B300 at about a 40% premium
- Nvidia supply commitments rose from $119 billion to $279 billion in one quarter, 'primarily related to the procurement of memory'
- Token prices Dec to late May: proprietary +95%, open-model +60%, blended paid price -7.5%; proprietary share of tokens fell from about 80% to about 25%
- Vera Rubin: 'up to 45X lower cost per million tokens' than GB300; revenue opportunity per gigawatt up from $18 billion (Hopper) to $40 billion (Rubin)
- 1968 analogue: average computer stock fell about 80% from the late-1968 high; IBM System/360 served 8 years 9 months on average; System/370 launched June 1970, five weeks after the market low
Key insights
- The y-axis on Nvidia's retained-value slide is not a second-hand price: it is Silicon Data's present value of current rent run out eight years by formula, compared against a five-year accelerated curve, which Burry calls wholly apples to oranges.
- The slide is internally inconsistent: an eight-year physical life in the model, a five-year depreciation curve on the chart, and a claimed 10-year A100 life two slides later.
- The B200's 158% uses a $45,000 third-party cost estimate; at Mercatus's $56,000 per GPU it is 127%, and the A100's 25% exists only because years seven and eight were added to the model.
- Every GPU generation was falling faster than scheduled depreciation (H100 -45% in under a year) until the late-2025 memory shortage; all generations, plus the GDDR6 index, turned up together from February 2026, which signals scarcity rather than durability.
- Higher memory prices explain only about 15-16 points of the H100 server's 58% residual; the real driver is that memory, not GPUs, is the pinch point limiting supply of new servers.
- Proprietary token prices rose 95% but the blended price paid fell 7.5% as proprietary share dropped from about 80% to about 25% of tokens, so users trade down hard while machine rents rise, squeezing the lessor.
- Nvidia will end the scarcity itself: Vera Rubin, shipping since August 2026, claims up to 45x lower cost per million tokens than the GB300, the same vendor-driven obsolescence that sank computer lessors after IBM's System/370.
- The 1968 lessors were right about machine life (8.75 years for the System/360) and still lost about 80% in the stocks; Burry expects today's equivalent pain to land in private credit and PE-owned captive insurers holding data center loans and bonds, echoing Leasco's Reliance Insurance, liquidated in 2001.
#046
Short Thoughts
Sep 28, 2026
Short Thoughts: A Wall Street Titan Hands Us Part V.5 of Our Heretic's Guide to AI's Stars.
The AI capex boom is financed through a fragile, circular web ultimately backstopped by more than $10 trillion of insurance float plus vendor residual-value guarantees on chips. Those guarantees exist because chip depreciation is real (the Nvidia H100 has lost 51% of its value in three years), not because vendors want to offer them. The entire structure rests on a single assumption: sustained AI capital spending. A single season of disappointing AI revenue could lead capital-allocating boards to conclude the highest-conviction bet has shifted, triggering the interconnected guarantees roughly simultaneously and precisely when guarantors are weakest. Burry frames this as the same vendor-financing correlation risk that turned the 2000 telecom fiber glut from a technology story into a systemic one, and as mainstream Wall Street (via Ares) converging on the bearish view of his Heretic's Guide to AI's Stars series.
aineocloudaisemimag5saas
Positions & performance
+30.8%avg return100% win · 5 tracked · S +30.8%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SNBIS:NASDAQ aineocloud Nebius REIT | NASDAQBitget | ISPT | 231.88closeSep 28 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +53.2% |
| WhyNeo-cloud upstart in the same vendor-financed, residual-value-dependent cohort as CoreWeave, named for coverage in Part VII of the Heretic's Guide series. |
| SCRWV:NASDAQ aineocloud CoreWeave REIT | NASDAQBitget | ITC | 85.07closeSep 28 '26 | 49.76 ● | 40.58138.25 | +43.4%APY +345.4% | +41.5% |
| WhyNeo-cloud whose efforts to show chip-pricing stamina are at odds with the depreciation reality lenders face; flagged as a Part VII short-thesis target dependent on the vendor-financing backstops. |
| SNVDA:NASDAQ aisemi Nvidia REIT | NASDAQS&PBitget | ISPTC | 228.86closeSep 28 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +32.0% |
| WhyThe H100 has lost 51% of its value in its first three years; Nvidia now offers residual-value guarantees not because it wants to but because lenders require them, confirming the chip-depreciation risk at the center of his AI-bubble short thesis. |
| SGOOGL:NASDAQ mag5 Alphabet (Google) REIT | NASDAQS&PBitget | ISPTC | 342.75closeSep 28 '26 | 283.9 ● | 272.11408.61 | +17.2%APY +217.9% | +17.2% |
| WhyGoogle Cloud is named among hyperscalers offering residual-value guarantees on chips of unproven durable value, a Big 5 node in the interconnected guarantee structure he argues would unwind simultaneously if AI capex falters. |
| SORCL:NYSE aineocloud Oracle REIT | S&PBitget | ISAPTC | 132.6closeSep 28 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +10.2% |
| WhyOne of the Big 5 hyperscalers behind the 'missing $3 trillion' of Part IV; extending residual-value guarantees on depreciating chips because it has to, a tell of weak underwriting embedded in the circular AI-financing web. |
Full summary 507 words
This Short Thought interrupts Burry's ongoing 'Heretic's Guide to AI's Stars' series (Parts IV and V are already out, and Part VI on the financing web is in progress) because he considers the material need-to-know now. The trigger is Ares Management's Fall 2026 Alternative Credit Newsletter, which he reviewed as part of his research and reads as mainstream Wall Street coming around to the structural view he has been developing. He urges readers to study Ares's pages 2-17 alongside his own Parts IV and V.
The core structural point is that the only pool of capital deep enough to fund the AI buildout sits on insurance company balance sheets, in the float created by the timing gap between premiums paid in and claims paid out. Between US and Bermuda insurers, that float amounts to a little over $10 trillion. He cites Ares's observation that the same risk the insurance industry finds too big to insure as a property liability, it swallows whole as an asset when it arrives as an investment-grade bond in the float. Burry says he has spent months digging up the insurance detail and will add the other half and then some when Part VI prints.
He then presents a detailed map of the circular transactions among AI companies, with categories of financing on one side and the big players on the other, but tells readers the structure and its risk matter more than any individual deal. Every obligation, backstop and guarantee rests on one assumption: sustained AI capex. A single season of disappointing AI revenue could lead boards predisposed to chase the highest-conviction bet to conclude that bet has shifted; the legal documents contemplate that decision, and because the players are interconnected, guarantees would trigger more or less simultaneously, exactly when guarantors are weakest. Chip depreciation is the crux: the Nvidia H100 has lost 51% of its value in its first three years. He echoes Ares's historical self-obsolescence examples, where dominant vendors obsolete their own legacy product through their upgrade cycles, citing IBM mainframes in 1979, the Airbus A380 in 2019, and solar PV modules in 2014. This is why Nvidia and hyperscalers such as Oracle and Google Cloud now offer residual-value guarantees, not because they want to but because lenders require them, replaying the vendor-financing circularity that turned the 2000 telecom fiber glut from a technology story into a risk-correlation story.
No new trades with prices or sizes are stated; the note reinforces the bearish AI-complex thesis rather than logging transactions. Part VII will cover neo-clouds such as CoreWeave and Nebius and similar upstarts, whose insistence on chip-pricing stamina is at odds with the reality lenders face. Burry warns that a bubble being widely and repeatedly documented (as in 1999 and before the housing bust) does not make it any less a bubble, and points to willful ignorance in the White House, on Wall Street and in Silicon Valley. The view would be invalidated if AI revenue keeps justifying the capex underwriting it, keeping chip residual values and the guarantee web intact.
Key metrics
- US + Bermuda insurance float: a little over $10 trillion
- Nvidia H100 has lost 51% of its value in its first 3 years
- Ares Fall 2026 Alternative Credit Newsletter: pages 2-17 are the key must-read sections
- Ares 'Ares On Hyperscalers And Risk' newsletter PDF: 6.65 MB
- 'Missing $3 Trillion' framing from Heretic's Guide Part IV (Big 5 hyperscalers)
- Historical self-obsolescence analogues cited by Ares: IBM mainframes 1979, Airbus A380 2019, solar PV modules 2014
- Telecom fiber glut of 2000 cited as the vendor-financing correlation-risk analogue
- 3 books on the 1999 bubble were published during 1999
Key insights
- Insurance float (US + Bermuda, a little over $10 trillion) is the only capital pool deep enough for the AI buildout; Ares notes insurers take on as investment-grade bonds the same catastrophe risk they will not insure as a property liability.
- The Nvidia H100 has lost 51% of its value in its first three years, hard evidence that chip depreciation, not durable pricing stamina, drives the financing structure.
- Residual-value guarantees from Nvidia, Oracle and Google Cloud exist because lenders demand them given depreciation; Burry treats them as a red flag, not reassurance.
- He likens the residual-value and vendor-financing circularity to the 2000 telecom fiber glut, which became a risk-correlation event rather than a technology story.
- The whole web of guarantees, backstops and obligations rests on one assumption, sustained AI capex; a revenue disappointment could trigger guarantees simultaneously when guarantors are weakest.
- Ares's historical self-obsolescence analogues span IBM mainframes (1979), the Airbus A380 (2019) and solar PV modules (2014).
- Part VII will target neo-clouds CoreWeave and Nebius, whose claims of chip-pricing stamina conflict with lender-required residual-value guarantees.
- Burry sees Ares as mainstream Wall Street converging on his bearish view, but warns that wide documentation of a bubble (1999, housing) has never prevented the bust.
#045
Deep Diveabridged
Sep 26, 2026
[Abridged] Capital Cycle IQ & Forensic Files on the Big 5 Hyperscalers (MSFT, AMZN, ORCL, META, GOOG)
Burry argues the Big 5 hyperscalers are funding a record AI-datacenter capital cycle increasingly through off-balance-sheet leases, purchase commitments, vendor-financing circles and accounting choices that flatter reported economics while pushing risk into the out-years. Market peaks have historically preceded peaks in net investment, and with S&P 500 net-capex versus GDP near a prior-cycle extreme, he places the group late in the capital cycle. Each name shows the same migration toward longer-dated, riskier backlog and rising leverage even as free cash flow turns negative (Amazon) or gross margins compress (Oracle), with useful-life extensions, finance-to-operating-lease reclassification and financing-laden prepayments used to manage optics. Ranking the universe by IV15/IV10 (the prices implying 15% and 10% long-run compounding), he concludes most are too expensive, keeping Alphabet relatively higher in his 'Castle' tier but marking it down for prompt-based LLM search-disruption risk. The setup is a build-up of hidden liabilities and 'bullwhip' demand that he expects to eventually discombobulate both the industry and the stocks.
aineocloudmag5
Positions & performance
Full summary 507 words
This is the abridged (~1,750-word) version of Part V of Burry's 'Heretic's Guide to AI's Stars' series, a forensic-accounting sweep of the five hyperscalers building the AI cloud (Microsoft, Amazon, Oracle, Meta and Alphabet). It follows Part IV, which tallied roughly $3 trillion in aggregate purchase commitments, future leases, guarantees backing third-party debt and other off-balance-sheet exposures. The framing is the capital cycle: charting S&P 500 net capital investment against U.S. GDP, Burry notes market peaks have historically arrived before peaks in net investment, and argues 'we are near one of those moments again.' Recent 10-Q/10-W filings, earnings calls and disclosures are then mined for how stressed each hyperscaler really is.
Company by company he builds the same pattern. At Microsoft, finance-lease liabilities jumped from $46.2B to $66.6B even as bond debt fell, and CFO Amy Hood said future data-center leases will 'shift from finance leases to operating leases'; Microsoft extends assumed useful lives to as high as 25 years to clear ASC 842's 75% threshold, which Burry calls a 'move of convenience.' He flags $329B of signed-but-uncommenced leases that went unmentioned on the July 29 call, and >1yr backlog growing 113% versus 37% near-term. Amazon's free cash flow has turned negative (~ -$8B) while uncommenced leases plus purchase commitments reached $267B (+81% in nine months) and backlog migrated from $200B at 3.8-year weighted life to $496B at 6.4 years; long-term debt issuance exploded from $746M to $81.9B year over year, leading him to conclude 'Amazon is simply too expensive.'
Meta funds its buildout through variable-interest entities backed by a $28B residual-value guarantee, carries ~$700B of uncommenced leases and obligations, has $46B of maximum off-balance-sheet exposure, and had to post $10.8B of cash collateral 'demanded by its counterparties, who are the ones that know these assets best.' Alphabet shows the largest hidden exposure of all, ~$900B including $707B of supply-chain purchase obligations, with long-term debt doubling to $98.2B and earnings flattered by ~$99B of equity-linked gains in the quarter; he highlights the Apollo/Anthropic/TPU circle ($38B raised for Google chips sold to a company Alphabet owns 14% of and is investing up to $40B in) as vendor financing dressed as demand. Oracle's new +$11.4B 'customer prepayments with significant financing component' line, which lets it accrete interest into contract liabilities and later book it as cloud revenue, he likens to Enron; backlog is $664B but roughly half sits beyond fiscal 2029 and gross margin fell five points.
On the calls, executives stress 'short-lived assets' and the option to 'just slow down,' but Burry counters that only 10-15% of a data center's cost is non-repeating (Jassy's own logic) and that obsolescence is rapid, so the reassurance is thin. Positioning follows from his IV15/IV10 rankings rather than named trades in this abridged cut: most of the group screens expensive, Alphabet is kept in the higher 'Castle' tier but lowered for LLM-search risk, and he notes his covered 'Universe' is skewed toward names he owns because they rank high. Part VI will extend the analysis to other AI players and leverage schemes.
Key metrics
- MSFT: finance lease liabilities rose $46.2B to $66.6B while bond debt fell $43.2B to $40.3B
- MSFT: $329B of signed-but-uncommenced leases not mentioned on the July 29 call; >1yr backlog +113% vs near-term backlog +37%
- MSFT extends data-center useful life to as high as 25 years (from 15) to clear ASC 842's 75% threshold (13-yr lease implies 17.3-yr minimum)
- AMZN: uncommenced leases + purchase commitments $267B (+81% over nine months); free cash flow roughly negative $8B
- AMZN backlog $200B at 3.8yr WAL (Sep 30 2025) migrated to $496B at 6.4yr (Jun 30 2026); long-term debt raised $746M -> $81.9B YoY
- META: ~$700B uncommenced leases/purchase obligations; VIE max exposure $46B off balance sheet; $28B residual-value-guarantee threshold; $10.8B cash collateral posted
- GOOG: $707B supply-chain purchase obligations, ~$900B total off-balance-sheet exposure (highest of Big 5); long-term debt doubled $46.5B -> $98.2B; $99B equity-linked gains last qtr ($136B over 6 months); Cloud backlog +112% in 6 months
- GOOG/Anthropic: Apollo raised $38B for Google TPUs sold to Anthropic; Alphabet owns 14% of Anthropic and is investing up to $40B
- ORCL: new +$11.4B 'customer prepayments with significant financing component' line; backlog $664B (+$209B YoY), ~$332B beyond FY2029, $106B (~16%) beyond 2031; gross margin down 5 points
Key insights
- Microsoft's CFO said future data-center leases will 'shift from finance leases to operating leases,' and MSFT extends useful lives to ~25 years to clear ASC 842's 75% threshold, which Burry calls a 'move of convenience,' not a valid accounting basis.
- MSFT left $329B of signed-but-uncommenced leases unmentioned on the July 29 call and analysts did not ask; long-term backlog grew 113% vs 37% near-term, so future revenue increasingly depends on customers who may never raise the capital to pay.
- Amazon's free cash flow is roughly negative $8B while uncommenced leases plus purchase commitments reached $267B (+81% in nine months) and long-term debt issuance jumped from $746M to $81.9B YoY; Burry says 'Amazon is simply too expensive.'
- Meta funds its buildout via VIEs with a $28B residual-value guarantee and $46B of max off-balance-sheet exposure, and posted $10.8B of cash collateral demanded by counterparties 'who know these assets best,' atop ~$700B of uncommenced obligations.
- Alphabet carries ~$900B of off-balance-sheet exposure (most of the Big 5), doubled long-term debt to $98.2B and booked ~$99B of equity-linked gains last quarter; the Apollo/Anthropic/TPU arrangement is a 'seamless circle' of vendor financing.
- Oracle's new +$11.4B 'customer prepayments with significant financing component' line lets it accrete interest into contract liabilities and later recognize it as cloud revenue, which Burry likens to Enron; ~$332B of its $664B backlog sits beyond FY2029 and gross margin fell 5 points.
- Executives lean on 'short-lived assets' and the ability to 'just slow down,' but Burry notes ~85% of a data center's cost repeats each server generation and warns of a 'bullwhip'/barbell-middle that will disrupt the industry and the stocks.
- Burry ranks the CU universe by IV15/IV10 (prices for 15%/10% long-run compounding), keeps Alphabet in the higher 'Castle' tier but marks it down for prompt-based LLM search risk, and previews Part VI on other AI players and leverage schemes.
#044
Essay
Sep 23, 2026
The Next Big Deadly Fire Nightmare is Being Incubated Right Now by the State of California
Burry, writing as president of his 501(c)(3) nonprofit CRED, argues that California's 'builder's remedy' regime is forcing cities to approve dense luxury housing in Very High Fire Hazard Severity Zones at the top of evacuation routes that already fail. Saratoga's own studies show Pierce Road cannot evacuate current residents before a fire arrives, yet the City Council reversed its Planning Commission's denial and approved the 25-home Masson Estates project out of fear of legal retribution from the State and developer. CRED has sued (CRED v. City of Saratoga) alleging violations of CEQA, the Subdivision Map Act, and the California Constitution's right to safety. Burry frames the 'affordable housing' justification as a fraud — the 'affordable' units are ~$3-4M homes for families earning up to $162,400 — and warns that this pattern is recreating, parcel by parcel, the exact conditions that killed 116 people in Paradise, Palisades and Altadena.
Positions & performance
No single-stock positions stated in this report.
Full summary 480 words
This is not a market note but an advocacy essay: Michael Burry writes as president of Citizens for Responsible and Equitable Development (CRED), a 501(c)(3) he helped found, announcing a lawsuit against the City of Saratoga, California over a hillside housing approval. It extends his recurring interest in the collision between state housing mandates and physical risk, but the subject here is public safety in his own neighborhood rather than a tradable security. Burry lives at the base of Pierce Road, the two-lane mountain road at the center of the case.
He opens with the historical record: the 2018 Camp Fire that killed 85 people in Paradise, where too few roads out of town left residents trapped in gridlock, and the January 2025 Palisades and Eaton fires that together killed 31 and destroyed over 16,000 structures. The common thread, he argues, was overdevelopment in fire-prone hills with narrow evacuation routes. After Paradise, California passed AB 747 requiring cities to analyze whether roads can handle mass evacuation. Saratoga did the study — and it found Pierce Road already fails, with residents needing five hours to evacuate against a four-hour fire.
Despite that finding, and despite its own EIR language conceding 'significant and unavoidable impacts' to emergency access and evacuation, the City Council on August 19 approved the 25-home Masson Estates subdivision in a CAL FIRE Very High Fire Hazard Severity Zone. Burry documents the specifics his group obtained: the Public Works Department flagged an active landslide along the project frontage; a Fehr & Peers study projected the project would add ~30 minutes (13%) to evacuation time; and outside expert KLD concluded the delay creates 'a significant threat to human life.' The Planning Commission denied the project on July 8; the Council reversed six weeks later, with members stating on the record they approved it out of fear of legal retribution from the State's housing agency or the developer.
Burry's legal argument is that fear of the State is not a lawful basis — the Housing Accountability Act itself (Gov. Code 65589.5(d)(2)) lets a city deny a project with a specific, adverse, unavoidable public-safety impact, and CEQA and the Subdivision Map Act require denial here. He then attacks the 'affordable housing' framing as a boondoggle: only 5 of 25 units are 'affordable,' defined against a $205,500-area median income, and even those are ~$2,400 sq ft homes likely selling for $3-4M, while market-rate homes reach ~$10.7M. He calls builder's remedy as practiced 'a mansion-entitlement program.' Finally he shows Masson is not isolated — a 255-unit winery redevelopment, a 97-unit Chadwick Heights project requiring 800,000 cubic yards of grading against a 1,000-yard limit, and 400+ units queued behind on the same failing road. The piece has no investment positioning; it is a call to support the petition at saratogacred.org and a warning that the seeds of the next deadly fire are being planted statewide.
Key metrics
- Camp Fire (2018): 85 deaths, ~19,000 structures burned; fire consumed one football field per second
- Palisades Fire (Jan 7, 2025): grew from ~10 acres to 200 acres in 20 minutes
- Palisades + Eaton fires: 31 deaths, 16,000+ structures destroyed, 100,000 displaced
- Masson Estates: 25-home subdivision in a CAL FIRE Very High Fire Hazard Severity Zone
- City study: residents take 5 hours to evacuate Pierce Road vs 4 hours for fire to arrive
- Fehr & Peers (May 2026) study: project adds ~30 minutes to evacuation, a 13% increase over baseline
- 5 of 25 Masson homes 'affordable' on ~2,100 sq ft lots; 20 market-rate homes up to 7,000+ sq ft on lots up to 2 acres
- San Jose/Sunnyvale/Santa Clara median family income $205,500; 'low income' family of four up to $162,400
- Saratoga existing homes sell ~$1,510/sq ft; a 7,091 sq ft home pencils to ~$10.7M
- Chadwick Heights: 97 units requiring ~800,000 cubic yards of earth moved (90-ft cuts, 70-ft fills) vs the 1,000 cubic-yard municipal limit; 400+ units queued behind Masson on Pierce Road
Key insights
- This report is an advocacy essay from Burry's nonprofit CRED, not a trading post — it contains no tickers, prices or stated positions.
- CRED filed CRED v. City of Saratoga (Case No. 26CV503751) in Santa Clara County Superior Court on Sept 18, 2026, alleging CEQA, Subdivision Map Act and California Constitution violations.
- Saratoga's own evacuation modeling shows Pierce Road needs 5 hours to evacuate residents while a fire would arrive in 4 — failing even before any new development.
- The City Council reversed its Planning Commission's July 8 denial six weeks later, with members stating on the record they approved out of fear of legal retribution from the State or developer.
- Burry attacks the 'affordable housing' justification as a fraud: 5 of 25 Masson units are 'affordable' but are ~$3-4M homes, defined against a $205,500 area median family income.
- He frames builder's remedy as practiced in Saratoga as 'a mansion-entitlement program sited where the state's own fire maps say no one should add people.'
- The Masson site sits on an active landslide (flagged by Public Works) and adds ~30 min / 13% to evacuation time per the Fehr & Peers study, which outside expert KLD called a life-threatening impact.
- Masson is one of several queued projects on the same corridor: a 255-unit Mountain Winery redevelopment, ~two dozen homes at 12991 Pierce Road, and 97-unit Chadwick Heights requiring 800,000 cubic yards of grading vs a 1,000-yard code limit — 400+ units behind Masson.
#043
Trading Post
Sep 22, 2026
Trading Post September 22, 2026
Burry argues the current chip 'shortage' is an artifact of fabs diverting normal-chip capacity to high-bandwidth AI memory, and that as production catches up over the next two years the shortage will blow off and memory will enter another down cycle. With the Nasdaq 100 at historically overvalued, top-heavy levels pushing to new highs, he added to AI/semiconductor shorts (MU, NBIS, SOXX, PLTR) via puts, sized meaningfully, while keeping optionality to cover on further melt-ups. Simultaneously he leaned long into beaten-down consumer/retail names (QXO, BBW, SFM, BIRK, MELI) that have corrected 'tremendously,' taking each to a full position.
aimemoryaineocloudairobotaisemiconsumerhealthconsumershoppingenergygridsaas
Positions & performance
+16.6%avg return67% win · 9 tracked · L -5.5% · S +44.4%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDputs · in some size | NASDAQBitget | I | 572.78closeSep 22 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +75.4% |
| WhyBroad semiconductor short expressing his view that the coming memory/normal-chip glut and overvalued semis lead to a down cycle; he will cover if the market pushes these to new highs again. |
| SNBIS:NASDAQ aineocloud Nebius Group ADDputs · in some size | NASDAQBitget | ISPT | 236.12closeSep 22 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +54.0% |
| WhyPart of his added AI-infrastructure short book alongside the memory-cycle call, betting the historically overvalued, top-heavy Nasdaq 100 AI complex is stretched as it drags toward a new all-time high. |
| SMU:NASDAQ aimemory Micron Technology ADDputs · in some size | NASDAQS&PBitget | ISAPTC | 1096.16closeSep 22 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +34.3% |
| WhyBelieves capacity for normal chips was diverted to high-bandwidth AI memory, creating a temporary shortage that will 'blow off' as production catches up over the next two years, sending memory into another down cycle. Cites Acer's CEO flagging Chinese DDR4 oversupply and disputing claims of a lasting shortage. |
| SPLTR:NASDAQ saas Palantir Technologies ADDputs · in some size | NASDAQS&PBitget | ISPTC | 184.99closeSep 22 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +13.8% |
| WhyAdded to his short as an expensive AI-narrative name within a historically overvalued, top-heavy Nasdaq 100; he continues to hold puts but will watch to cover on a further melt-up to new highs. |
| LBIRK Birkenstock ADDcommon · full position | — | IA | 31.68closeSep 22 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +12.4% |
| WhyAdded on a steep correction at what he views as an attractive market price; position is now full-sized. |
| LBBW:NYSE consumershopping Build-A-Bear Workshop ADDcommon · full position | — | S | 26.37closeSep 22 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | +11.6% |
| WhyAdded on a large correction at prices he finds attractive; position is now full-sized. |
| LQXO:NYSE energygrid QXO ADDcommon · full position | — | IT | 12.78closeSep 22 '26 | 12.12 ● | 10.0427.61 | -0.7%APY -4.1% | -5.2% |
| WhyBought more after the stock 'corrected tremendously'; finds the market's price attractive and has now built it to a full position. |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon · full position | NASDAQBitget | ISAPC | 1826.58closeSep 22 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -20.0% |
| WhyBought more of the corrected name at prices he finds attractive; now a full position. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADDcommon · full position | NASDAQ | IAPT | 68.68closeSep 22 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -26.5% |
| WhyBought more after a sharp pullback he considers attractively priced; now a full position. |
Full summary 408 words
This Trading Post is a short, action-focused update posted while 'watching the big game' - the Nasdaq 100, which Burry calls historically overvalued and top-heavy, dragging its weight toward a new all-time high. The hook is a Taiwanese report quoting Acer CEO Chen Chun-sheng on components: price increases this year are 'unprecedented,' with customers going from complaining to scrambling to buy, and Q4 PC prices set to rise 5-20%. Crucially, Chen disputes the narrative of a lasting memory shortage, pointing to mainland Chinese memory makers as 'price disruptors' releasing unlimited capacity, DDR4 already in oversupply (more sellers than buyers), and only select parts (LPDDR5, DDR5 9600) still tight. He also implies the three big memory suppliers are signaling price hikes through 2027 as a coordinated posture under antitrust constraints ('those who understand will understand'), and notes even Apple accepted Samsung's contract increase.
Burry reads this as confirmation of his own view: capacity for producing normal chips was redirected to high-bandwidth chips for AI, which created the shortage in normal chips. Over the next two years, he expects that shortage to 'blow off' as production catches up, sending memory into another down cycle. He is candid that interpreting the news - which predates the prior day's big run-up - is 'fraught,' but says it aligns with what he believes to be true.
On the back of that thesis he shorted more Micron (MU), Nebius (NBIS), the iShares Semiconductor ETF (SOXX), and Palantir (PLTR), and says he did so 'in some size.' He frames these as tactical: he will watch and look to cover should the market take them up to new highs again, and he continues to hold his puts - indicating the short exposure is expressed largely through options rather than outright stock.
The other half of the update is a set of long adds in beaten-down consumer and retail names. He purchased more QXO, Build-A-Bear (BBW), Sprouts (SFM), Birkenstock (BIRK), and MercadoLibre (MELI), noting all have 'corrected tremendously' and that he finds the prices the market is offering rather attractive. He states that all of these are now full positions for him. No specific transaction prices or sizes are given beyond 'in some size' for the shorts and 'full positions' for the longs. The view would be invalidated if the AI/memory complex keeps melting up (his stated trigger to cover the semis shorts) or if the corrected consumer names continue lower on deteriorating fundamentals rather than sentiment.
Key metrics
- Acer expects Q4 PC price increases of 5% to 20% across products
- DDR4 in oversupply (more sellers than buyers); LPDDR5 / DDR5 9600 still short
- Nasdaq 100 near a new all-time high, described as historically overvalued and top-heavy
- Memory suppliers guiding price increases through 2027; Apple accepted Samsung's contract price increase
- Burry sees the normal-chip shortage blowing off over ~2 years as production catches up
- Long adds (QXO, BBW, SFM, BIRK, MELI) all taken to full positions after large corrections
Key insights
- Burry's core call is a memory down-cycle: he thinks fabs shifted normal-chip capacity to high-bandwidth AI memory, and that as capacity catches up over ~2 years the shortage 'blows off' - the opposite of the suppliers' guide-to-2027 narrative.
- He leans on Acer CEO Chen Chun-sheng's remarks - Q4 PC prices +5-20%, DDR4 oversupplied, Chinese makers as 'price disruptors' releasing unlimited capacity - as external confirmation, while acknowledging the read is 'fraught.'
- The added shorts (MU, NBIS, SOXX, PLTR) were put on 'in some size' and are expressed through puts he continues to hold, not outright stock.
- These are explicitly tactical shorts: he says he will look to cover if the market takes them to new highs again, signaling defined risk management rather than a permanent bearish anchor.
- He hints at coordinated-signaling among the three big memory suppliers ('under antitrust regulations, companies can only communicate through such signals... those who understand will understand'), and notes even Apple accepted Samsung's price increase.
- On the long side he bought more QXO, BBW, SFM, BIRK and MELI - all names that have 'corrected tremendously' - and took each to a full position, a value-on-weakness stance in consumer/retail.
- The framing sets an overvalued, top-heavy Nasdaq 100 pushing new highs against beaten-down consumer names, a barbell of AI-complex shorts vs. corrected-consumer longs.
#042
Trading Post
Sep 21, 2026
Trading Post September 21, 2026: A Number of Value Buys, Including Two New to the CU Universe
Ignoring the AI melt-up ("the woo-hoos"), Burry goes value-hunting and puts on a batch of buys, two of them new to the Cassandra Unchained universe. He initiates Ero Copper as a leveraged play on a structural copper squeeze: large new discoveries have dried up, mines take ~18 years to build, and AI data-center demand is tightening the market on a 2-3 year cycle, so a discounted higher-cost producer with clean accounting and a completing capex program has outsized upside. He also starts QXO, Brad Jacobs's building-products roll-up, preferring its mispriced mandatorily convertible 5.5% preferred (7.4% yield) alongside common, betting on Jacobs's consolidation playbook and a re-rating once housing/rates noise clears. He adds to Temple & Webster, Sprouts and Zoetis. No shorts were covered and no puts sold.
consumerhealthenergygridhealth
Positions & performance
-6.0%avg return20% win · 5 tracked · L -6.0%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LZTS:NYSE health Zoetis ADDcommon · a little more | S&P | IAPTC | 71.33closeSep 21 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +18.9% |
| WhyAdded a little more to a name he has always admired that is extremely out of favor, with no COVID bump, no AI attachment and no current momentum. |
| LQXO:NYSE energygrid QXO Inc NEW5.5% mandatorily convertible preferred (now 7.4% current cumulative yield) | — | IT | 12.2closeSep 21 '26 | 12.12 ● | 10.0427.61 | -0.7%APY -4.1% | -0.7% |
| WhyHe favors the mandatorily convertible 5.5% preferred, which he argues is mispriced with no natural institutional holders (Bloomberg shows a ~27% premium to parity when the real gap is ~4% because it misreads the floating conversion ratio). Mandatory converts issued after tight financing typically trade better once the dust settles, and the dividend offers downside protection. |
| LQXO:NYSE energygrid QXO Inc NEWcommon | — | IT | 12.2closeSep 21 '26 | 12.12 ● | 10.0427.61 | -0.7%APY -4.1% | -0.7% |
| WhyA building-products roll-up run by Brad Jacobs (of United Rentals and XPO), his new venture since 2024, with the stock way down. Scale makes sense in the business and Jacobs knows how to execute consolidation; Burry sees a re-rating opportunity once housing, rates and diesel noise fade. |
| LERO Ero Copper NEWcommon · mid-size position | — | I | 34.7closeSep 21 '26 | 27.46 ● | 20.6140.83 | -20.9%APY -77.7% | -20.9% |
| WhyNew management installed in early 2025 delivered on three difficult promises and a forensic pass on the accounting came up clean, rare in mining. As a higher-cost, Brazil-based producer trading at a big discount to larger peers, it benefits most as the copper supply squeeze intensifies, and it is on the far side of a large capex program that should start cash flowing. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADDcommon · a little more | NASDAQ | IAPT | 68.66closeSep 21 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -26.4% |
| WhyRounded out the buys by adding a little more to his existing Sprouts position. |
| LTPW AU Temple & Webster Group ADDcommon · fairly large position | — | — | —Sep 21 '26 | — | — | — | n/a |
| WhyHe bought more of the Australian online furniture retailer, now one of his larger positions, and plans to write out the full Temple & Webster thesis relatively soon. |
Full summary 507 words
Burry's September 21 Trading Post opens with one of his signature vignettes -- a night in the backyard with a flashlight, catching crawfish and musing on copper -- as a frame for tuning out the AI-driven market ('the house party is packed... but I am largely ignoring the woo-hoos') and hunting instead for value. The post is a list of buys, two of them entirely new to the Cassandra Unchained universe, and it closes flatly: neither shorts covered nor puts sold.
The centerpiece new position is Ero Copper (ERO), a Brazil-focused producer. Burry's case rests on a structural supply squeeze he backs with Apollo chief economist Torsten Slok: major copper discoveries of at least 500,000 tonnes have collapsed from double digits a year in the 1990s-2000s to one or two recently, with none in 2025, and because a new deposit takes roughly 18 years to reach production, today's shortfall dictates supply in the 2040s. Against that, AI data-center construction is adding copper-intensive load on a two-to-three-year cycle, pointing to a structurally tighter market. The biggest copper names have already run, so Burry reaches for a higher-cost producer that gains most as the squeeze intensifies and still trades at a big discount to larger peers. He is encouraged that new management installed in early 2025 delivered on three difficult promises, that a forensic pass on the accounting came up clean, and that regulation is finally breaking the company's way; the stock is also on the far side of a large capex program that should start cash flowing. He takes ERO common as a mid-size position.
The second new name is QXO, Brad Jacobs's building-products roll-up (Jacobs previously built United Rentals and XPO). Burry favors and owns the mandatorily convertible 5.5% preferred -- now a 7.4% current cumulative yield, payable in cash or stock -- and also bought common. His edge is a familiar one: mandatory convertibles are mispriced because they have no natural institutional holders and standard tools misread them. He notes Bloomberg shows a ~27% preferred-to-parity gap when the real gap is only about 4%, because it interprets the floating conversion ratio wrong. Mandatory converts, he argues, are usually issued when financing is tight (so ratings agencies forgive), which means waiting after issuance can get a better price -- 'here we are.' He cites a fellow Substack writer (Bearstone) and Jacobs's own book, How to Make a Few Billion Dollars, and expects execution-driven consolidation to re-rate the name once it stops being dragged by housing, rates and diesel noise.
He rounds out the buys by adding to three existing names: Temple & Webster (TPW AU / TPLWF), the Australian online furniture retailer, now a fairly large position with a full thesis promised soon; a little more Sprouts Farmers Market (SFM); and a little more Zoetis (ZTS), a name he has long admired that is 'extremely out of favor' with no COVID bump, no AI attachment and no current mojo. The post ends by flagging that Part V of his Heretic's Guide series will follow within 24-48 hours.
Key metrics
- Ero Copper operations entirely in Brazil; new management (early 2025) delivered on three difficult promises
- Major copper discoveries (>=500,000 tonnes contained) fell from double digits/year in the 1990s-2000s to one or two recently, with none in 2025 (Torsten Slok, Apollo)
- A new copper deposit takes roughly 18 years to reach production -- today's shortfall describes supply in the 2040s
- AI data-center buildout adds copper-intensive load on a 2-3 year cycle vs multi-decade mine lead times
- QXO 5.5% mandatorily convertible preferred now yields 7.4% current cumulative, payable cash or stock
- Bloomberg shows a ~27% preferred-vs-parity gap; Burry says the real gap is only ~4% (floating conversion ratio misread)
- Brad Jacobs (United Rentals, XPO) has run QXO since 2024; stock way down
- Temple & Webster now a fairly large position for him
- No shorts covered and no puts sold this post
Key insights
- Two brand-new positions this post -- Ero Copper (ERO) and QXO -- both classic Burry value/contrarian setups against an AI-obsessed tape he is deliberately ignoring.
- The copper thesis leans on Torsten Slok's data: near-zero large discoveries (none in 2025) plus ~18-year mine lead times versus AI copper demand on a 2-3 year cycle imply a structural squeeze into the 2040s.
- Burry deliberately chose a higher-cost producer (ERO) over the big miners that already ran, because a higher-cost name has more torque as the squeeze intensifies -- and it still trades at a big discount to peers.
- What made ERO investable to him: new 2025 management delivering on three hard promises, a clean forensic accounting pass, favorable regulation, and a capex program about to start cash flowing.
- On QXO he prefers the mandatorily convertible 5.5% preferred (7.4% current yield) -- a recurring Burry edge: mandatory converts are mispriced with no natural institutional holders and Bloomberg misreads the floating conversion ratio (27% quoted gap vs ~4% real).
- He frames QXO as a Brad Jacobs roll-up bet (United Rentals, XPO pedigree), a stock 'way down' since his 2024 start, expecting a re-rate on execution once housing/rates/diesel noise clears.
- Temple & Webster (TPW AU) is now a fairly large position and he promises a standalone thesis soon -- worth watching as a future deep dive.
- He explicitly did not cover any shorts or sell any puts, so the short book is unchanged; this post is purely accumulation on the long side.
#041
Deep Dive
Sep 19, 2026
The Heretic's Guide to AI's Stars Part IV: The Big 5 Hyperscalers & the Missing $3 Trillion
The five hyperscalers (Microsoft, Meta, Alphabet, Amazon, Oracle) have quietly loaded roughly $3 trillion of off-balance-sheet obligations — ~$1.2T in mostly non-cancelable uncommenced data-center leases, >$1.5T in purchase commitments, plus SPVs and guarantees — against combined annual earnings of under $400 billion. Under ASC 842 an uncommenced lease creates no asset and no liability until the keys turn, so these commitments sit invisible in footnotes while growing many times faster than the companies themselves. Meanwhile >$400B of capitalized Construction-in-Progress depreciates nothing until switched on, flattering current earnings, and GPUs iterate every 12-18 months while purpose-built halls take 3-5 years to build — a duration mismatch that leaves scant secondary market when demand softens. When the AI capex cycle rolls over (booms historically peak near the midpoint of the spending cycle), these commitments convert into impairments, termination costs and stranded capacity very fast, and the write-offs would dwarf the group's earnings.
aineocloudaisemimag5
Positions & performance
No single-stock positions stated in this report.
Full summary 546 words
Part IV of Burry's 'Heretic's Guide to AI's Stars' series is a forensic read of the 10-K and 10-Q footnotes of the five hyperscalers — Microsoft, Meta, Alphabet, Amazon and Oracle — timed to wait for Oracle's report so the set is complete. It follows Part I (supply-side gluttony as the cardinal sign of a bubble) and Part III (Tracepalooza & the Bezzle, on benchmarking-driven demand and captive-insurer financing). Burry opens by mocking Wall Street: using two decades of sell-side price targets on lululemon, he shows analyst target-vs-price spreads peak near tops, and notes Oracle now sits at a $98 target spread on a ~$250 stock (39% premium) with almost no sell ratings — a contrarian tell. He argues most analysts covering these ex-software/AdTech names have never modeled maintenance capex and still aren't doing it well.
He is explicit that he is not alleging fraud — CEOs and CFOs certified these filings under penalty of perjury — and even hands out some credit: Meta's capex is well-defined, Microsoft depreciates buildings conservatively (after raising building lives to 25 years) and discloses backlog cuts, and Oracle's own risk factors candidly describe the dangers. The core of the piece is what the filings hide. Under ASC 842, a non-cancelable lease sits off the balance sheet until it commences. Uncommenced leases have exploded: Microsoft from $92.7B to $329.1B, Meta from $103.8B to $279.0B (plus ~$68B signed in July), Oracle disclosing $288.0B for the first time, Amazon to $137.2B and Alphabet to $85.2B — nearly $1.2 trillion in total, at least $857B non-cancelable and invisible.
Purchase commitments add another layer, and here Alphabet is the whale at $811B, with Meta $349B, Microsoft $229B, Amazon $130B and Oracle $34.2B — over $1.5 trillion combined. Between the two, the group carries ~$2.7 trillion in off-balance-sheet contractual commitments; add SPVs, guarantees and contingent backstops and it clears $3 trillion — against combined earnings of under $400 billion. Burry stresses these liabilities are compounding far faster than the businesses, and could reach $4-5 trillion by 2028 (only 15 months out). He layers on Construction-in-Progress: capitalized but depreciating nothing until in service. Alphabet holds $122.8B (29% of gross property), Meta $80.3B, Amazon $71.7B, Oracle $40.0B (+142%), and Microsoft — which hasn't disclosed CiP since ~2001 — an estimated $85B-$127B, deduced from the gap between implied ($43B) and actual ($34B) depreciation. Total non-depreciating CiP tops $400B, much of it Nvidia GPUs 'economically disappeared' in warehouses.
The punchline is the duration mismatch: GPUs iterate every 12-18 months while custom-cooled, custom-powered halls take 3-5 years to build, and when demand softens a purpose-built 130kW-rack data hall has no deep resale market — unlike office space, which Meta could still only sublease after $5B of 2022 impairments (and it wrote off $1.34B of canceled data-center builds even then). Booms peak near the midpoint of capex cycles (echoing Chancellor's Capital Account on Marathon), so when the music stops these off-balance-sheet commitments become impairments, termination costs and stranded capacity almost overnight, producing write-offs more impressive in absolute terms than anything else on the tape. Positioning is thesis-level rather than a stated trade log here — Burry frames all five hyperscalers plus Nvidia as the objects of a bearish structural case, with Parts V (more hyperscalers) and VI (demand and financing) to follow.
Key metrics
- Uncommenced leases across the Big 5 total ~$1.2 trillion, of which at least $857 billion is non-cancelable and off the balance sheet
- Purchase commitments across the Big 5 exceed $1.5 trillion; combined off-balance-sheet contractual commitments ~$2.7 trillion, and above $3 trillion including SPVs, guarantees and backstops
- Big 5 combined earnings over the past year are less than $400 billion
- Microsoft uncommenced leases jumped from $92.7B to $329.1B (10-K); 13-year weighted-average remaining lease term; only 'some' cancelable at 1 year
- Meta uncommenced leases $103.8B prior to $279.0B at June 30, plus ~$68B more signed in July (subsequent event)
- Oracle disclosed $288.0B of uncommenced leases for the first time in its latest 10-Q; Amazon $96.4B->$137.2B; Alphabet $58.5B->$85.2B
- Purchase commitments: Alphabet $811B (the whale), Meta $349B, Microsoft $229B, Amazon $130B, Oracle $34.2B
- Non-depreciating Construction-in-Progress tops $400B: Alphabet $122.8B (29% of gross property), Meta $80.3B (27%), Amazon $71.7B, Oracle $40.0B (+142% y/y), Microsoft estimated $85B-$127B
- Microsoft gross property grew 11x from $38B to $432B over ten years; implied blended asset life 8.25 years on a 6-year server life; MSFT has not disclosed CiP since ~2001
- Microsoft targets 32 GW of data-center capacity by 2032 (~3x current), yet only 2 of its current 12 GW is AI-specific; Oracle carries a $98 price-target spread on a ~$250 stock (39% premium) with vanishingly few sell ratings
- Oracle's reported 121% cloud-infrastructure growth is not apples-to-apples and includes an undisclosed circular-demand share; off-balance-sheet liabilities could reach $4-5 trillion by 2028
Key insights
- Uncommenced leases are the tell: ASC 842 keeps a non-cancelable 20-year data-center lease off the balance sheet until it commences, so ~$1.2T of Big 5 lease obligations (>=$857B non-cancelable) sit only in footnotes.
- Combined off-balance-sheet commitments (~$2.7T of leases + purchases, >$3T with SPVs/guarantees) dwarf the group's <$400B annual earnings and are growing many times faster than the companies.
- Alphabet is the purchase-commitment whale at $811B — far above Meta ($349B), Microsoft ($229B), Amazon ($130B) and Oracle ($34.2B) — an unusual physical-supply-chain risk for a former AdTech name.
- Microsoft's uncommenced leases tripled to $329.1B and it hasn't disclosed Construction-in-Progress since ~2001; Burry back-solves ~$85B-$127B of hidden CiP from its depreciation gap.
- Over $400B of capitalized CiP across the group depreciates nothing until switched on, flattering current earnings while Nvidia GPUs 'economically disappear' in warehouses.
- Duration mismatch is the fault line: GPUs iterate in 12-18 months but custom-cooled halls take 3-5 years to build, and a purpose-built data center has no deep resale market when unneeded (unlike Meta's subleasable offices).
- Oracle is the wobbly one — 121% cloud-infra growth is not apples-to-apples and carries undisclosed circular demand; Burry likens its inscrutable dual-purpose balance sheet to pre-GFC AIG, and flags its $98 target spread with almost no sell ratings as a contrarian top signal.
- Booms peak near the midpoint of capex cycles; when the AI buildout rolls over, off-balance-sheet commitments convert to impairments, termination costs and stranded capacity fast, with absolute write-offs potentially exceeding anything else on the tape.
#040
Community
Sep 16, 2026
The Sickle Cell Development Fund
This is Cassandra Unchained's September 2026 Charity of the Month post, not an investment note. Burry draws on his time as a medical intern in Memphis to spotlight sickle cell disease, argues that although cures exist (bone marrow transplant, 2023 gene therapies) the real problem is now access given costs of $2.2M-$3.1M per patient, and champions Dr. Michael DeBaun's Vanderbilt-Meharry program that uses cheap transcranial Doppler screening and generic medicine to prevent strokes in at-risk Nigerian children at a fraction of the cost. He commits 5% of the month's gross revenues to the fund.
Positions & performance
No single-stock positions stated in this report.
Full summary 323 words
This is a charity post rather than a trading note: the Cassandra Unchained Charity of the Month for September 2026. Michael Burry opens with a personal memory from his time as a medical intern at the Elvis Presley Memorial Trauma Center in Memphis, where he saw dozens of sickle cell patients lining the walls, breathing oxygen through masks for what relief it could offer. He uses the scene to explain the disease: a single-letter mutation in the hemoglobin gene produces hemoglobin S, which in low-oxygen capillaries turns red blood cells into rigid, sickle-shaped crescents that pile up and jam small vessels, starving downstream tissue of oxygen and triggering a self-feeding cycle of crushingly painful, recurrent, lifelong attacks.
Burry then frames the core argument as one of access rather than science. A cure exists — bone marrow transplantation has been curative for decades, and new gene therapies were approved in 2023 — but transplants cost hundreds of thousands of dollars with donor-matching and procedural risks, and gene therapies list at $2.2 million to $3.1 million per patient and are available only at a handful of Western centers. Because most children with sickle cell are born in sub-Saharan Africa, and roughly 150,000 annually in Nigeria alone, transplant and gene therapy are not feasible at scale there.
The post highlights the work of Dr. Michael R. DeBaun and the Vanderbilt-Meharry Center of Excellence in Sickle Cell Disease, which over the past decade has become a focal point for treatment in the US and abroad. DeBaun's Nigeria effort screens children with transcranial Doppler ultrasound and treats at-risk children with generic medicine to prevent sickling before a first stroke — dramatically cheaper per child, with more than 11% of untreated children suffering a stroke before age 20. Burry closes by committing 5% of Cassandra Unchained's gross September revenues to the Sickle Cell Development Fund, noting donations are tax deductible and also support research into better care and cures.
Key metrics
- Approved gene therapies for sickle cell list at $2.2M-$3.1M per patient
- Bone marrow transplantation, curative for decades, costs hundreds of thousands of dollars
- ~150,000 children born with sickle cell disease annually in Nigeria
- Over 11% of children with sickle cell suffer a stroke before age 20 absent intervention
- Cassandra Unchained will donate 5% of September gross revenues to the fund
Key insights
- Not an investment post — this is Cassandra Unchained's September 2026 Charity of the Month, with no stated positions, prices or tickers.
- Burry reframes sickle cell as an access problem, not a science problem: cures exist but gene therapies list at $2.2M-$3.1M per patient and are confined to a few Western centers.
- The featured intervention is high-leverage philanthropy: transcranial Doppler screening plus generic medicine to prevent strokes in Nigerian children at a fraction of transplant/gene-therapy cost.
- Scale of need is concentrated in sub-Saharan Africa — roughly 150,000 children born with sickle cell annually in Nigeria, where over 11% would stroke before age 20 without intervention.
- Cassandra Unchained is donating 5% of September gross revenues to Dr. DeBaun's Vanderbilt-Meharry Sickle Cell Development Fund.
#039
Essay
Sep 11, 2026
Shorting the Dollar by Drinking Free First Growth Bordeaux For Life
Burry argues bonded fine wine held in Europe (e.g. at London City Bond) is a scarce, borderless real asset that functions structurally as a long-horizon short on the U.S. dollar. Fine wine has fallen ~25-30% from its October 2022 peak, offering a cyclically depressed entry; academic data show ~4.1% net real returns over 113 years, and buying well below market plus continuous supply destruction (bottles are consumed) adds to that. Because the wine is priced by a multi-currency global bid and held outside U.S. jurisdiction, a 30-40% dollar decline over two decades layers a large FX kicker on top of the wine's own return, while European-vs-U.S. price spreads (45-80%) let a buyer effectively drink First Growth Bordeaux 'for free' by capturing the arbitrage through consumption. He frames it as an uncorrelated hedge against fiscal debasement and AI/quantum-era risks to digital assets, and discloses he has executed and continues to execute the strategy.
Positions & performance
No single-stock positions stated in this report.
Full summary 580 words
Re-published cleanly after a formatting error garbled the prior day's version, this essay lays out Burry's most unconventional idea in some time: buying bonded fine wine in Europe as a long-dated short on the U.S. dollar. He motivates it with a near-future scenario in which AI and quantum computing (he cites IBM's plan to deliver its fault-tolerant 'Starling' machine by 2029) threaten the integrity of digital claims — bank accounts, pensions, Social Security — accelerating the dollar's loss of reserve status to real assets. Gold is already responding; wine, he argues, is a scarce, consumable real asset with a deep global bid that few investors think of as a currency hedge.
The thesis rests on 'three legs.' Leg one is a bear market: fine wine has fallen roughly 25-30% on the Liv-ex indices from its October 2022 peak, the deepest broad correction of the indexed era, with Bordeaux especially depressed while Italian wines lead a tentative 2025-26 recovery. He grounds the return case in the reference paper The Price of Wine (Dimson, Rousseau & Spaenjers, Journal of Financial Economics, 2015), which assembled 36,271 prices for the five First Growths from 1900-2012 and found a 5.3% real return before costs (4.1% net), beating British stamps, art, bonds and bills over 113 years. Leg two is 'free wine for life': the same bottle costs 45-80% more in the U.S. than at the European trade level (Margaux 2015 ~$1,999 in America vs ~£1,120 in the UK trade), so a buyer who purchases at an 18-20% discount to the European market and holds in bond can carve out 15-20% of the book over time for personal consumption and gifting, with the day-one value bonus covering the entire drinking tranche — effectively free First Growth Bordeaux and DRC Burgundy.
Leg three is the currency story he says actually got him started. Wine held in European bond is priced in sterling, euros and francs, so a 35% dollar decline turns even a flat wine market into a ~+54% gain in dollar terms on geography alone. He presents correlation work: essentially zero monthly correlation between the Liv-ex 100 and the DXY, but a ~-0.74 correlation across overlapping five-year periods (block-bootstrap p=0.01), plus wine returning 10.7% on average in nine down-dollar years versus 4.4% in thirteen up-dollar years. He layers this on macro alarm — the dollar's reserve share has slid from 72% (2001) to under 57% (2025), DXY fell 10.8% in H1 2025, and U.S. FY2026 interest ran ~$867B through May — arguing central banks are crowding into gold and other neutral, sanction-proof hard assets.
The execution section is granular: an 'autopsy' of a 2010 Lafite that returned less than nothing over 15 years because the winery priced in future appreciation at issue during peak-China Veblen demand and now releases at industrial scale (up to 20,000 cases/year); guidance to buy only from large, reputable dealers (AVU, Millésima, Justerini & Brooks, Bordeaux Index, Berry Bros. & Rudd) and store at a third-party bonded facility like London City Bond to defeat the three classic wine frauds; a 'three shelves' framework (overpriced trophies, fairly priced middle, and the discounted 'fireflies' where he buys ~1 in 20, e.g. Château Pavie 2021 at ~15-20% under trade). He notes estate step-up and collectibles-tax considerations, rates his own thesis '92 points,' discloses he has executed and will keep executing it, and frames wine as one sleeve of a broader real-asset portfolio alongside gold and real estate. There are no stock positions in this piece.
Key metrics
- Fine wine down ~25-30% on the Liv-ex indices from the October 2022 peak (deepest broad correction of the modern era)
- The Price of Wine (Dimson, Rousseau, Spaenjers, JFE Nov 2015): 5.3% annual real return before costs, 4.1% net of storage/insurance, five First Growths 1900-2012 (36,271 prices)
- 1900-2012 British real returns: equities 5.2%, stamps 2.8%, art 2.4%, govt bonds 1.5%, T-bills 0.9% — fine wine beat all collectibles/fixed income
- DXY fell 10.8% in H1 2025 (worst half since 1973), incl. -9.6% vs Swiss franc in one quarter
- Dollar share of allocated FX reserves fell from 72% (2001) to just under 57% (Q3 FY2025) — first sub-57% reading since 1995
- Wine vs dollar: ~-0.74 correlation across overlapping 5-year periods (block-bootstrap p=0.01)
- Liv-ex 100 vs DXY: essentially zero monthly correlation; wine returned avg 10.7% in 9 down-dollar years vs 4.4% in 13 up-dollar years
- Same bottle runs 45-80% more in America than at the European trade level (Margaux 2015 ~$1,999 US vs ~£1,120 UK); 15% EU wine tariff since 2025
- US Treasury flagged ~$867B in FY2026 interest through May 2026; fiscal policy called long-term unsustainable
- Target buy discounts of ~18-20% to European trade level; bought ~40 of ~700 wines considered this year
- Lafite produces up to ~20,000 cases/year; Bordeaux fell from ~95% to ~38% of the traded market
- Liv-ex Fine Wine 100 vs S&P 500: +0.136 monthly correlation (p=0.02) — near-perfect diversifier that still roughly matches S&P performance
Key insights
- Burry pitches bonded European fine wine as a structural, unleveraged, no-decay short on the dollar — a real asset priced by a multi-currency global bid and held outside U.S. jurisdiction.
- Entry timing is the edge: wine is ~25-30% below its Oct-2022 peak, the deepest broad drawdown of the indexed era, so today is a cyclically depressed (non-random) entry point he thinks can lift long-run real returns toward 6-8%.
- The 45-80% Europe-vs-US price gap lets a holder capture 'free wine for life' through consumption/gifting (carve out 15-20% of the book), while the day-one discount covers the entire drinking tranche.
- Currency kicker: a 35% dollar decline over 20 years turns a flat wine market into ~+54% in dollar terms; he shows ~-0.74 wine/dollar correlation over overlapping 5-year windows (p=0.01).
- His 2010 Lafite 'autopsy' warns that First Growth wineries now price in future appreciation at release and dump ex-château supply at industrial scale (Lafite up to ~20,000 cases/yr) — so buy distressed later, never at release.
- Provenance is everything: buy from large reputable dealers and store at a third-party bonded warehouse (London City Bond) to defeat the three frauds (wine that was never yours, wine that never existed, wine that isn't what the label says).
- He buys selectively — ~40 of ~700 wines considered this year, targeting ~18-20% discounts to the European trade level — and treats the 'fireflies' shelf (e.g. Pavie 2021) as the only investable one.
- Framed as macro insurance: dollar reserve share down from 72% to <57%, DXY -10.8% in H1 2025, ~$867B FY2026 interest — with wine a near-perfect S&P diversifier (+0.136 correlation) that still roughly matches equity returns.
#038
Trading Post
Sep 11, 2026
Trading Post Friday September 11, 2026: Selling Out of One to Pile Into Another
This is a relative-value rotation: Burry sold his entire Flutter (FLTR) position because it looks less attractive at current prices than other stocks that have fallen further, and redeployed the majority of the proceeds into lululemon (LULU) under $100. He argues LULU is highly profitable with a badly compressed valuation whose troubles stem from poor management, making it an easy target for an acquirer who could run it better — a take-private he views as plausible enough to act on with urgency. He also added to Zoetis (ZTS) in the $72s as a cheap, basing name.
consumershoppinghealth
Positions & performance
+11.6%avg return100% win · 2 tracked · L +11.6%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LZTS:NYSE health Zoetis ADDcommon | S&P | IAPTC | 72.0his fillSep 11 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +17.8% |
| WhyBought more Zoetis in the $72s, which he believes is very cheap for this company; it is trying to put in a base here on high volume, though he concedes it may bleed lower first. |
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon · majority of the Flutter proceeds | NASDAQS&P | ISAPT | 100.0his fillSep 11 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | +5.4% |
| WhyPiled the majority of the Flutter sale into lululemon under $100. As profitable as it is and with a valuation this compressed, he suspects LULU may not stay public long — problems have been driven by bad management, so a better operator could easily buy and run it — hence his urgency, though he stresses he has no inside information. |
| LFLTR Flutter Entertainment EXITcommon · entire position | — | — | 25.55closeSep 11 '26 | 25.51 ● | 25.3525.61 | -0.2%APY -0.8% | -0.2%closed |
| WhySold Flutter in its entirety because he judges it no longer as attractive at this price relative to the valuations to which some other stocks have now fallen; a pure relative-value rotation rather than a change of heart on the business. |
Full summary 360 words
This short Trading Post, published September 11, 2026, documents a single decisive portfolio move: Burry sold one position in full to substantially increase another. He opens by flagging it will be brief — he simply exited one name to pile into a second.
The exit is Flutter (FLTR), sold in its entirety. His reasoning is purely relative: Flutter is 'not as attractive at this price relative to the valuations to which some other stocks have fallen.' It is a rotation driven by opportunity cost rather than a negative thesis on the underlying business — other things have gotten cheaper faster, so the capital is better used elsewhere.
That elsewhere is lululemon (LULU), where he put the majority of the Flutter proceeds to work under $100. He ties this to his earlier deep dive on the company ('lululemon athletica (LULU): An Ode to Bad Management & The Politics of Skin-Tight Leggings'), published roughly three months earlier, which framed LULU as a repeat crisis case driven by management failure. Here he sharpens the angle: as profitable as LULU is, and with its valuation this compressed, he believes it 'may not be public for very much longer.' He is explicit that he has no inside information to that effect, but reasons that because the problems have been largely self-inflicted by bad management, other owners could easily buy the company and run it far better — which is what lends his purchase its urgency.
He also added to Zoetis (ZTS) in the $72s, calling it very cheap for the quality of the company. Technically, he notes ZTS is trying to put in a base on high volume, while cautioning it 'may bleed lower' before it turns — a characteristic Burry hedge that he is buying value, not calling an exact bottom.
He closes on housekeeping: with Oracle's results in, earnings season is effectively over for the names he follows, and the next installment of his Heretic's Guide series — Part 4 — should publish early the following week. The positioning takeaway is a concentrated conviction rotation: exit Flutter entirely, concentrate into LULU sub-$100 on a bad-management-plus-takeout thesis, and top up a cheap, basing Zoetis.
Key metrics
- Sold Flutter (FLTR) in full
- Added to lululemon (LULU) under $100
- Bought more Zoetis (ZTS) in the $72s
- LULU deep dive 'An Ode to Bad Management' published ~3 months prior (420 likes, 163 comments)
- Next Heretic's Guide (Part 4) due early the following week, after Oracle's results closed out earnings
Key insights
- Burry exited Flutter (FLTR) entirely — not on a soured thesis but because other stocks have fallen to more attractive valuations, a relative-value rotation.
- The bulk of the Flutter proceeds went into lululemon (LULU) under $100, concentrating the book into a single high-conviction idea.
- His LULU case is a bad-management-plus-takeout thesis: profitable, compressed valuation, troubles caused by management, so an acquirer could buy and run it better; he flags possible take-private with 'urgency' but stresses no inside information.
- He added to Zoetis (ZTS) in the $72s as 'very cheap,' noting it is basing on high volume but may bleed lower first — buying value without calling the exact low.
- The move shows Burry running a concentrated book, willing to sell a full position to fund a higher-conviction one rather than trimming at the margin.
- Earnings season wrapped with Oracle's print; he signaled the next Heretic's Guide (Part 4) for early the following week.
#037
Essay
Sep 10, 2026
Shorting the Dollar by Drinking Free First Growth Bordeaux
Burry argues that European bonded fine wine, bought during its deepest drawdown in the indexed era, is a low-decay, unleveraged short on the US dollar and a hedge against fiscal debasement and AI/quantum-driven digital-asset risk. Because top wines are priced by a global multi-currency bid and stored in bond outside US jurisdiction, a weakening dollar translates directly into dollar-denominated gains on top of the asset's ~4% real return and its built-in supply depletion. Buying 100-point First Growths and top Super Tuscans at 18-20% discounts in a cyclical trough stacks a valuation discount, a currency kicker and an aging 'dividend'. The consumption arm is effectively free: the day-one discount to market covers a lifetime of gifting and special-occasion bottles, while the bulk stays in bonded storage as the investment. The downside case — dollar strength — still leaves scarce real assets compounding at mid-single-digit real rates with near-zero correlation to equities.
saas
Positions & performance
Full summary 508 words
This is not a stock trading post but a long-form investment essay: Burry lays out a three-legged thesis for buying European bonded fine wine as a hard-asset hedge, and discloses he has already executed and will keep executing it. The framing is macro and cyberpunk — he opens on the risk that AI and IBM-style fault-tolerant quantum computing (Starling, targeted for 2029) could eventually threaten the integrity of digital claims (bank, brokerage, pension, Social Security balances), pushing capital toward neutral, sanction-proof real assets. Gold has already 'joined the financial markets' as a marker of too much debt and too little monetary restraint; wine, he argues, is a less-crowded, depleting, borderless version of the same trade.
Leg one is the bear market: fine wine is down roughly 25-30% from its October 2022 Liv-ex peak, the deepest broad correction of the modern era, with Bordeaux especially unbounced. He anchors valuation in the reference academic paper, Dimson, Rousseau and Spaenjers' 'The Price of Wine' (JFE, 2015), which assembled 36,271 prices for the five First Growths from 1900-2012 and found a 4.1% net real return that handily beat stamps, art, gilts and T-bills — and argues that entering at today's trough could lift that to 6-8% real. He distinguishes wine from bourbon (fungible, non-aging) on horizontal (region/producer) and vertical (vintage) differentiation plus continuous supply destruction as bottles are drunk.
Leg two is the 'free wine' arithmetic: on a theoretical CHF 200,000 book mirroring his own recent buys, an average 18% discount to realizable market and a ~22% day-one value bonus fund the entire drinking/gifting tranche, so consumption is effectively paid for by the investment discount. He details the cross-Atlantic mechanics — the same bottle costs 45-80% more in the US (Margaux 2015 ~$1,999 vs ~$1,120 UK trade) thanks to the 15% EU tariff, freight and the three-tier chain — and stresses that only the European trade comp matters and that imported wine can never re-enter the European bonded market.
Leg three is the currency story he says actually motivated the piece: wine in bond is priced by a global sterling/euro/franc/Asian bid, so a 30-40% dollar decline over two decades alone adds +45-65% in dollar terms. He backs it with data — the dollar's reserve share has slid from 72% (2001) to under 57%, DXY fell 10.8% in H1 2025, and over overlapping five-year windows the Liv-ex 100 shows a -0.74 correlation to the dollar (p=0.01), with wine returning 10.7% in DXY-down years versus 4.4% in up years. Execution guidance dominates the back half: a '2010 Lafite autopsy' showing wineries now capture the appreciation at release; buy only deep-discount 'fireflies' (he bought ~40 of ~700 wines this year), avoid trophy value traps and urgency pitches, insist on third-party bonded storage at London City Bond to defeat the three classic wine frauds, and mind collectibles-rate taxes offset by the estate step-up. He rates the thesis 92 points; the main caveat is that it takes five-to-seven figures to get dealer access, and the view stalls (but does not break) if central banks resume accumulating dollars.
Key metrics
- Fine wine ~25-30% below the October 2022 Liv-ex peak; deepest broad correction of the modern indexed era, and now below long-term trend
- The Price of Wine (Dimson, Rousseau, Spaenjers; JFE 2015): 36,271 auction/dealer prices for the five First Growths, 1900-2012; 5.3% annual real return before costs, 4.1% net of storage and insurance
- 1900-2012 UK real returns: equities 5.2%, stamps 2.8%, art 2.4%, gilts 1.5%, T-bills 0.9%; wine 113-yr nominal mean 10.9% vs equities 11.2%
- Depressed-trough entry could lift long-run real returns to 6-8%; a 35% dollar decline over 20 years adds a ~+54% dollar kicker on a flat-to-4% real wine book
- Dollar share of allocated FX reserves fell from 72% (2001) to <57% (Q3 FY2025), the first sub-57% reading since 1995; gold is now the second-largest reserve asset
- DXY fell 10.8% in H1 2025, its worst half since 1973, including 9.6% against the Swiss franc in one quarter
- Over overlapping 5-year windows, Liv-ex Fine Wine 100 shows -0.74 correlation to the dollar index (p=0.01); in 9 DXY-down years wine returned 10.7% avg vs 4.4% in 13 DXY-up years
- Same bottle runs 45-80% more in the US than at the European trade level (Margaux 2015 ~$1,999 US vs ~$1,120 UK trade); 15% EU-wine tariff in force since 2025
- Burry targets 18-20% purchase discounts to the European trade bid; of ~700 wines considered this year he bought ~40 (buys roughly 1 of 20 'fireflies')
- US Treasury flagged fiscal policy as long-term unsustainable; ~$867B of FY2026 interest through May 2026; Bordeaux has fallen from ~95% to 38% of the traded fine-wine market
Key insights
- Burry is not trading a stock here — the 'position' is physical bonded fine wine in Europe, which he frames as an unleveraged, no-margin, no-decay short on the dollar layered on top of a ~4% real return and continuous supply depletion.
- The tell for value is the European trade bid, never US retail: he rejected Latour 2000 pitched at '35% off' because it was discounted to the inflated American price, which he calls 'no information at all'.
- The 'free wine' works only because the day-one discount (he targets 18-20%, on a book carrying a ~22% value bonus) covers the entire cost of the bottles pulled for drinking and gifting; the rest stays in bond as the investment.
- His dollar-hedge evidence firms up only over long horizons — no monthly correlation, but -0.74 over overlapping 5-year windows (p=0.01) and 10.7% wine returns in dollar-down years vs 4.4% in up years.
- The 2010 Lafite 'autopsy': a perfect-score First Growth returned less than nothing over 15 years because, post-GFC, the château now prices at release to capture the appreciation itself — so gains accrue to distressed later buyers, not original subscribers.
- Supply/rotation matters: Lafite makes up to 20,000 cases a year (vs 6,000-8,000 across all seven Burgundy grand crus) and periodically releases ex-château stock, and Bordeaux has fallen from ~95% to 38% of the traded market — the rotation away is the opportunity.
- Fraud defense is structural: buy from large reputable dealers (AVU, Justerini & Brooks, Bordeaux Index, Berry Bros. & Rudd, etc.) and force third-party bonded storage at London City Bond with case photos and rotation numbers to defeat the three fraud types.
- Macro glue: he ties tariffs to reserve status — a weakening dollar makes the 15% EU-wine tariff politically unsustainable, so the American bonded owner effectively holds a call option on future tariff reversal while capturing the currency move regardless.
#036
Trading Post
Sep 09, 2026
Trading Post September 9, 2026: Another Quick One
Burry is pulling in risk heading into the fall. He closed his December 2026 NVDA and Palantir puts outright to sidestep the accelerating time decay that hits short-dated options in September, and rather than rolling them he let the exposure come off. He then trimmed every remaining long and short across the book to raise cash, keeping the overall sizing order intact rather than making directional bets. He retains longer-dated 2027 puts on Palantir and QQQ as standing downside hedges. The message is defensive positioning and dry powder ahead of what he expects to be an interesting, volatile fall market — with a falling dollar index flagged as the subject of an upcoming piece.
aimemoryaineocloudairobotaisemiconsumerhealthconsumershoppingenergygridfinancefintechhealthindexussaas
Positions & performance
+15.5%avg return83% win · 24 tracked · L +4.4% · S +31.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF TRIMcommon | NASDAQBitget | I | 532.0closeSep 09 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +73.5% |
| WhyTrimmed the semiconductor-ETF short as part of the across-the-board reduction of the short book. |
| SNBIS:NASDAQ aineocloud Nebius TRIMcommon | NASDAQBitget | ISPT | 240.35closeSep 09 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +54.8% |
| WhyTrimmed the Nebius AI-infrastructure short along with the rest of the short book. |
| SCRWV:NASDAQ aineocloud CoreWeave TRIMcommon | NASDAQBitget | ITC | 94.94closeSep 09 '26 | 49.76 ● | 40.58138.25 | +43.4%APY +345.4% | +47.6% |
| WhyTrimmed his smallest short, CoreWeave, in the portfolio-wide reduction of AI-infrastructure exposure. |
| SNVDA:NASDAQ aisemi Nvidia TRIMcommon | NASDAQS&PBitget | ISPTC | 223.42closeSep 09 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +30.4% |
| WhyTrimmed the Nvidia short (distinct from the closed Dec-2026 puts) in the general de-risking; NVDA remains a mid-sized short. |
| SCAT:NYSE energygrid Caterpillar TRIMcommon | S&PBitget | IAPT | 815.56closeSep 09 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +30.0% |
| WhyTrimmed the Caterpillar short along with the rest of the short positions. |
| SMU:NASDAQ aimemory Micron TRIMcommon | NASDAQS&PBitget | ISAPTC | 1027.77closeSep 09 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +29.9% |
| WhyTrimmed the Micron short in the broad de-risking while keeping it on. |
| LFISV:NASDAQ financefintech Fiserv TRIMcommon | NASDAQS&PBitget | ISAPT | 48.65closeSep 09 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +28.5% |
| WhyTrimmed alongside every other long as he pulls in risk. |
| SORCL:NYSE aineocloud Oracle TRIMcommon | S&PBitget | ISAPTC | 161.63closeSep 09 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +26.4% |
| WhyTrimmed his largest short as he reduces gross exposure, but keeps Oracle as the top short in the book. |
| LHCA:NYSE health HCA Healthcare TRIMcommon | S&P | IAPT | 421.83closeSep 09 '26 | 511.78 ● | 353.99556.52 | +41.4%APY +174.8% | +21.3% |
| WhyTrimmed in the across-the-board cut; remains part of the core long book. |
| LMOH:NYSE health Molina Healthcare TRIMcommon | — | ITC | 194.13closeSep 09 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +20.1% |
| WhyTrimmed with every other long to pull in risk; stays among his top long positions. |
| LZTS:NYSE health Zoetis TRIMcommon | S&P | IAPTC | 73.31closeSep 09 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +15.7% |
| WhyTrimmed as part of the portfolio-wide de-risking; position rank unchanged. |
| LVEEV:NYSE health Veeva Systems TRIMcommon | S&PBitget | IT | 260.8closeSep 09 '26 | 298.04 ● | 148.05325.99 | +18.0%APY +87.4% | +14.3% |
| WhyTrimmed with all longs to lower exposure and hold more cash. |
| LBIRK Birkenstock TRIMcommon | — | IA | 31.56closeSep 09 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +12.8% |
| WhyTrimmed as part of the blanket reduction across long positions. |
| LPYPL:NASDAQ financefintech PayPal TRIMcommon | NASDAQS&PBitget | ISPT | 52.17closeSep 09 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | +12.5% |
| WhyTrimmed with the full long book; remains a smaller tail position in the sizing order. |
| SQQQ:NASDAQ indexus Invesco QQQ Trust (Nasdaq-100) HOLD2027 puts | NASDAQBitget | SC | 715.56closeSep 09 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +7.4% |
| WhyRetains his 2027 QQQ puts as a standing index hedge and left them untrimmed, keeping downside protection on the Nasdaq-100 while he raises cash elsewhere. |
| SPLTR:NASDAQ saas Palantir HOLD2027 puts | NASDAQS&PBitget | ISPTC | 169.53closeSep 09 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +5.9% |
| WhyContinues to hold his 2027 Palantir puts and deliberately did NOT trim them, since selling the December 2026 puts already reduced his PLTR bearish exposure; the longer-dated puts have room before decay bites. |
| SPLTR:NASDAQ saas Palantir TRIMcommon | NASDAQS&PBitget | ISPTC | 169.53closeSep 09 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +5.9% |
| WhyTrimmed the Palantir short (separate from the Dec-2026 puts he closed) as part of the portfolio-wide short reduction; PLTR stays his second-largest short. |
| LLULU:NASDAQ consumershopping Lululemon Athletica TRIMcommon | NASDAQS&P | ISAPT | 99.72closeSep 09 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | +5.7% |
| WhyTrimmed as part of an across-the-board reduction; remains his largest long position, so sizing rank is unchanged even after the cut. |
| LBBW:NYSE consumershopping Build-A-Bear Workshop TRIMcommon | — | S | 28.2closeSep 09 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | +4.3% |
| WhyTrimmed as part of the portfolio-wide reduction. |
| LJD:NASDAQ airobot JD.com TRIMcommon | NASDAQBitget | IS | 27.0closeSep 09 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | +1.5% |
| WhyTrimmed in the general risk pull-back; the China e-commerce long stays in his larger holdings. |
| LADBE:NASDAQ saas Adobe TRIMcommon | NASDAQS&PBitget | IPT | 254.86closeSep 09 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | -2.5% |
| WhyTrimmed with the rest of the book to reduce net exposure and raise cash. |
| LFLUT Flutter Entertainment TRIMcommon | — | AT | 99.33closeSep 09 '26 | 79.15 ● | 72.56229.86 | -26.0%APY -56.6% | -20.3% |
| WhyTrimmed in the general de-risking while keeping the position on. |
| LMELI:NASDAQ airobot MercadoLibre TRIMcommon | NASDAQBitget | ISAPC | 1876.33closeSep 09 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -22.1% |
| WhyTrimmed in the broad exposure reduction while keeping it a top-tier long. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market TRIMcommon | NASDAQ | IAPT | 73.11closeSep 09 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -30.9% |
| WhyTrimmed with all longs to build a cash buffer ahead of an uncertain fall. |
| SNVDA:NASDAQ aisemi Nvidia EXITDecember 2026 puts · entire December 2026 put position | NASDAQS&PBitget | ISPTC | 223.42closeSep 09 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +30.4%closed |
| WhySold the December 2026 NVDA puts in full to avoid the rapid time-based decay that accelerates once September arrives, and chose not to roll them because he is pulling back exposure across the whole book. |
| SPLTR:NASDAQ saas Palantir EXITDecember 2026 puts · entire December 2026 put position | NASDAQS&PBitget | ISPTC | 169.53closeSep 09 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +5.9%closed |
| WhySold the December 2026 PLTR puts entirely for the same reason as NVDA — avoiding late-year theta decay — and did not replace or roll them as he reduces overall risk. |
| LTPW AU Temple & Webster TRIMcommon | — | — | —Sep 09 '26 | — | — | — | n/a |
| WhyTrimmed alongside the rest of the book; the Australian e-commerce name remains one of his larger longs (also held via US OTC line TPLWF). |
| LFNMA Fannie Mae TRIMcommon | NASDAQ | — | —Sep 09 '26 | — | — | — | n/a |
| WhyTrimmed one of the 'toxic twins' GSE longs in the broad cut, but keeps the position as a standing bet. |
| LFMCC Freddie Mac TRIMcommon | NASDAQ | — | —Sep 09 '26 | — | — | — | n/a |
| WhyTrimmed the second GSE 'toxic twin' along with the rest of the longs while retaining the position. |
Full summary 416 words
This is a short 'Another Quick One' Trading Post in which Burry does not open any new thesis but tells subscribers how he repositioned the portfolio on September 9, 2026. He frames it as pulling in risk while he finishes 'a couple big pieces' due out soon. The single most consequential move is closing his December 2026 puts on both Nvidia (NVDA) and Palantir (PLTR) in full. His stated reason is mechanical rather than a change of view: now that the calendar has turned to September, short-dated options bleed value to rapid time-based (theta) decay, so he took the bearish exposure off rather than pay that carry. Crucially, he did not replace or roll either position, because he is reducing exposure across the whole book, not just rotating it.
Beyond those two complete sales, he trimmed every single position — every long and every short. He is careful to note that the relative sizing order did not change: the largest longs remain, in order, Lululemon (LULU), Molina Healthcare (MOH), MercadoLibre (MELI), Temple & Webster (TPW AU / TPLWF), Zoetis (ZTS), Sprouts (SFM), JD.com (JD), Birkenstock (BIRK), Adobe (ADBE), HCA, Fiserv (FISV), Flutter (FLUT), Build-A-Bear (BBW), Veeva (VEEV), the GSE 'toxic twins' Fannie Mae (FNMA) and Freddie Mac (FMCC), and PayPal (PYPL). On the short side, in order of size, he remains short Oracle (ORCL), Palantir (PLTR), Nebius (NBIS), Nvidia (NVDA), the iShares Semiconductor ETF (SOXX), Micron (MU), Caterpillar (CAT), and CoreWeave (CRWV) — each trimmed the same way as the longs.
He explicitly did NOT trim the longer-dated hedges: he continues to hold 2027 puts on Palantir and on QQQ, and left those untouched precisely because selling the December 2026 NVDA and PLTR puts had already reduced that bearish exposure. So the AI-infrastructure and index-downside thesis stays expressed through 2027-dated options even as the near-term put bets come off.
The net effect is that he now holds some cash, which he says he is 'happy to sit on' while he watches how the market unfolds, calling this 'an interesting market this fall.' He closes with a macro tease: he is watching the US dollar spot index fall again, and that will feature in a post he is close to publishing. There are no prices, targets or position sizes in this note — it is purely a risk-reduction and cash-raising update, and nothing here reverses or rearranges his core long/short framework; it just dials down gross exposure and pushes the bearish tech expression out to 2027-dated puts.
Key metrics
- Sold 100% of December 2026 NVDA puts (no roll)
- Sold 100% of December 2026 PLTR puts (no roll)
- Trimmed every long and every short position in the portfolio
- Largest longs (order): LULU > MOH > MELI > TPW AU > ZTS > SFM > JD > BIRK > ADBE > HCA > FISV > FLUT > BBW > VEEV > FNMA/FMCC > PYPL
- Short book (order): ORCL > PLTR > NBIS > NVDA > SOXX > MU > CAT > CRWV
- Still holds 2027 puts on PLTR and QQQ (left untrimmed)
- Net result: raised cash, happy to sit on it
- Watching the US dollar spot index (DXY) fall again
Key insights
- Burry closed his December 2026 NVDA and PLTR puts entirely — the move was about avoiding September theta decay on short-dated options, not a change in his bearish view on either name.
- He deliberately did not roll the closed puts, signaling he wants less gross exposure overall rather than a rotation.
- Every long and every short was trimmed, but the relative sizing order was left unchanged — this is de-grossing, not repositioning.
- His largest long remains Lululemon (LULU); his largest short remains Oracle (ORCL).
- The AI-infrastructure short book (ORCL, PLTR, NBIS, NVDA, SOXX, MU, CRWV) is intact but smaller after the trims.
- He keeps 2027 puts on Palantir and QQQ untrimmed, so his index and PLTR downside hedges now sit in longer-dated options.
- The trims raised cash he is 'happy to sit on,' explicitly positioning defensively for a volatile fall.
- A falling US dollar spot index (DXY) is flagged as the subject of an imminent post — a macro thread to watch next.
#035
Trading Post
Sep 04, 2026
Trading Post 9/4/2026: lululemon (LULU) Is a Fat Pitch Under $100
lululemon's fundamentals have deteriorated sharply — a -9% global comp, leggings down ~20%, and mid-single-digit operating-margin guidance, with the headline EPS 'beat' flattered by a one-time tariff benefit. But Burry argues the guidance has been deliberately kitchen-sinked ahead of an incoming CEO who inherits the lowest possible bar, while the stock at ~$99 trades just over 2x tangible book with $1.3B cash and no debt for a brand he still believes in. Under every scenario in which LULU ever grows again, $99.50 sits below his IV15 and implies 15-20% annual returns over 15-20 years, making it a 'fat pitch' under $100. He is adding LULU and flags JD (owned) and Alibaba (to be bought) as the only other clear 'super fat pitches' in his universe.
aichina-aiairobotconsumershopping
Positions & performance
-5.6%avg return33% win · 3 tracked · L -5.6%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon | NASDAQS&P | ISAPT | 99.5his fillSep 04 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | +5.9% |
| WhyAfter reworking his IV15 model on the latest 10-Q, call transcript and 8-K, Burry says that under every scenario where LULU grows again $99.50 sits below intrinsic value, offering 15-20% annual returns over 15-20 years; at ~$99 the stock trades just over 2x tangible book with $1.3B cash and no financial debt, and a new CEO enters with a deliberately kitchen-sinked, lowest-possible bar. He plans to buy more under $100. |
| LJD:NASDAQ airobot JD.com HOLDcommon | NASDAQBitget | IS | 28.26closeSep 04 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | -3.0% |
| WhyBurry names JD one of the only other 'clear super fat pitches' in his universe alongside Alibaba, and states he already owns it. |
| LBABA:NYSE aichina-ai Alibaba Group NEWcommon | Bitget | ISAP | 113.24closeSep 04 '26 | 90.92 ● | 90.57181.10 | -4.4%APY -10.6% | -19.7% |
| WhyHe calls Alibaba a clear 'super fat pitch' that he does not yet own and plans to buy the stock soon. |
Full summary 443 words
This is a short, single-name Trading Post written overnight and centered almost entirely on lululemon athletica (LULU). Burry says he has been working through LULU's latest 10-Q, conference-call transcript and 8-K press release and concludes that the business has clearly changed for the worse. His own intrinsic-value model, the IV15, fell dramatically under new assumptions: negative growth in China, negative growth in the US despite continued store expansion, a -9% global comparable-sales figure, leggings sales down roughly 20%, and operating margin guided down to the mid single digits. He also argues the headline earnings 'beat' was in fact a big miss once a one-time (or two- or three-time) tariff benefit is stripped out.
Despite that grim read, the post is a bullish one, and the reason is his view of management and the setup. A new CEO is arriving, and Burry 'smell[s] the stench of a kitchen sink left with no running water for far too long' — i.e., he thinks guidance has been deliberately loaded with bad news so the new executive starts from the lowest possible bar. He points to front-loaded marketing expense, guidance that excludes re-orders of products that are actually selling, and suspicious 'expense timing' language as tells. He frames the incoming CEO — hand-picked by a Chairwoman who, in his telling, silenced her biggest critic and chose a candidate who could not start for six months — as inheriting expectations set so low 'one can hardly trip' on them.
The valuation case follows. The company retired 7% of its shares at $142, which Burry notes is well below the per-share IV8-IV10 range he considers intrinsic value. He sees $1.3 billion of cash, no financial debt, and shares near $99 trading at just over 2x tangible book for a brand he believes still means something. Across every scenario in which LULU grows again, $99.50 is below his IV15 and offers potential returns of 15-20% annually over 15-20 years. He likens the bet to having wagered that Abercrombie & Fitch, Ralph Lauren, or LULU itself (back in 2017) would never grow again — bets that would have been wrong.
Positioning is explicit: he states plainly that LULU is a 'fat pitch' under $100 assuming it ever grows again, and that he plans to buy more that morning under $100. The clear invalidation is embedded in his own conditional — the thesis rests on LULU eventually returning to growth; a permanent structural decline of the brand would break it. He closes by naming the only other 'clear super fat pitches' in his universe: JD, which he already owns, and Alibaba, which he does not yet own but plans to buy soon.
Key metrics
- LULU global comp -9%; leggings sales down ~20%
- Operating margin guided to mid single digits
- Negative growth in both China and the US despite store expansion
- Company retired 7% of shares at $142, below Burry's IV8-IV10 intrinsic range
- $1.3B cash and no financial debt
- Shares at ~$99 = just over 2x tangible book value
- $99.50 below his IV15; est. 15-20% annual returns over 15-20 years
- Buyback/entry: plans to add LULU under $100
Key insights
- Burry's IV15 model on LULU fell sharply after the latest 10-Q/call/8-K: -9% global comp, leggings -20%, and mid-single-digit operating-margin guidance.
- He argues the headline EPS beat was really a miss once a one-time tariff benefit is excluded.
- The bull case is a 'kitchen sink' setup: he thinks guidance was deliberately loaded with bad news so the incoming CEO starts from the lowest possible bar.
- Tells he cites: front-loaded marketing expense, guidance excluding re-orders of products that are selling, and suspicious 'expense timing' language.
- Valuation: ~$99 is just over 2x tangible book, with $1.3B cash and no financial debt; the company bought back 7% of shares at $142, above where it trades now and below his IV8-IV10 range.
- Under scenarios where LULU grows again, $99.50 is below his IV15 and implies 15-20% annual returns over 15-20 years — he calls it a 'fat pitch' under $100 and plans to add.
- He compares the wager to betting ANF, Ralph Lauren, or LULU (in 2017) would never grow again — bets that proved wrong.
- Beyond LULU, the only other 'clear super fat pitches' he names are JD (owned) and Alibaba (plans to buy soon).
#034
Short Thoughts
Sep 03, 2026
Short Thoughts - Cassandra in Flames?
LULU, Burry's largest position, has fallen roughly 20% after a weak quarter he expected, and he treats the drop as a buying opportunity rather than a reason to exit. His logic: the stock is now well under his IV15 intrinsic value, the thesis is intact, and a new CEO arrives this month, with a private-equity takeout as a backstop. He frames enduring pain in a top holding as his M.O., citing Avanti in 2001 and recent Cassandra Unchained names (Adobe, Molina, Veeva) that crashed then rebounded sharply, versus Tailored Brands as a rare permanent loss. His decision rule at a ~20% drawdown is binary: buy more or sell, and here he chooses to buy aggressively under $100.
consumershoppinghealthsaas
Positions & performance
-13.4%avg return0% win · 1 tracked · L -13.4%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon · largest position; will buy aggressively under $100 | NASDAQS&P | ISAPT | 121.77closeSep 03 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -13.4% |
| WhyLULU is his largest position and is down ~20% from his last purchase after a bad quarter he had anticipated. He says it is now cheaper and trades well under IV15, his thesis holds, and a new CEO arrives this month; his rule at a 20% drawdown is to either buy more or sell, and he will buy aggressively under $100, waiting for a PE firm to take him out if nothing else works. |
Full summary 400 words
This is a short, reflective Short Thoughts post written the evening of a sharp LULU drop, prompted by readers asking Burry what it feels like to lose money and to review Tailored Brands, a position that ended in a permanent loss of capital. He calls Tailored Brands a mistake made worse by that permanence: a small-cap where owned land and buildings might have hidden value, but management chose bankruptcy over the liquidity he believed existed. He takes few lessons from it and moves on with a characteristic shrug: "So let it be written. So it was done."
The heart of the post is lululemon (LULU), which he calls the trickster in his portfolio and, this time, his largest position. He asks whether a top position has ever cost him this dearly and answers that it is practically his M.O. He recounts Avanti, which he bought at $12 in June 2001 after its top five executives were convicted, rode all the way down into the $2s, and then watched get acquired at $22/share around Christmas, with the deal negotiated near the lows. He also invokes The Big Short as a trade that almost ate him alive before it paid off, and points to recent Cassandra Unchained names, Adobe, Molina and Veeva, that crashed hard and rebounded dramatically.
His message on holding is about grip rather than mere patience: patience alone does not win the rodeo, and eight seconds is never enough, so the hand may gangrene while he keeps holding. He had told CU readers he expected a bad quarter into the earnings release, and it arrived. LULU is now down about 20% from his last purchase, trading under $100 after hours.
Positioning follows directly. He continues to hold LULU and states his rule at a 20% drawdown: either buy more or sell. He chooses to buy, aggressively under $100, because the stock is cheaper, well under his IV15 intrinsic value, his thesis holds, and a new CEO arrives this month; failing all else, he is content to wait for a private-equity firm to take him out. What would invalidate the view is a break of the thesis rather than price alone. He closes with domestic color, carpet beetles in his home office, a long unfed day, a need for Mexican food, and a promise that the detour will not delay a batch of articles that are coming, like the beetles.
Key metrics
- LULU is Burry's largest position
- LULU down ~20% from his last purchase; trading under $100 after hours
- Stated plan: buy aggressively if LULU trades under $100
- LULU trades well under IV15 (his intrinsic-value measure)
- New LULU CEO arrives September 2026
- Avanti (2001): bought at $12, fell to the $2s, acquired at $22/share
- Adobe, Molina and Veeva cited as prior CU names that crashed then rebounded
- Tailored Brands cited as a rare permanent loss of capital via bankruptcy
#033
Trading Post
Aug 28, 2026
Trading Post August 28, 2026
After a near-doubling in Veeva and a collapsed deal in PayPal, Burry books his first real sells of the week. He trims Veeva to a ~3% 'house money' stake because its valuation has stretched past his IV15 fair value, harvesting capital to redeploy into more spring-loaded ideas, while keeping a small position since it is still cheap on IV8. He holds a reduced ~3% PayPal position after the stock fell into the $53s, planning to add in the mid-$40s. The through-line is valuation discipline and after-tax return comparison: he sells only when a name is uncomfortably expensive and better opportunities exist, not to chase long-term capital-gains treatment.
financefintechhealthsaas
Positions & performance
+9.2%avg return100% win · 2 tracked · L +9.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LPYPL:NASDAQ financefintech PayPal Holdings HOLDcommon · ~3% position | NASDAQS&PBitget | ISPT | 53.0his fillAug 28 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | +10.8% |
| WhyA pending deal fell apart and the stock dropped nearly $8 into the $53s; because he had already reduced the position to balance exactly this kind of risk, he holds the remaining ~3% stake and plans to add if it falls to the mid-$40s. |
| LVEEV:NYSE health Veeva Systems TRIMcommon · ~3% position | S&PBitget | IT | 276.69closeAug 28 '26 | 298.04 ● | 148.05325.99 | +18.0%APY +87.4% | +7.7% |
| WhyVeeva nearly doubled over the last few months and now trades well above his buy price and above what he considers fair on IV15, so he sold most of the position and is 'playing with the house's money'; he keeps a ~3% stake because it is not yet expensive relative to its IV8 and he still believes in the business and its competitive position. |
Full summary 445 words
This Trading Post marks Burry's first real sells of the week — trimming what had been two of his most significant positions rather than exiting them outright. He frames the decision inside the valuation process he has laid out repeatedly since Cassandra Unchained launched nine months ago, anchored on IV15 (his 15x-normalized intrinsic-value gauge) read alongside a company's competitive position and enduring return on invested capital. He points readers back to his 'Software & Payments Stocks Part II' work, noting the Cassandra software-and-payments universe is now 50 names (the SW50) after adding Synopsys, Cadence, JFrog and Samsara.
Before detailing the trades he makes a point about the mechanics of selling: you never realize the screen price, because taxes take 20-50% of the gain. That after-tax reality has to be weighed against other opportunities — but, he cautions, it should not tempt you into holding a position you are uncomfortable with just to reach a long-term capital-gains rate, which he calls 'a good way to get burned.'
The first sell is Veeva (VEEV). The stock nearly doubled over the last few months and now trades well above his buy price and above what he views as fair. He cuts it to roughly a 3% position, describing himself as 'playing with the house's money' on a name he still likes. He plans to keep the smaller stake because Veeva is not yet expensive relative to its intrinsic value on an IV8 basis and because he believes in the business and its competitive position; information is never perfect, he adds, and it generally pays to hold very good businesses. The harvested capital is earmarked for redeployment into 'more spring-loaded positions.'
The second name is PayPal (PYPL), where 'the deal fell apart' and the stock traded down nearly $8 into the $53s. Here he had already reduced the position to balance precisely this kind of event risk, so it too sits at about 3%. He holds at current levels and states he will add if PYPL falls into the mid-$40s.
Positioning follows directly from the valuation-and-tax logic: sell into strength when a name outruns fair value and better ideas exist (Veeva), and hold — with a defined lower add point — a name that has been de-risked and re-rated cheaper by a broken catalyst (PayPal). What would invalidate the approach is straightforward: a further Veeva re-rating that still leaves it cheap on IV15 would argue against trimming, and a PayPal recovery above his hold zone without reaching the mid-$40s simply leaves the smaller position in place. It is a housekeeping post — no new shorts, no fresh longs — that illustrates his sell discipline more than it adds exposure.
Key metrics
- Veeva (VEEV) nearly doubled over the last few months
- VEEV trimmed to a ~3% position ('house's money')
- PYPL fell nearly $8 to the $53s after a deal collapsed
- PYPL held at a ~3% position; would add in the mid-$40s
- Taxes take 20-50% of the gain on a sale, factored into sell decisions
- Cassandra Unchained software & payments universe now 50 stocks (the SW50)
- Uses IV15 as the primary valuation guide, plus IV8 for a nearer-term read
- SW50 recently added Synopsys, Cadence, JFrog and Samsara
Key insights
- Burry's first real sells of the week are trims, not exits: both Veeva and PayPal are cut to ~3% 'house money' stakes rather than closed.
- Veeva was sold down after nearly doubling and running past his IV15 fair value, but he keeps a stake because it is still cheap on IV8.
- He explicitly warns against holding an uncomfortable position just to reach long-term capital-gains treatment — 'a good way to get burned.'
- Taxes (20-50% of the gain) are treated as a first-class input to sell decisions, compared against the after-tax return of alternative ideas.
- PayPal fell nearly $8 into the $53s after a deal collapsed; he had pre-emptively reduced it and now holds ~3%, with a stated add point in the mid-$40s.
- Capital harvested from Veeva is earmarked for 'more spring-loaded positions' — signaling he sees better risk/reward elsewhere.
- The Cassandra software & payments universe (SW50) now spans 50 names after adding Synopsys, Cadence, JFrog and Samsara.
- No new longs or shorts this post — it is a discipline/housekeeping note showcasing his valuation framework in action.
#032
Trading Post
Aug 27, 2026
Trading Post August 27, 2026
Burry treats this as a real-time Trading Post. On Nvidia he monetizes a very short-term call hedge that worked (up 50-60% into earnings), legging out because the event has passed while his long-term bearish thesis on the stock is intact. On Build-a-Bear, a ~28% earnings-day drop trips his 20%-down discipline; rather than react, he holds his existing long, points to the stock trading below historic IV15 as valuation support, and defers any add until he can read the 10-Q — arguing the brand-dilution/wholesale problems that caused the guide-down are fixable under new management.
aisemiconsumershopping
Positions & performance
-14.2%avg return50% win · 2 tracked · L -14.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LBBW:NYSE consumershopping Build-a-Bear Workshop HOLDcommon | — | S | 28.44closeAug 27 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | +3.4% |
| WhyStock is down ~28% intraday on a weak earnings release and call (commercial revenue guide cut to flat from +20%, Walmart wholesale non-renewal, international franchising down 33%), which triggers his 20%-down rule. He continues to hold the position he took recently, notes it now trades below historic IV15, but wants to see the not-yet-filed 10-Q before deciding whether to add; his 20% rule 'does not say I must act immediately or rashly.' |
| LNVDA:NASDAQ aisemi Nvidia TRIMcalls · half | NASDAQS&PBitget | ISPTC | 227.98closeAug 27 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | -31.7% |
| WhySold half of an NVDA call position established less than 24 hours earlier purely as a hedge into earnings; the calls are up 50-60% and 'did their job.' He is legging out (alerting CU subscribers after the first leg) and expects to sell the remaining half, since he sees no good reason to keep them and remains long-term bearish on the stock. |
Full summary 460 words
This is a same-day Trading Post in which Burry addresses 'the issues of the day in real-time,' covering two names. The first is Nvidia, a continuation of his August 26 paywalled post 'Nvidia, Friends, and the Rebel Alliance.' He reveals he sold half of an NVDA call position that he had put on less than 24 hours earlier purely as a hedge into the earnings event; the calls are up 50-60% and, in his words, 'did their job.' He is legging out deliberately, having alerted Cassandra Unchained subscribers as soon as the first leg was done, and expects to sell the remaining half because he has 'no good reason to keep them.' He concedes there is 'little standing in the way of further stock appreciation' and that the print did not damage — indeed in places furthers — his long-term bearish thesis, but 'no head shots' landed and the bullish narrative 'can continue for a fraction of an eon.' He attaches Nvidia's earnings release, transcript and 10-Q but declines to get into the earnings weeds in a Trading Post.
The more urgent matter is Build-a-Bear Workshop (BBW), which is 'having a day' — the stock is down about 28% intraday after a weak earnings release and a conference call that 'did not help.' The move triggers his 20%-down rule ('do something or get off the pot'). His biggest concerns are the commercial-segment revenue guide cut to roughly flat from +20% previously, a Walmart wholesale non-renewal, and international franchising down 33% with partner operators also declining. He quotes CFO Voin Todorovic tying the commercial cut to the 'traditional wholesale mess,' a Q2 miss driven by summer trend-product performance and persistent traffic pressure. Burry pushes back on the 'kitchen sink' read: there were no impairment charges or inventory write-downs.
His diagnosis is brand dilution: a company built around the ceremony of building a bear had been selling generic pre-made stuffed animals to chase the trendy Squishmallows, boosting sales temporarily while watering down the brand. That, he argues, is why there is a new CEO as of May and why the Chief Growth Officer responsible for the 'plushy push' was fired without cause; he finds it 'curiouser' that the growth plan (50+ new locations, ~$25M capex for a micro cap) remains intact. Positioning: he continues to hold the BBW long he took recently and notes it now trades below historic IV15, but is awaiting the 10-Q (expected today or tomorrow) before deciding whether to add — his 20% rule does not require rash action. He says he will fold all of this into his full 'Burry Bullish on Build-a-Bear' post once the filing lands. The clear invalidation risk would be a 10-Q that reveals the quarter was worse than the clean-looking release suggested.
Key metrics
- Sold half of NVDA calls up 50-60%; position was established <24h earlier as an earnings hedge
- Expects to sell the remaining half of the NVDA calls (no good reason to hold; remains long-term bearish)
- BBW down ~11 points (~28%) intraday, triggering his 20%-down rule
- BBW commercial/wholesale segment guide cut to flat from +20% previously
- BBW international franchising down 33%; partner operators also declined
- Walmart wholesale relationship non-renewed
- New BBW CEO as of May 2026; Chief Growth Officer (behind the Squishmallows-chasing 'plushy push') fired without cause
- BBW growth plan intact: 50+ new locations, capex target ~$25M (a micro cap)
- No impairment charges or inventory write-downs in the quarter; 10-Q not yet filed (expected today/tomorrow)
- BBW trades below historic IV15
Key insights
- The NVDA calls were an explicit, sub-24-hour earnings hedge — not a directional bull bet; Burry booked +50-60%, sold half, and plans to exit the rest, keeping his long-term bearish stance on the stock.
- He openly concedes 'little standing in the way of further stock appreciation' in NVDA and that the print did no thesis damage ('no head shots'), a notably candid acknowledgment that the bull narrative can persist.
- BBW's ~28% drop tripped his 20%-down rule, but he stresses the rule mandates a decision, not immediate action — he is holding, not panic-selling.
- The bear case that caused the guide-down is, in his read, self-inflicted brand dilution: chasing Squishmallows with generic pre-made plush undercut the 'build-a-bear' experience for a temporary sales pop.
- Management change is central to his bullish setup: new CEO since May and the Chief Growth Officer behind the plush push fired without cause signal a strategy reset.
- He flags a tension — the growth plan (50+ new stores, ~$25M capex) stays intact even as commercial revenue is guided to flat — as something the 10-Q needs to justify.
- Valuation anchor: BBW now trades below historic IV15 (his intrinsic-value screen), which is why he is inclined to add rather than exit — pending the 10-Q.
- He explicitly separates a clean-looking earnings release (no impairments or write-downs) from the guidance shock, resisting the 'kitchen sink' interpretation until the full filing confirms it.
#031
Trading Post
Aug 26, 2026
Trading Post August 26, 2026 - Nvidia, Friends, and the Rebel Alliance
Burry believes Nvidia's price is incongruent with the market's own narrative: on the surface it looks wildly undervalued (low PE, monopoly rents), yet the stock is only treading water, which he reads as the market quietly disbelieving the story. His analysis points to a short runway for Nvidia's monopoly power, falling margins, and a capex/investment focus that will starve shareholder distributions — so he is fundamentally short NVDA and the broader AI-infrastructure complex (ORCL, PLTR, NBIS), plus cyclical CAT. Because that short-and-put position is so large, he bought December NVDA calls as a pure earnings-day hedge, not a directional long. On the other side of the book, his 'Rebel Alliance' of longs — Birkenstock and the GSEs Freddie Mac and Fannie Mae — got added to.
aineocloudaisemienergygridsaas
Positions & performance
+16.0%avg return86% win · 7 tracked · L -12.1% · S +27.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SNBIS:NASDAQ aineocloud Nebius ADDcommon | NASDAQBitget | ISPT | 213.93closeAug 26 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +49.2% |
| WhyTopped off the short on this AI/GPU-cloud name, consistent with his view that spending through the top of the bubble will hurt these businesses. |
| SCAT:NYSE energygrid Caterpillar ADDcommon | S&PBitget | IAPT | 821.93closeAug 26 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +30.6% |
| WhyAdded to the short — a cyclical/industrials expression alongside the AI-bubble shorts, part of the >21% short stock position. |
| SNVDA:NASDAQ aisemi Nvidia ADDcommon | NASDAQS&PBitget | ISPTC | 209.66closeAug 26 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +25.8% |
| WhyShorted a little more NVDA this week: he argues the stock's theoretical value is far below market, that its monopoly power has a short runway, that margins will fall, and that capex/investment focus means it will not distribute enough to shareholders. |
| SORCL:NYSE aineocloud Oracle ADDcommon | S&PBitget | ISAPTC | 148.87closeAug 26 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +20.1% |
| WhyTopped off the short this week as part of his bet against the AI-infrastructure/hyperscaler capex bubble; the combined short stock book is now over 21% of portfolios, excluding puts. |
| SPLTR:NASDAQ saas Palantir ADDcommon | NASDAQS&PBitget | ISPTC | 177.5closeAug 26 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +10.1% |
| WhyAdded to the short as one of the richly valued AI names he is positioned against; part of the >21% short stock book. |
| LBIRK Birkenstock ADDcommon · full position | — | IA | 35.0his fillAug 26 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +1.7% |
| WhyAdded in the mid-$30s to bring it up to a full position — part of his 'Rebel Alliance' of longs held against the large short book. |
| LNVDA:NASDAQ aisemi Nvidia NEWDecember calls, strike mid-to-high $200s (low implied vol, premium paid single digits) | NASDAQS&PBitget | ISPTC | 209.66closeAug 26 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | -25.8% |
| WhyAn earnings-day hedge, not a bullish bet: he bought the calls only because he carries a very large NVDA short-plus-put position (3.5-4% of portfolios). The call premium is entirely offset by the short, and then some. He notes NVDA has fallen more than risen after recent earnings and that the option is a coin-flip play on a single day's event. |
| LFMCC Freddie Mac ADDcommon · large position (with FNMA) | NASDAQ | — | —Aug 26 '26 | — | — | — | n/a |
| WhyAdded in the mid-$5s; together with Fannie Mae this is a large GSE position, a continuation of his long-running housing/GSE recap thesis. |
Full summary 428 words
This paywalled Trading Post, published the day of Nvidia's earnings, is a snapshot of Burry's book on a 'special day.' The centerpiece is a trade he flags up front: he bought December NVDA calls, struck in the mid-to-high $200s with relatively low implied volatility, paying single-digit premium. He is emphatic that this is a hedge, not a bullish bet — he would not do it if he did not already carry such a large NVDA short and put position, which he sizes at 3.5-4% of his portfolios. The call premium, he says, is entirely offset by the short, and then some. He concedes his record buying options around specific earnings dates is 'hit and miss,' exactly what one should expect when wagering on a single day's event, and notes NVDA has fallen more than risen after recent earnings.
The bulk of the note restates his bearish Nvidia thesis. Stock markets are supposed to be discounting machines, and on his analysis Nvidia's theoretical value — future cash flows discounted back — sits well below today's price. He argues its monopoly power has only a short runway from here, which means falling earnings margins, and that Nvidia will increasingly divert cash into capex and investments to diversify and sustain sales growth rather than distribute to shareholders. He points readers to his July 8th long 'Short Thoughts' for the fuller case, references a revenue chart with a 100% year-over-year growth line, and wraps the argument in his usual color, casting Nvidia as 'the mafia don at the center of a Byzantine grift on society.' His conclusion for the day is that the market 'does not believe,' and that a post-earnings pop, if it comes, would only temporarily restore that belief.
On positioning, the 'Rebel Alliance' longs were added to: Birkenstock in the mid-$30s, now a full position, and Freddie Mac in the mid-$5s, which together with Fannie Mae constitutes a large GSE position. On the short side he 'topped off' this week, adding to Oracle, Palantir, Nebius, NVDA and Caterpillar, bringing the total short stock position to over 21% of the portfolios — and that figure excludes puts, which he describes as far out-of-the-money, distant in expiry, and an asymmetric, leveraged, directional strategy with minimal current market value. He closes by teasing that his 'next big post' is very close. What would invalidate the view is straightforward from his framing: a durable re-acceleration in Nvidia's margins and monopoly economics, or the market genuinely re-embracing the AI-capex narrative, would undercut both the NVDA short and the broader complex he is positioned against.
Key metrics
- NVDA short-plus-put position is 3.5-4% of portfolios
- Bought December NVDA calls, strike mid-to-high $200s, low implied vol, premium paid 'single digits'
- Total short stock position over 21% of portfolios, not including puts
- Added BIRK in the mid-$30s to a full position
- Added FMCC in the mid-$5s; large GSE position with FNMA
- Topped off shorts this week in ORCL, PLTR, NBIS, NVDA and CAT
- References NVDA revenue chart with a 100% year-over-year growth line
- Puts described as far out-of-the-money, distant expiry, minimal current market value
Key insights
- The December NVDA calls are explicitly a hedge on an earnings-day coin flip, fully offset by his short-plus-put position of 3.5-4% of portfolios — not a change of heart on Nvidia.
- His core NVDA bear case: theoretical (DCF) value well below market, a short runway for monopoly power, falling margins, and capex crowding out shareholder distributions.
- He reads NVDA merely 'treading water' despite a low PE as evidence the market itself does not believe the bull narrative.
- The short book is large and broadening: total short stock position now exceeds 21% of portfolios, excluding puts.
- This week he added to shorts in ORCL, PLTR, NBIS, NVDA and CAT — the AI-infrastructure/hyperscaler-capex complex plus a cyclical.
- On the long side, the 'Rebel Alliance' grew: BIRK added in the mid-$30s to a full position, FMCC added in the mid-$5s as part of a large GSE position with FNMA.
- His puts are deliberately far out-of-the-money and distant in expiry — asymmetric, leveraged and directional with minimal current mark.
- He admits a mediocre track record trading single-day earnings options, framing this NVDA call purely as insurance around a large short.
#030
Trading Post
Aug 21, 2026
Trading Post & Short Thoughts to the Max, August 18, 19, 20, 2026
Burry argues US GDP is dangerously dependent on a single, debt-financed AI data-center buildout even as the economics that justify it are eroding. Citing new academic work on 'intelligence per watt,' he says small local/open-weight inference models now handle the vast majority of queries far more cheaply, so billable inference will migrate off hyperscaler GPUs and expose a duration mismatch in six-year server depreciation. That undercuts the returns underwriting the buildout, which he likens to pre-GFC leverage, fraud and off-balance-sheet financing (now via captive insurers). He positions accordingly: shorting AI-infrastructure and richly valued newer names (CRWV, MU, SOXX, PLTR) plus far-OTM index/NVDA/PLTR puts, while building large longs in cheap, misunderstood consumer and healthcare businesses (LULU, MELI, Temple & Webster, SFM, BIRK, BBW, MOH).
aimemoryaineocloudairobotaisemiconsumerhealthconsumershoppingfinancefintechhealthindexusmag5saas
Positions & performance
+9.5%avg return67% win · 15 tracked · L -7.2% · S +28.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDcommon | NASDAQBitget | I | 520.05closeAug 21 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +72.9% |
| WhyShorted more of the semiconductor index ETF as a broad expression of the AI-capex/compression thesis. |
| SCRWV:NASDAQ aineocloud CoreWeave NEWcommon | NASDAQBitget | ITC | 87.85closeAug 21 '26 | 49.76 ● | 40.58138.25 | +43.4%APY +345.4% | +43.4% |
| WhyAdded a new short earlier in the week; part of his AI-infrastructure/compression bear case, and a name that has not been public very long. |
| SNVDA:NASDAQ aisemi Nvidia HOLDputs, 2027 expiries, far out of the money · part of ~6.5% put book | NASDAQS&PBitget | ISPTC | 214.72closeAug 21 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +27.5% |
| WhyHolds far-OTM 2027 puts on Nvidia as part of a put book worth ~6.5% of the portfolio, betting the debt-financed data-center buildout is over-earning relative to durable inference demand. |
| SMU:NASDAQ aimemory Micron Technology ADDcommon | NASDAQS&PBitget | ISAPTC | 966.78closeAug 21 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +25.5% |
| WhyIncreased the size of the existing Micron short; a long-public memory/semis name tied to his view that AI-hardware demand will disappoint as inference migrates to local chips. |
| LVEEV:NYSE health Veeva Systems TRIMcommon · 4.8% position | S&PBitget | IT | 247.9closeAug 21 '26 | 298.04 ● | 148.05325.99 | +18.0%APY +87.4% | +20.2% |
| WhyShaved down to a 4.8% position; he does not currently expect to continue selling. |
| LMOH:NYSE health Molina Healthcare ADDcommon · 9% position | — | ITC | 201.0his fillAug 21 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +16.0% |
| WhyAdded at $201 to a 9% position; a year-plus volume swell supports his view that the stock and business should be bottoming. |
| SPLTR:NASDAQ saas Palantir Technologies ADDcommon | NASDAQS&PBitget | ISPTC | 179.94closeAug 21 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +11.4% |
| WhyAdded to the Palantir short, a name public 'not too very long'; total shorts excluding puts are now a little over 21% of the portfolio. |
| SPLTR:NASDAQ saas Palantir Technologies HOLDputs, 2027 expiries plus a portion expiring December 2026, far out of the money · part of ~6.5% put book | NASDAQS&PBitget | ISPTC | 179.94closeAug 21 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +11.4% |
| WhyIn addition to the common short, holds far-OTM Palantir puts — mostly 2027 expiries with a portion expiring December 2026 — within the ~6.5% put book. |
| SQQQ:NASDAQ indexus Invesco QQQ (Nasdaq-100 ETF) HOLDputs, 2027 expiries, far out of the money · part of ~6.5% put book | NASDAQBitget | SC | 713.44closeAug 21 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +7.2% |
| WhyHolds far-OTM 2027 QQQ puts as an index hedge against the AI-concentrated Nasdaq within the ~6.5% put book. |
| LBIRK Birkenstock ADDcommon · 5.2% position | — | IA | 35.0his fillAug 21 '26 | 35.61 ● | 29.9350.73 | +1.7%APY +7.4% | +1.7% |
| WhyIncreased his small stake to a 5.2% position as the stock fell into the $35s; a longtime personal favorite with a durable, differentiated product and a write-up he says is coming soon. |
| LPYPL:NASDAQ financefintech PayPal TRIMcommon · 4% position | NASDAQS&PBitget | ISPT | 61.55closeAug 21 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | -4.6% |
| WhySold down to a 4% position with the stock trading above its deal price; he thinks a better deal is possible but is wary of handicapping the outcome and does not expect to sell further. |
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon · 17.4% position | NASDAQS&P | ISAPT | 115.0his fillAug 21 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -8.3% |
| WhyAdded in the $115s, taking it to a very large 17.4% position ahead of earnings on Thursday, September 3 (after the close). |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon · 12% position | NASDAQBitget | ISAPC | 1890.0his fillAug 21 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -22.6% |
| WhyAdded in the $1890s to a 12% position; one of his four largest longs. |
| LBBW:NYSE consumershopping Build-A-Bear Workshop NEWcommon · full position | — | S | 38.89closeAug 21 '26 | 29.42 ● | 23.1972.19 | -24.4%APY -68.6% | -24.4% |
| WhyOpened a new full long in the mid-to-upper $30s, calling it a remarkably simple and yet misunderstood business, with a dedicated post to follow. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADDcommon · 5.2% position | NASDAQ | IAPT | 78.0his fillAug 21 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -35.3% |
| WhyAdded in the high $70s to a full 5.2% position; a simple story where he thinks private equity 'has to be sniffing around' and the current valuation will not last. |
| LTPLWF Temple & Webster (US pink sheets) ADDcommon · part of a now-large position | — | — | —Aug 21 '26 | — | — | — | n/a |
| WhyBought the US pink-sheet TPLWF listing in the low $3s over the last few days; 'scooping with both hands' — Temple & Webster is now one of his four largest positions (~9%). |
| LTPW AU Temple & Webster (ASX) ADDcommon · part of a now-large position | — | — | —Aug 21 '26 | — | — | — | n/a |
| WhyAlso added to his Australian-listed TPW AU shares alongside the US pink sheets; the combined Temple & Webster stake is now ~9%, a large position. |
Full summary 509 words
This mid-August Trading Post opens not with trades but with a macro warning. Burry riffs on an Apollo/Torsten Slok 'Daily Spark' chart showing that when the Fed began hiking in 2022, rate-sensitive sectors like office construction rolled over quickly, yet data-center construction kept surging because hyperscalers judged AI returns would exceed a higher cost of capital. Burry reframes the same chart darkly: private non-residential construction excluding data centers fell 7.9% year over year in June, so GDP growth is 'really counting on' the data-center buildout. He draws an explicit GFC analogy — fraud, leverage, and over-dependence on a single factor — and flags complex off-balance-sheet SPV financing and a newer twist, captive offshore and onshore insurers soaking up risky or fraudulent paper, echoing warnings in the BIS 2026 annual report.
The core of the note is his 'compression' thesis. Burry argues American businesses are quietly moving to small language inference models built on open-source or open-weight foundations (Chinese if cheapest), driven by cost. He leans on an August 7 paper, 'Intelligence per Watt' by Jon Saad-Falcon, Avanika Narayan and others, which found that 88.7% of single-turn chat and reasoning queries can be handled by small local models, that intelligence-per-watt improved 5.3x over two years (3.1x from models, 1.7x from hardware), that local-model coverage rose from 23.2% in 2023 to 71.3% in 2025, and that hybrid local-cloud routing cuts energy, compute and cost 60–80%. He notes OpenAI (a GPT-5.x internal router), Perplexity, Intel (on-chip inference) and Apple (an internal router since 2024) are already routing tasks, calling cost-cutting 'insanely powerful and durable.'
He then ties this to hardware economics and philosophy. The efficiency gains create a 'duration mismatch': six-year server depreciation is far too long because inference load will increasingly land on local chips, undermining the billable-inference-per-GPU assumptions that the debt-financed hyperscaler buildout is underwritten on. He revisits 'Ballard's Test' from his February post — an entity lacks true understanding until reason is demonstrated without language — to argue LLMs merely simulate reason and AGI may never arrive, and that most businesses don't need it anyway.
Positioning follows directly. Longs: he added Birkenstock to a 5.2% position in the $35s, opened a new full Build-A-Bear position in the mid-upper $30s, bought Temple & Webster (TPLWF pink sheets in the low $3s plus more TPW AU) to a ~9% position, added Sprouts to 5.2% in the high $70s (expecting PE interest), took lululemon to a very large 17.4% in the $115s ahead of Sept 3 earnings, and added Molina to 9% at $201. He trimmed PayPal to 4% (trading above its deal price) and Veeva to 4.8%, and added MercadoLibre in the $1890s to 12%. Shorts: total shorts excluding puts now exceed 21% of the book — a new CoreWeave short plus larger Micron, SOXX and Palantir shorts — with far-OTM puts (~6.5% of the portfolio) on NVDA, PLTR and QQQ, mostly 2027 expiries. He is mid-high-teens percent on margin, plans to bring it back toward 0–10%, and repeats his standard caution that most people should never short or use margin.
Key metrics
- BIRK increased to a 5.2% position as the stock fell into the $35s
- BBW new full long in the mid-to-upper $30s
- TPLWF bought in the low $3s; Temple & Webster now a ~9% position
- SFM to a full 5.2% position in the high $70s
- LULU added in the $115s to a very large 17.4% position; earnings Sept 3
- MOH added at $201, now a 9% position
- PYPL trimmed to a 4% position (trading above deal price); MELI added in the $1890s to 12%
- Total shorts excluding puts a little over 21% of the portfolio; puts ~6.5% of the portfolio
- Local-model coverage rose from 23.2% (2023) to 71.3% (2025); 88.7% of single-turn queries handled locally
- Intelligence-per-watt improved 5.3x in two years (3.1x model, 1.7x hardware); private non-residential construction ex-data-centers fell 7.9% YoY in June
Key insights
- Burry's central macro claim: with private non-residential construction ex-data-centers down 7.9% YoY in June, US GDP growth is riding almost entirely on the AI data-center buildout — a single-factor dependence he compares to pre-GFC housing.
- He flags captive offshore/onshore insurers absorbing risky or fraudulent paper and off-balance-sheet SPV financing as the hidden leverage in the AI buildout, echoing the BIS 2026 annual report.
- The 'Intelligence per Watt' paper is his evidence base: 88.7% of single-turn queries handled by small local models, IPW up 5.3x in two years, local coverage up from 23.2% (2023) to 71.3% (2025).
- His 'duration mismatch' point: six-year server depreciation is too long because inference migrates to local chips, breaking the billable-inference-per-GPU math behind debt-financed hyperscaler capex.
- Shorts are concentrated in newer, richly valued AI-infra names (CoreWeave new short, added Palantir) plus long-public semis (Micron, SOXX); total shorts ex-puts now just over 21% of the portfolio.
- The put book (~6.5% of the portfolio) is far-OTM NVDA, PLTR and QQQ, mostly 2027 expiries with a portion of PLTR puts expiring December 2026.
- Long book is a cheap-consumer/healthcare barbell: LULU (17.4%), MELI (12%), Temple & Webster (~9%), MOH (9%), SFM (5.2%), BIRK (5.2%), plus a new BBW full position.
- He is running mid-high-teens percent margin to accommodate a larger long book against the big short, but says he'll bring margin back toward 0–10% and warns retail investors away from shorting and margin.
#029
Community
Aug 18, 2026
Acres of Diamonds
This is Cassandra Unchained's August 2026 Charity of the Month, not an investment post. Burry vets Acres of Diamonds, a small Duvall, Washington charity that houses homeless mothers and children, by pulling its Form 990 and applying the same forensic-disclosure lens he uses on companies. He concludes the filings are clean and non-disqualifying, program efficiency (79%) clears every major watchdog threshold, and donor support is genuine at 94.5% of revenue. As promised, 5% of Cassandra Unchained's August gross revenue will be donated to the charity.
Positions & performance
No single-stock positions stated in this report.
Full summary 264 words
This is Cassandra Unchained's monthly charity post for August 2026, not a trading note. Burry highlights Acres of Diamonds, a small charity in Duvall, Washington that takes in homeless mothers and their children, houses them, and works with them for as long as necessary to get back on their feet — 44 families and 50 children last year, over three decades of operation. The charity was suggested by one of his own readers.
Characteristically, Burry vets it the way he vets a company. He pulled the Form 990 and read Part IX, finding public support at 94.5% of revenue, fundraising costs of about $10.80 per $100 raised, and an executive director base salary of just $54,000 out of $3.3 million in total spending — roughly $50,000 per family served. Contributions have doubled over five years to $4 million in 2025. The 21% expense ratio looks a touch high at first, but he notes this is an operations-heavy charity with real overhead; the resulting 79% program efficiency clears CharityWatch (75%), Charity Navigator (70%), and the BBB standard (60%).
He frames the exercise around how hard homelessness charities are to vet — "sometimes it seems there are more galas than beds provided" — and says he looked through dozens of charities and was reminded how often he finds disqualifying disclosures; this one checks the boxes and is clean. As is now custom, 5% of Cassandra Unchained's gross revenue for August 2026 will be donated to Acres of Diamonds, and he asks readers to keep suggesting smaller, effective charities without celebrity sponsorship where donations move the needle.
Key metrics
- 44 families and 50 children housed last year
- Public support = 94.5% of revenue
- Fundraising cost ~$10.80 per $100 raised
- Executive director base salary $54,000
- ~$3.3 million total spent by the charity
- ~$50,000 spent per family
- Contributions doubled over 5 years to $4 million in 2025
- Expense ratio 21% (79% program efficiency)
- Clears CharityWatch 75%, Charity Navigator 70%, BBB 60% standards
- In operation for three decades
Key insights
- Charity-of-the-month post, not an investment thesis — no tickers or positions.
- Burry applies forensic-filing scrutiny to charities, reading the Form 990 Part IX just as he would a 10-K.
- Acres of Diamonds: 94.5% public support, ~$10.80 fundraising cost per $100 raised, $54k ED salary.
- 79% program efficiency clears CharityWatch (75%), Charity Navigator (70%) and BBB (60%) thresholds.
- Contributions doubled in five years to $4 million in 2025; ~$50,000 spent per family served.
- He deliberately seeks small, effective charities without celebrity sponsorship where exposure moves the needle.
- 5% of Cassandra Unchained's August 2026 gross revenue will be donated to the charity.
#028
Trading Post
Aug 13, 2026
Trading Post August 13, 2026: Welcome to the Sausage Factories
Burry argues the AI trade is being financed circularly — AI companies funding each other to buy from each other, so injected capital returns as revenue and Wall Street rewards the resulting growth — and that the capital funding it is increasingly debt, which puts the bubble on a clock with a concrete cost of capital. His base case for the turn is 2028, when compute becomes too much, though he notes markets historically catch on well before that. In the meantime the rally has tipped his short book from roughly breakeven into a loss, so he is de-grossing: selling volatility where it still exists, buying cheap volatility where he can, and covering the shorts that have already paid. The signature trade is rotating out of expensive-IV SOXX puts into a larger, longer-dated QQQ put position at 25 IV, betting that compressed index volatility spreads on further rallies. He simultaneously trims every long to raise ~12% cash, explicitly preparing for a larger market fall while keeping his short bias intact.
aimemoryaineocloudairobotaisemiconsumerhealthconsumershoppingenergygridfinancefintechhealthindexussaas
Positions & performance
+12.0%avg return63% win · 19 tracked · L -3.7% · S +33.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF HOLDcommon · about 7% of the portfolio, the largest short | NASDAQBitget | I | 550.74closeAug 13 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +74.4% |
| WhyLeft the outright SOXX short in place as the largest position in the short book even after exiting the SOXX puts, keeping the semiconductor bearish exposure via shares rather than options. |
| SNBIS:NASDAQ aineocloud Nebius Group HOLDcommon | NASDAQBitget | ISPT | 255.04closeAug 13 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +57.4% |
| WhyRetained as part of the ranked short book behind SOXX and Micron, an AI-neocloud expression of the circular-financing thesis. |
| SCAT:NYSE energygrid Caterpillar TRIMcommon · 25% of the short position | S&PBitget | IAPT | 854.6closeAug 13 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +33.2% |
| WhyTrimmed a quarter of the Caterpillar short as part of the broad reduction in gross exposure. |
| SNVDA:NASDAQ aisemi Nvidia HOLDputs | NASDAQS&PBitget | ISPTC | 225.3closeAug 13 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +30.9% |
| WhySpared the Nvidia puts during the de-grossing, keeping them alongside the Palantir puts as core AI-bubble expressions. |
| SMU:NASDAQ aimemory Micron Technology ADDcommon · second-largest short after SOXX | NASDAQS&PBitget | ISAPTC | 1000.0his fillAug 13 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +28.0% |
| WhyMade one move counter to the de-grossing trend, increasing the Micron short on its move back up near $1,000. |
| LHCA:NYSE health HCA Healthcare TRIMcommon · second tier, ~1pp below the top group | S&P | IAPT | 411.74closeAug 13 '26 | 511.78 ● | 353.99556.52 | +41.4%APY +174.8% | +24.3% |
| WhyTrimmed with the rest of the long book; sits alongside the other second-tier healthcare and consumer names. |
| SORCL:NYSE aineocloud Oracle HOLDcommon | S&PBitget | ISAPTC | 156.22closeAug 13 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +23.8% |
| WhyOracle's puts remain very expensive, so he continues to express the short through the shares rather than options. |
| LVEEV:NYSE health Veeva Systems TRIMcommon · second tier, ~1pp below the top group | S&PBitget | IT | 252.64closeAug 13 '26 | 298.04 ● | 148.05325.99 | +18.0%APY +87.4% | +18.0% |
| WhyReduced with the rest of the long book in the de-grossing; second-tier sizing separated from peers by basis points. |
| LMOH:NYSE health Molina Healthcare TRIMcommon · second tier, ~1pp below the top group | — | ITC | 205.71closeAug 13 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +13.3% |
| WhyTrimmed as part of the across-the-board long reduction; remains in the second size tier. |
| LZTS:NYSE health Zoetis TRIMcommon · top position (5-8% full position) | S&P | IAPTC | 75.84closeAug 13 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +11.9% |
| WhySold down somewhat evenly across every long to raise a ~12% cash position; Zoetis remains one of the four top holdings separated by only basis points. |
| SPLTR:NASDAQ saas Palantir HOLDputs | NASDAQS&PBitget | ISPTC | 179.01closeAug 13 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +10.9% |
| WhySpared the Palantir puts alongside Nvidia while cutting elsewhere in the short book. |
| SQQQ:NASDAQ indexus Invesco QQQ Trust ADDputs — rolled January 2027 strikes in the mid-high $500s to June 2027 strikes in the mid-high $600s · about 6% of the portfolio | NASDAQBitget | SC | 732.07closeAug 13 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +9.5% |
| WhyExchanged the SOXX puts for a much larger QQQ put position, rolling from Jan 2027 mid-high $500 strikes to June 2027 mid-high $600 strikes at a substantial debit. The trade moved him from 45 IV in the SOX to 25 IV in the QQQs on the thesis that index volatility's recent compression will spread on further rallies. |
| LPYPL:NASDAQ financefintech PayPal TRIMcommon · second tier, ~1pp below the top group | NASDAQS&PBitget | ISPT | 60.59closeAug 13 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | -3.1% |
| WhyReduced evenly with the other longs during the de-grossing while keeping the position intact. |
| LJD:NASDAQ airobot JD.com HOLDcommon · top position, roughly tied with ZTS/MELI/ADBE | NASDAQBitget | IS | 29.3closeAug 13 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | -6.5% |
| WhyJD fell almost 10% on its first revenue decline (-2.9%) since its 2014 IPO, but he sees nothing wrong with the earnings — the comparison is against last year's subsidy bonanza. He believes China has already hit the nadir of its Great Depression and consumer names like JD will catch a tailwind while remaining remarkably cheap and dominant. |
| LADBE:NASDAQ saas Adobe TRIMcommon · top position (5-8% full position) | NASDAQS&PBitget | IPT | 270.49closeAug 13 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | -8.2% |
| WhyTrimmed with the rest of the long book to free up cash while maintaining short bias; remains a top-four holding. |
| LLULU:NASDAQ consumershopping lululemon athletica TRIMcommon · second tier, ~1pp below the top group | NASDAQS&P | ISAPT | 119.56closeAug 13 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -11.8% |
| WhyPart of the even sell-down across longs; sits in the second size tier roughly one percentage point below the top four. |
| LFLUT Flutter Entertainment TRIMcommon · second tier, ~1pp below the top group | — | AT | 98.32closeAug 13 '26 | 79.15 ● | 72.56229.86 | -26.0%APY -56.6% | -19.5% |
| WhyReduced evenly with all longs to build the cash buffer ahead of an expected larger market fall. |
| LMELI:NASDAQ airobot MercadoLibre TRIMcommon · top position (5-8% full position) | NASDAQBitget | ISAPC | 1828.25closeAug 13 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -20.0% |
| WhyTrimmed as part of the even reduction across all longs; still one of the four largest positions. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market TRIMcommon · about 5% | NASDAQ | IAPT | 82.68closeAug 13 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -38.9% |
| WhyCut modestly with the rest of the book; he still considers the ~5% weight a full position. |
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF EXITputs · entire position | NASDAQBitget | I | 550.74closeAug 13 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +74.4%closed |
| WhySold the losing SOXX put position entirely. Implied vol stayed stubbornly high which had protected the value, but on a further rally he expects to be doubly burned by both falling implied vol and adverse price action in the underlying. |
| SAMAT:NASDAQ aisemi Applied Materials COVERcommon · entire position | NASDAQS&PBitget | ISPTC | 534.54closeAug 13 '26 | 349.0 ● | 215.50739.67 | +52.2%APY +198.7% | +34.7%closed |
| WhyCovered the Applied Materials short on the same logic as Tesla — taking a quick sizable gain rather than pressing it in a rallying market. |
| STSLA:NASDAQ airobot Tesla COVERcommon · entire position | NASDAQS&PBitget | ISPTC | 339.96closeAug 13 '26 | 379.8 ● | 297.38498.83 | +8.8%APY +24.4% | -11.7%closed |
| WhyCovered the Tesla short after a decent gain — quick, sizable short-sale gains are "gift horses in this market." |
| LFNMA Fannie Mae TRIMcommon · about 5% | NASDAQ | — | —Aug 13 '26 | — | — | — | n/a |
| WhyTrimmed slightly in the even long reduction; the GSE position remains a full ~5% weight. |
| LFMCC Freddie Mac TRIMcommon · about 5% | NASDAQ | — | —Aug 13 '26 | — | — | — | n/a |
| WhyReduced alongside Fannie Mae in the across-the-board sell-down, still held as a full position. |
Full summary 706 words
Burry opens by pointing readers to Bloomberg's latest visualization of circular financing in the AI trade, arguing the idea that AI companies fund each other to buy from each other is now well understood — the Bank for International Settlements said as much in its June annual report. His tell that it is a real problem is that Wall Street itself never discusses it. The mechanism, as he describes it, is a concerted effort to inject capital into the AI ecosystem that comes straight back to those same companies as revenue, after which the market judges the participants on their amazing revenue growth. What changes the calculus is that the capital is increasingly debt: that puts the bubble on a clock with a concrete cost of capital. He declines to develop the argument fully in a Trading Post but states his base case plainly — 2028 is when he thinks compute becomes too much, while acknowledging history says markets catch on well before that sort of turn.
On JD.com, he addresses the day's near-10% decline, which he says fits the pattern of the last few years: the stock builds a head of steam into earnings and then crashes on the print. He sees nothing wrong with the numbers. The headline is JD's first revenue fall (-2.9%) since going public in 2014, but the real story is that the year-ago comparison contained a subsidies bonanza that juiced sales. His view remains that China is going through its Great Depression, that it has already hit the nadir, and that consumer names like JD will catch a tailwind — while the company stays remarkably cheap and dominant in its space.
The body of the post is a de-grossing. He describes a day that happens on occasion in the life of someone who sells short: a short book that had been roughly breakeven is tipping into a loss as the market and certain stocks rally. His approach is to sell volatility where it still exists, buy cheap volatility where he can, and manage the options both to realize losses and to balance exposures against outright shorts. Concretely, he sold his losing iShares Semiconductor ETF (SOXX) put position entirely — implied vol had stayed stubbornly high and protected the value, but on a further rally he expects to be doubly burned by lower IV and adverse underlying price action. He exchanged it for a much larger Invesco QQQ position, rolling January 2027 puts struck in the mid-high $500s into June 2027 puts struck in the mid-high $600s, moving from 45 IV in the SOX to 25 IV in the QQQs on the thesis that index volatility's recent compression will spread as the rally continues. The roll cost a substantial debit and restored his value at risk roughly to where the SOXX puts had it; the QQQ put position is now about 6% of the portfolio. He spared the Nvidia and Palantir puts.
Across the rest of the short book he pulled back selectively, with one move against the trend: he increased the Micron short on its move back up near $1,000. The SOXX short remains the largest at about 7% of the portfolio, followed by Micron, then Nebius, Nvidia, Oracle and Palantir. He covered Tesla after a decent gain and covered Applied Materials as well, calling quick sizable short-sale gains "gift horses in this market," and trimmed 25% of the Caterpillar short. Oracle's puts remain very expensive, so he continues to express that one through the shares.
He then turned to the longs, selling down somewhat evenly across every position — reducing each by relatively little but overall raising about a 12% cash position. The resulting top holdings, all about the same size and separated by basis points, are Zoetis, MercadoLibre, JD.com and Adobe. Roughly a percentage point lower sit lululemon, Veeva, Flutter, Molina Healthcare, HCA and PayPal. At about 5% each are Sprouts, Fannie Mae and Freddie Mac; he considers all of them full positions, ranging from 5% to 8%. His stated goal was to reduce gross exposure and free up cash while maintaining the short bias — and he says explicitly that if this looks like preparation for a larger fall in the market, that is because it is.
Key metrics
- JD revenue -2.9%, the first decline since its 2014 IPO; stock down almost 10% on the day
- QQQ put roll: January 2027 strikes in the mid-high $500s rolled to June 2027 strikes in the mid-high $600s, at a substantial debit
- Volatility trade moved from 45 IV in the SOX to 25 IV in the QQQs
- QQQ put position now about 6% of the portfolio
- SOXX short is the largest short at about 7% of the portfolio
- Micron short increased on its move back up near $1,000
- Caterpillar short trimmed by 25%
- Cash position raised to roughly 12% via an even sell-down across all longs
- Top four longs (ZTS, MELI, JD, ADBE) all about the same size, separated by basis points; full positions range 5% to 8%
- Base case for the AI bubble turning: 2028, when he thinks compute becomes too much
Key insights
- Burry puts a date on it: his base case for the AI bubble's turn is 2028, when compute becomes too much — but he notes markets historically catch on well before the actual turn.
- The reason debt matters more than the circularity itself is that it gives the bubble a concrete cost of capital, converting a sentiment story into a clock.
- The core trade is a volatility rotation, not a directional change: out of 45 IV SOXX puts and into 25 IV QQQ puts, on the view that compressed index vol spreads once rallies force it wider.
- He extended duration as well as strike — January 2027 mid-high $500 strikes rolled to June 2027 mid-high $600s at a substantial debit, buying time and moneyness rather than reducing the bet.
- The short book is now ranked SOXX (~7%, largest) > Micron > Nebius > Nvidia > Oracle > Palantir, with the QQQ puts a separate ~6% sleeve.
- Micron was the one counter-trend add, increased on its move back near $1,000 while nearly everything else was cut.
- Oracle is shorted in shares rather than options purely because its puts remain too expensive — a mechanical, not thematic, distinction.
- He treats JD's first-ever revenue decline (-2.9%) as an artifact of lapping a subsidy bonanza, not deterioration, and holds it as a top-four long on the view that China's Great Depression has already hit its nadir.
- The ~12% cash position was raised by shaving every long a little rather than exiting any — preserving the full long roster while explicitly preparing for a larger fall.
#027
Trading Post
Aug 12, 2026
Trading Post August 12, 2026: More Buys, More Shorts
Nebius is inadvertently publishing the evidence against its own asset base: wholesale compute is sold in power terms because power is scarce, yet short-dated compute clears at roughly twice the price of multi-year compute. That is backwardation, and backwardation exists because the underlying commodity is depreciating. Power does not depreciate, so either the GPU is depreciating, the customer is depreciating, or — Burry's view — both. Management's decision to deliberately withhold 2027 capacity to keep selling short-term contracts confirms it knows the value sits at the front of the curve, even as it lengthens server depreciation from four years to five and books the difference. Burry therefore presses the AI-complex short book (ORCL, MU, NBIS, SOXX puts) into strength while adding to the two non-AI longs, MELI and ZTS.
aimemoryaineocloudairobotaisemihealth
Positions & performance
+28.1%avg return83% win · 6 tracked · L -2.8% · S +43.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDMarch 2027 puts struck in the low $400s | NASDAQBitget | I | 546.61closeAug 12 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +74.2% |
| WhyAdded to March 2027 SOXX puts struck in the low $400s — an index-level, defined-risk expression of the same view, dated far enough out to survive the current momentum and to capture a semi cycle rollover if compute demand proves front-loaded. |
| SNBIS:NASDAQ aineocloud Nebius ADDcommon | NASDAQBitget | ISPT | 247.0his fillAug 12 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +56.0% |
| WhySold short more Nebius at $247, calling it 'what the top of a boom looks like'. His specific objection is that Nebius is pricing 6-month compute at $40-50M per megawatt versus $20-25M for 1-3 year contracts — backwardation that implies the GPU, the customer, or both are depreciating fast, even as Nebius extends server depreciation from four years to five. |
| SMU:NASDAQ aimemory Micron ADDcommon | NASDAQS&PBitget | ISAPTC | 924.0his fillAug 12 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +22.1% |
| WhyAdded to the Micron short at $924, again on AI-trade strength. Memory is the most cyclical link in the AI capex chain and is priced as if the front-loaded demand pull persists. |
| SORCL:NYSE aineocloud Oracle ADDcommon | S&PBitget | ISAPTC | 152.0his fillAug 12 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +21.7% |
| WhySold short more Oracle at $152 into strength, explicitly on a day the AI trade was propelled by CoreWeave's banner revenue. Oracle is his vehicle for the leveraged, contract-backed neocloud buildout he thinks is being financed against depreciating assets. |
| LZTS:NYSE health Zoetis ADDcommon | S&P | IAPTC | 73.6his fillAug 12 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +15.3% |
| WhyHe added to the Zoetis long at $73.60 with the same framing — no further elucidation needed, very bullish. Animal health is the defensive, cash-generative side of the book funded against his growing AI-complex shorts. |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon | NASDAQBitget | ISAPC | 1850.0his fillAug 12 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -21.0% |
| WhyBurry added to the existing MercadoLibre long at $1,850, saying it needs no further elucidation and that he is very bullish on the company. It is one of the two non-AI longs he is scaling into on a day when the AI trade is running. |
Full summary 482 words
This is a short Trading Post published into a strong AI tape, which Burry attributes mostly to CoreWeave's banner revenue. He does two things on the long side and four on the short side. He added to MercadoLibre at $1,850 and Zoetis at $73.60, saying both need no further elucidation and that he is very bullish on each — the same posture he has held on his non-AI compounders in prior posts. Everything else in the note is an escalation of the short book against the AI capex complex: more Oracle at $152, more Micron at $924, more Nebius at $247, and an addition to iShares Semiconductor ETF March 2027 puts struck in the low $400s.
The analytical content is entirely about Nebius, which he calls 'what the top of a boom looks like'. Rather than dismissing management, he quotes them at length and takes their disclosure seriously. Nebius says it closed four landmark deals in the quarter — with Reflection, Cohere, a scaled US new lab and a large US-based quant trading firm — averaging more than $1 billion each, at a yield of $20-25 million per megawatt on 1-3 year terms, with upfront payments covering 50-60% of the associated capex. It also says it could sell its entire 2027 capacity on those terms today but is choosing not to, because a second class of shorter-duration deals — up to six months, for customers with immediate time-bounded needs — is being negotiated at $40-50 million per megawatt and sometimes above.
Burry's inference is the heart of the post. Wholesale compute is priced in power terms because power is the scarce asset, and a longer contract should be cheaper per unit, not more expensive. Buyers paying twice the rate for six months rather than three years are placing enormous value on near-term use and very little on longer-term use. In credit and commodity terms, that is backwardation, and backwardation exists because the commodity is depreciating. Power does not depreciate — so either the GPU is depreciating, the customer is depreciating, or both. He is blunt that 'customers that would just not quit depreciating is what extended the bust 25 years ago', drawing the direct line to the telecom and dot-com unwind.
The accounting kicker is that Nebius extended its own server depreciation schedule this year from four years to five. In Burry's framing the company is simultaneously pricing its contracts as though roughly 50% annual decay is real and lengthening the period over which it charges that decay through the income statement. That gap between what the pricing implies and what the books assume is the short. He does not give a size or target for any of the positions, and he does not state a stop or invalidation — the implicit invalidation is straightforward, namely that the term structure normalizes and long-dated compute stops trading at a discount to short-dated compute.
Key metrics
- Added to MELI long at $1,850
- Added to ZTS long at $73.60
- Sold short more ORCL at $152 and more MU at $924
- Sold short more NBIS at $247
- Nebius closed 4 landmark deals in the quarter — Reflection, Cohere, a scaled US lab and a large US quant trading firm — averaging more than $1 billion each
- Nebius midterm (1-3 year) contracts yield $20-25 million per megawatt, with upfront payments covering 50-60% of associated capex
- Nebius short-duration deals (up to 6 months) are being negotiated at $40-50 million per megawatt and sometimes above — roughly 2x the midterm rate
- Nebius extended its server depreciation schedule this year from four years to five
- Added to SOXX March 2027 puts struck in the low $400s
Key insights
- Nebius's own disclosure is the short thesis: $40-50M per megawatt for up-to-6-month capacity versus $20-25M per megawatt for 1-3 year contracts is roughly 2x, an inverted term structure in a market where the scarce input — power — does not depreciate.
- Backwardation in compute implies the depreciating asset is the GPU, the customer, or both; Burry explicitly invokes the dot-com precedent that 'customers that would just not quit depreciating is what extended the bust 25 years ago'.
- Management deliberately declining to sell its entire 2027 capacity, despite being able to, is treated as an admission that Nebius itself believes value sits at the front of the curve.
- The accounting contradiction: Nebius's pricing implies roughly 50% annual decay while the company extended server depreciation from four to five years this year.
- The 50-60% upfront capex coverage on the billion-dollar midterm deals is not read as a positive — it is consistent with customers paying to secure near-term delivery rather than committing to long-duration value.
- The short expression is broadening from single names to the index: ORCL at $152 and MU at $924 are single-stock adds, but the SOXX March 2027 low-$400s puts are a defined-risk, longer-dated bet on the whole semi complex.
- The long book is being funded in the opposite direction — MELI at $1,850 and ZTS at $73.60 are non-AI compounders added on the same day the AI trade rallied on CoreWeave's results.
#026
Trading Post
Aug 11, 2026
Trading Post August 11, 2026
Burry is pressing the same trade he has been building since November 2025: the AI buildout is now being sustained by credit engineering rather than cash flow, and that is the late-cycle tell. The $500 billion bank-financed vehicle for Nvidia chip purchases matters less for whether it is technically circular than for what it rhymes with — Goldman clearing its $3 billion synthetic CDO in late 2005, which opened the floodgates to the 2006 vintage that destroyed Merrill, and Enron's push to make wholesale power an investable asset class. His deeper accounting work introduces the 'perimeter of debt': hyperscalers and their suppliers pushing obligations off balance sheet and locking in non-cancellable long-term supply contracts, which amplifies the bullwhip on the way down. Layered on top, BTIG's Krinsky notes 182 consecutive sessions without an 80%+ downside volume day — the longest such streak in 30 years — meaning dispersion has masked risk and a correlation-1 selloff has been deferred, not avoided. So he adds to his three largest shorts (expressed through SOXX, NVDA and PLTR puts, PLTR/CAT/ORCL common) while topping up Molina on the long side.
aineocloudaisemienergygridhealthsaas
Positions & performance
+26.8%avg return100% win · 7 tracked · L +17.7% · S +28.3%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDcommon | NASDAQBitget | I | 533.0his fillAug 11 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +73.5% |
| WhyPart of adding to his three largest shorts on the day — a broad expression of the view that the AI capex cycle is being prolonged by increasingly unnatural financing rather than by durable end demand. |
| SCAT:NYSE energygrid Caterpillar ADDcommon | S&PBitget | IAPT | 844.0his fillAug 11 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +32.4% |
| WhyHis stated reason is blunt: 'Data center plans are already aging.' Caterpillar is the pick-and-shovel proxy for the buildout, so a rolling over of data center plans hits the order book with a lag the stock has not discounted. |
| SNVDA:NASDAQ aisemi NVIDIA ADDDecember 2026 and June 2027 puts, strikes in the low $100s | NASDAQS&PBitget | ISPTC | 217.5closeAug 11 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +28.5% |
| WhyHe is less concerned with whether the $500 billion deal with Wall Street banks to finance customer purchases of Nvidia chips is literally circular financing than with what it signals: structuring unnatural credit to prolong momentum late in a bull phase. He likens it to Goldman Sachs offloading a $3 billion synthetic RMBS CDO in late 2005, which opened the floodgates to 2006, and to Enron's attempt to make wholesale power an investable asset class. |
| SORCL:NYSE aineocloud Oracle ADDcommon | S&PBitget | ISAPTC | 145.0his fillAug 11 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +17.9% |
| WhyWork on his next article has produced what he calls the 'perimeter of debt' — a wider interpretation of liabilities. He argues Oracle and peers are expanding that perimeter with a plethora of off-balance-sheet tricks and by reaching deep into supply chains for non-cancellable long-term contracts, whipping the bullwhip as hard as they can, which 'will hurt on the way back.' |
| LMOH:NYSE health Molina Healthcare ADDcommon · full position, back on par with his largest positions | — | ITC | 198.0his fillAug 11 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +17.7% |
| WhyThe position had shrunk in relative size due to his own capital movements rather than any change in view. He believes in Molina long term and thinks the political season may shape up favorably for the company, which would be an upside surprise; he points readers back to his late-December write-up 'Molina Healthcare: Ghosts of GEICO Past'. |
| SPLTR:NASDAQ saas Palantir ADDcommon | NASDAQS&PBitget | ISPTC | 175.0his fillAug 11 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +8.9% |
| WhyShorted more stock at $175 alongside the put additions, so the position is expressed in both common and options rather than options alone. |
| SPLTR:NASDAQ saas Palantir ADDMarch 2027 puts and December 2026 puts, strikes in the low $100s | NASDAQS&PBitget | ISPTC | 174.94closeAug 11 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +8.8% |
| WhyAdded to both expiries of low-$100s strike puts as part of adding to his three largest shorts, extending the bearish AI-complex positioning across multiple 2026-2027 expiries. |
Full summary 683 words
This is a short, dated trade log rather than an essay, and it opens with the line that frames it: 'Today I added to my three largest shorts. But first I bought stock.' The long is Molina Healthcare, added at $198. Burry is explicit that this is not a new idea or a changed view — the position had simply fallen off in relative size because of his own capital movements, and the add brings it back on par with his largest positions. He reiterates long-term belief in the company and adds a light catalyst: the political season may be shaping up as a good one for Molina, which he frames as a potential upside surprise rather than the basis of the thesis. He points readers back to his late-December 2025 write-up, 'Molina Healthcare: Ghosts of GEICO Past.'
The short book is where the work is. He shorted more of the iShares Semiconductor ETF (SOXX) at $533, added to NVIDIA December 2026 and June 2027 puts at strikes in the low $100s, added to Palantir March 2027 and December 2026 puts also at low-$100s strikes, and shorted more Palantir stock at $175. He shorted more Caterpillar at $844 with a one-line rationale — 'Data center plans are already aging' — and shorted more Oracle at $145.
The Nvidia section carries the argument. Burry sets aside the debate over whether the $500 billion deal with Wall Street banks to raise capital for chip purchases is genuinely circular financing, because in his view that is not the issue. The issue is the pattern. He recounts late 2005, when Goldman Sachs was choking on a $3 billion synthetic RMBS CDS CDO it had been warehousing, looking like the house of cards might crash only months after he had put on his big short — until Goldman managed to sell it a week later and the floodgates opened to 2006, the year Merrill Lynch moved onto its balance sheet the risk that ultimately doomed it, as did many others. He also invokes Enron's effort to make wholesale power an investable asset class. Structuring credit, he allows, is a natural part of the system; structuring unnatural credits to prolong momentum late in the bull phase is where the worry begins. His dry summary of the prevailing mood: 'The proposal is to fail to embrace more debt here is un-American. Well, ok then. We will see.'
The Oracle add previews his next article and introduces the concept he calls the 'perimeter of debt' — a deliberately wider interpretation of liabilities. He argues these companies are expanding that perimeter through off-balance-sheet techniques and by reaching deep into their supply chains for non-cancellable long-term contracts, whipping the bullwhip as hard as they possibly can. That, he says, 'will hurt on the way back' — the same mechanic that makes Caterpillar vulnerable once data center plans stop being extended.
He closes with a technical overlay from BTIG's Jonathan Krinsky, whom he calls the most fundamentals-based technical analyst he has come across. Krinsky notes that dispersion has been extreme — high single-stock volatility against low index volatility — so there has been no correlation-1 selloff in a long time: by NYSE all-securities volume, this is the 182nd consecutive session without an 80%+ downside volume day, the longest streak in at least 30 years by nearly 50 trading days, and finishing 2026 without one would be unprecedented in a record where no year has had fewer than five. Burry concedes such a technical factor is easy to dismiss on its own, but says he has been writing about fundamental reasons for exactly this since November 2025, pointing back to 'The Blessed Fraud Recurrence' and the history of electrification. His takeaway is about duration rather than timing: in cycles this big, months or even years are nothing, and if something revolutionary is going to happen it will happen — over a long enough period for everyone to be right and almost everyone to go bankrupt. The practical rule he draws: avoid stepping into someone else's folly, and avoiding leverage is how you avoid the folly.
Key metrics
- Added to Molina Healthcare at $198
- Shorted more SOXX at $533
- Shorted more Palantir stock at $175
- Shorted more Caterpillar at $844
- Shorted more Oracle at $145
- NVDA and PLTR puts added at strikes in the low $100s, expiries December 2026, March 2027 and June 2027
- $500 billion deal with Wall Street banks to raise capital for purchases of Nvidia chips
- Goldman Sachs' $3 billion synthetic RMBS CDS CDO warehoused in late 2005
- 182nd consecutive trading day without an 80%+ downside volume day on NYSE — the longest streak in at least 30 years by nearly 50 trading days (Jonathan Krinsky, BTIG)
- No year in at least 30 years has had fewer than five 80%+ downside volume days
- 'The Blessed Fraud Recurrence' had nearly 41,000 views and ~300 comments
Key insights
- The Molina add is a sizing correction, not a new thesis — the position had shrunk relative to the book because of his own capital movements, and $198 brings it back on par with his largest positions.
- He explicitly refuses the circular-financing debate on the $500B Nvidia chip-purchase facility: the question is not whether it is circular but that unnatural credit is being structured to prolong momentum late in the bull phase.
- The historical template is late 2005 — Goldman warehousing a $3B synthetic RMBS CDS CDO, nearly cracking, then clearing it and opening the floodgates to the 2006 vintage that put fatal risk on Merrill's balance sheet.
- 'Perimeter of debt' is the new analytical frame for his next article: off-balance-sheet structures plus non-cancellable long-term supply contracts reaching deep into the chain, which amplifies the bullwhip on the downswing. Oracle at $145 is the expression.
- Caterpillar at $844 is the physical-world short on the same view, justified in five words: 'Data center plans are already aging.'
- Palantir is now expressed in both instruments — stock shorted at $175 plus added December 2026 and March 2027 puts at low-$100s strikes — implying he wants convexity and carry-free downside across two expiries.
- The put strikes across NVDA and PLTR cluster in the low $100s with expiries from December 2026 to June 2027, so the positions need a large drawdown within roughly 4-10 months to pay, not merely a de-rating.
- Krinsky's statistic is the market-structure argument: 182 consecutive sessions without an 80%+ NYSE downside volume day is the longest streak in 30+ years by nearly 50 days, and no year on record has had fewer than five such days — dispersion has deferred a correlation-1 selloff, not cancelled it.
- His stated survival rule for a cycle he expects to last years: 'Avoid the leverage, and one is more likely to avoid the folly' — everyone can be right and almost everyone can still go bankrupt.
#025
Trading Post
Aug 10, 2026
Trading Post August 10, 2026
With Palantir put implied vol crushed to yearly lows while the stock sits at 69x sales, Burry re-establishes his out-of-the-money put position on top of the remaining half of his common short, arguing PLTR's true economics (employee share delivery ~6x reported SBC, tripling off-balance-sheet infrastructure commitments) make the shares worth under $1 long-run. He funds patience on the long side with washed-out quality — adding Zoetis in the low $70s, now cheaper than leveraged competitor Elanco despite far better margins and pipeline. Market-wide, he sees a 1987-like setup at much higher valuations, keeps QQQ and SOXX puts for a step-function break, and expects that when 'price matters again' fallen value names will be the ones that rise.
aisemienergygridhealthindexussaas
Positions & performance
+26.0%avg return100% win · 4 tracked · L +13.4% · S +30.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF REITputs | NASDAQBitget | I | 529.39closeAug 10 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +73.4% |
| WhySOXX puts reiterated alongside QQQ puts as the index-level expression of the short book against a possible 1987-style step function down. |
| LZTS:NYSE health Zoetis ADDcommon | S&P | IAPTC | 74.83closeAug 10 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +13.4% |
| WhyAdded in the low $70s. The stock has traded 8.3x shares outstanding since its 2021 peak and 2.8x float just since late 2025 — weak hands are washed out — while ZTS trades at ~10x EBITDA/~12x earnings, a discount to leveraged, lower-margin Elanco; he expects the US business to stabilize over the next couple of years. |
| SPLTR:NASDAQ saas Palantir ADDMarch 2027 puts, strikes in the low-to-mid $100s | NASDAQS&PBitget | ISPTC | 175.23closeAug 10 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +9.0% |
| WhyRe-entering the out-of-the-money put trade (he had no puts for a while, and still holds half his common short since covering half near $107). Implied vol came in hard and puts in that strike range fell to levels not seen this year, while PLTR is back to 69x sales; his updated work after earnings finds shares worth under $1 long-run. |
| SQQQ:NASDAQ indexus Invesco QQQ (Nasdaq-100 ETF) REITputs | NASDAQBitget | SC | 720.87closeAug 10 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +8.1% |
| WhyPositioned for step-function developments akin to 1987 but at much higher overall valuations; QQQ puts are part of the significant short book while off-balance-sheet liabilities 'put both 2000 and 2008 to shame'. |
Full summary 575 words
This Trading Post records a small number of trades built around one decision: it is 'time to re-establish' the Palantir put position. Burry has held half his PLTR common short since covering half near $107, but has carried no puts for a while. With implied volatility having come in hard — puts in his target range falling to levels not seen this year — he bought March 2027 puts struck in the low and mid $100s while the stock sits back at 69x sales. He points readers to his two principal Palantir pieces (Palantir's New Clothes, Feb 12; Palantir: An Accounting, Feb 19) and previews new work: after the latest earnings he finds PLTR reported ~$3.0 billion of GAAP net income over the past year while delivering 31.3 million shares to employees worth about $5 billion — roughly six times the GAAP stock-based-compensation expense, the widest such gap among the 66 companies he has re-worked post-earnings. He also flags that Palantir's off-balance-sheet, non-cancellable infrastructure purchase commitments have more than tripled this year, and concludes that considering commitments, liabilities and long-term competitive position, the shares are worth under $1 in the long run — with the nuance promised in Part IV of the Heretic's Guide and Part 3 of the Software & Securities Stocks series. He is also attempting, but has not yet executed, a first-time purchase of out-of-the-money Caterpillar puts, held back by wide spreads and his insistence on paying low implied vol.
On the long side he added to Zoetis in the low $70s, reprinting last Trading Post's case: since the 2021 peak near $250 the stock has traded 8.3x its shares outstanding and over 9x its float (5.1x float since late 2024, 2.8x since late 2025), meaning the weak hands have been washed out. Zoetis now trades at a discount to Elanco (~10x EBITDA and ~12x earnings vs ~15x and ~20x) despite far better free-cash-flow conversion, a cheaper balance sheet (4% of EBITDA to interest vs Elanco's 22%), 80%+ US gross margin and 66% operating margin against Elanco's projected 55-56% gross margin, and a ~2:1 research-dollar advantage. The 10-point YoY loss of US in-clinic dermatology share — only a third of which Elanco captured — is real but not terminal decline, and if a price war comes, Zoetis wins. He quibbles with the $124 buyback funded by the $2 billion 0.25% December 2025 convertible ($148.20 conversion, hedged with calls capped at $211.72 costing $187 million) but likes the buying at $90. Management would not commit to a 6-, 12- or 18-month US stabilization timeline; Wall Street's impatience is the opportunity.
The market section restates his structure work (Foundations, Mar 3 and Jul 31): plenty of cheap stocks, plenty of very expensive ones, and risks piling up in a setup he likens to 1987 — at much higher valuations. He teases a coming piece on the 'expanded liability perimeter': off-balance-sheet liabilities today put both 2000 and 2008 to shame, much of it driven by component inflation. He makes no imminent predictions but is positioned for step-function developments via his significant short book including QQQ and SOXX puts, while advising most investors to sell expensive, buy cheap, hold cash or short-term Treasury money-market funds. Coming out of a 2000-style episode, he argues, the fallen value stocks doubled and tripled 'entirely because price mattered again' — today price does not matter, and he believes he can already see the field from which many stocks will rise.
Key metrics
- PLTR back up to 69x sales (range cited: 50-100x)
- PLTR GAAP net income ~$3.0B over the last year vs 31.3M shares delivered to employees worth ~$5B — about 6x the GAAP stock-based-comp expense, the widest gap of the 66 companies he has worked up post-earnings
- PLTR off-balance-sheet non-cancellable infrastructure purchase commitments more than tripled this year
- Burry's long-run value estimate for PLTR: under $1 per share
- ZTS has traded 8.3x its shares outstanding (9x+ float) since the 2021 peak near $250; 5.1x float since late 2024; 2.8x float since late 2025
- Elanco: 22% of EBITDA goes to interest expense vs 4% at Zoetis; ELAN ~15x EBITDA / ~20x earnings vs ZTS ~10x / ~12x
- ZTS US segment: 80%+ gross margin and 66% operating margin vs Elanco's projected 55-56% gross margin
- ZTS lost 10 points of US in-clinic dermatology share YoY; Elanco captured maybe 1/3 of the lost dollars
- ZTS $2B December 2025 convertible at 0.25% through June 2029, $148.20 conversion; dilution-hedge calls struck $148.20, capped $211.72, cost $187M
- Zoetis retains a ~2:1 research-dollar advantage over Elanco
Key insights
- Vol timing drove the PLTR re-entry: he waited until implied volatility collapsed and March 2027 low/mid-$100 strike puts hit their cheapest levels of the year before re-establishing, rather than chasing when the stock first re-rated to 69x sales.
- His post-earnings screen of 66 companies found PLTR has the widest gap between GAAP SBC expense and actual value of shares delivered to employees: 31.3M shares worth ~$5B against ~$3.0B GAAP net income — real dilution running ~6x the reported expense.
- Palantir's off-balance-sheet non-cancellable infrastructure purchase commitments more than tripled this year — part of a broader 'expanded liability perimeter' he says makes today's off-BS liabilities worse than 2000 and 2008.
- Share-turnover as a bottom signal for ZTS: 8.3x shares outstanding traded since the 2021 peak and 2.8x float just since late 2025 means the shareholder base's weak hands are exhausted — his stated rationale for sizing up in the low $70s.
- The ZTS/ELAN pair has inverted: the historically favored Zoetis now trades at ~10x EBITDA vs Elanco's ~15x, despite Elanco paying 22% of EBITDA in interest vs Zoetis's 4% — and in a price war Zoetis's 80%+/66% US margins beat Elanco's 55-56% gross margin.
- A new short target is forming: he is trying to buy his first-ever out-of-the-money Caterpillar puts but refuses to cross wide spreads or pay up for implied vol — watch for execution in a future post.
- He sees 1987 with today's valuations: no imminent prediction, but QQQ and SOXX puts held for a step-function break, with a recovery he expects would be drawn out like post-2000 given record index valuations.
- The regime call: 'Today, price does not matter' — echoing late-1998-to-2000 — and when price matters again he expects the out-of-favor cheap stocks (the field 'from which many stocks will rise') to double or triple as in 2001-2005.
#024
Trading Post
Aug 07, 2026
Trading Post August 7, 2026
Burry uses post-earnings and post-downgrade selloffs as his entry mechanism, adding to five longs where the narrative has deteriorated faster than the economics. The clearest expression is Zoetis: the market is treating lost U.S. dermatology share as terminal decline, but Zoetis lost 10 points of share to a fragmented field and category contraction rather than to Elanco alone, and it enters any price war with 80%+ gross and 66% operating margins against Elanco's 55-56% gross margins and a balance sheet consuming 22% of EBITDA in interest. Management's refusal to timeline U.S. stabilization is what created the discount, since Wall Street will not wait 6 to 18 months. On the GSEs, he argues the absence of IPO news is itself the opportunity — Fannie and Freddie compound book value at double digits while the IPO slips, so delay depresses sentiment without destroying value. Against those longs he is pressing a semiconductor short, adding both outright SOXX and March 2027 low-$400s puts specifically because the index's sharp bounce off recent lows improved the entry.
airobotaisemiconsumershoppingfinancefintechhealth
Positions & performance
+24.7%avg return67% win · 6 tracked · L +0.1% · S +74.0%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDMarch 2027 puts struck in the low $400s | NASDAQBitget | I | 543.27closeAug 07 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +74.0% |
| WhyHe added to his existing March 2027 low-$400s puts, using the semiconductor index's hard bounce off recent lows as a cheaper entry for downside exposure. |
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDcommon (ETF short) | NASDAQBitget | I | 541.0his fillAug 07 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +73.9% |
| WhyHe shorted more of the ETF outright at $541, explicitly because the Semiconductor Index bounced hard off recent lows and that rally provided an attractive entry point for the short. |
| LFISV:NASDAQ financefintech Fiserv ADDcommon | NASDAQS&PBitget | ISAPT | 51.93his fillAug 07 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +20.4% |
| WhyHe calls August of a transition year the perfect time to accumulate, and notes Mizuho's same-morning view that the stock has bottomed, that a second guidance cut in 2026 isn't necessary, and that Clover strength makes the Merchant two-year stack achievable — adding that the long-term setup is better than the 2H setup Mizuho describes. |
| LZTS:NYSE health Zoetis ADDcommon · large position | S&P | IAPTC | 72.72his fillAug 07 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +16.7% |
| WhyZoetis is falling on post-earnings downgrades and now trades at ~10x EBITDA and ~12x earnings versus Elanco at ~15x and ~20x, despite better free cash flow conversion, a far cleaner balance sheet (4% of EBITDA to interest vs 22% at Elanco) and a ~2:1 research spending advantage; he reads the 10 points of lost U.S. in-clinic dermatology share as perception damage, not terminal decline. |
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon | NASDAQS&P | ISAPT | 127.45his fillAug 07 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -17.3% |
| WhyA further add to the long he has built out in prior posts on the name; no new reasoning given beyond continuing to accumulate at this level. |
| LMELI:NASDAQ airobot MercadoLibre ADDcommon | NASDAQBitget | ISAPC | 1812.39his fillAug 07 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -19.3% |
| WhyHe topped off the existing position the day after earnings, using the post-print reaction as the entry. |
| LFMCC Freddie Mac ADDcommon | NASDAQ | — | —Aug 07 '26 | — | — | — | n/a |
| WhyHe argues the lull in IPO news is the buying opportunity he flagged in December's 'Fannie & Freddie, Toxic Twins No More No More' — delay does not destroy value because both GSEs keep building capital and shrinking their ERCF shortfalls, growing book value at double-digit rates while sentiment pushes the shares down. |
Full summary 725 words
This is a straightforward trade-log post: five adds to existing longs, one increase to a short, and one add to puts. The framing is consistent with how Burry has been operating all year — let the market hand him a worse narrative on a name he already owns, then size up.
The largest block of reasoning goes to the GSEs and to Zoetis. On Freddie Mac, bought at $5.43, he explicitly points back to his December 2025 piece 'Fannie & Freddie, Toxic Twins No More No More' and to his July 10th follow-up 'Fannie, Freddie, Trump & The Housing Crisis.' The December argument was that privatization is neither automatic nor quick for common and junior preferred holders, and that if the companies do not go public they simply keep building capital reserves and shrinking their ERCF funding shortfalls — meaning there is little internal financial pressure to move the IPO up the President's agenda, and the entities get better capitalized the longer it is deferred. His conclusion then was that further delay does not destroy value while both grow book value at double-digit rates, even though sentiment would likely drive the securities down through an unproductive year. He now says that predicted lull has arrived and is the long-term investor's opportunity.
Zoetis, bought at $72.72 and described as a large position, gets the most detailed work-up, structured as a pair comparison against Elanco, a company he says he knows well from owning it when it was the out-of-favor half. The valuation inversion is the hook: Elanco now trades at roughly 15x EBITDA and 20x earnings while Zoetis — historically the favored name — trades at about 10x and 12x, despite better free cash flow conversion. The balance sheets are not comparable: 22% of Elanco's EBITDA goes to interest expense versus 4% at Zoetis. On competitive dynamics he concedes Elanco is winning a bit in the vet's office, but quantifies it against the narrative: Zoetis lost 10 points of U.S. in-clinic dermatology share year over year, and Elanco captured only about a third of those lost dollars, with the rest going to smaller competitors and category contraction — 'not a story of terminal decline.' In a price war, Zoetis's 80%+ U.S. gross margins and 66% operating margins beat Elanco's guided 55-56% gross margins, and Elanco is already guiding down with more leverage and less room. He also credits Zoetis with a ~2:1 research dollar advantage and a better historical batting average on returns and pipeline productivity. The criticism is capital allocation: buying stock at $124 with proceeds from the $2 billion December 2025 convertible (0.25% coupon through June 2029, $148.20 conversion) was probably too aggressive, though he expects it to look fine long term, and the $187 million of capped calls struck at $148.20 and capped at $211.72 sit in paid-in capital and will never hit the income statement. Repurchases at $90 earlier this year he has no quarrel with. The proximate cause of the discount, in his read, is management refusing to say whether U.S. stabilization takes 6, 12 or 18 months.
The three remaining longs get brief treatment. MercadoLibre at $1,812.39 was a top-off the day after earnings. lululemon at $127.45 is a continued accumulation. Fiserv at $51.93 comes with the observation that August of a transition year is the ideal time to accumulate, plus an unusual nod to sell-side confirmation: Mizuho called the bottom pre-market the day before and followed up arguing a second 2026 cut isn't necessary, with Clover strength making the ambitious Merchant two-year stack doable and easy Financial Solutions comps supporting an attractive 2H setup. Burry says he doesn't need the validation but was surprised enough to share it, and thinks the long-term setup is much better than the 2H one Mizuho describes.
On the short side he added to March 2027 SOXX puts struck in the low $400s and shorted more SOXX outright at $541, in both cases justified by the same technical point — the Semiconductor Index bounced hard off recent lows, and that bounce is what made the entry attractive. The implicit risk is symmetric: the semis short is a timing-and-level trade, so continued index strength works against both legs, and the long book depends on narratives (GSE privatization, Zoetis U.S. stabilization, Fiserv's transition year) resolving on a longer clock than the market is willing to grant.
Key metrics
- Added Freddie Mac (FMCC) at $5.43
- Added MercadoLibre (MELI) at $1,812.39, the day after earnings
- Added lululemon (LULU) at $127.45
- Added Fiserv (FISV) at $51.93
- Added Zoetis (ZTS) at $72.72; Zoetis is described as a large position
- Shorted more SOXX at $541 and added to March 2027 SOXX puts struck in the low $400s
- Elanco pays ~22% of EBITDA in interest expense vs ~4% at Zoetis
- Elanco ~15x EBITDA / ~20x earnings vs Zoetis ~10x EBITDA / ~12x earnings
- Zoetis lost 10 points of U.S. in-clinic dermatology share year over year; Elanco captured only about one third of those lost dollars
- Zoetis U.S. segment: 80%+ gross margin and 66% operating margin vs Elanco gross margin guided to 55-56%
- Zoetis $2B December 2025 convertible note at 0.25% through June 2029 with a $148.20 conversion price; capped calls struck $148.20 and capped at $211.72 cost $187M booked to paid-in capital
- Zoetis repurchased stock at $124 with convert proceeds and a lot more at $90 earlier this year
Key insights
- The Zoetis add is a valuation-inversion trade: the historically premium name now trades at ~10x EBITDA / ~12x earnings while Elanco, the former value half, trades at ~15x / ~20x, with Zoetis converting free cash flow better and carrying far less debt.
- He quantifies the Zoetis bear case to defuse it — 10 points of U.S. in-clinic dermatology share lost, but Elanco took only about one third of those dollars; the rest went to small competitors and category contraction.
- Any Zoetis-Elanco price war favors Zoetis structurally: 80%+ U.S. gross margins and 66% operating margins against Elanco's guided 55-56% gross margins, plus 4% vs 22% of EBITDA going to interest.
- The GSE thesis is that delay is not damage: with no IPO, Fannie and Freddie keep building capital and shrinking ERCF shortfalls while compounding book value at double digits, so a news lull depresses sentiment without impairing value — exactly the setup he flagged in December 2025.
- He is explicit that GSE privatization is 'neither an automatic nor a necessarily quick win' for common and junior preferred holders, which is why he treats weakness as an accumulation window rather than a catalyst trade.
- The Fiserv add leans on seasonality and a transition-year framing, with Mizuho's same-morning call that the bottom is in and that a second 2026 guidance cut isn't needed — Clover strength carrying the Merchant two-year stack.
- His capital-allocation critique of Zoetis is specific: buybacks at $124 funded by the $2B / 0.25% December 2025 convert (conversion $148.20) were too aggressive, and the $187M of capped calls struck $148.20 / capped $211.72 never touch the income statement.
- The semiconductor short was pressed on strength, not weakness — more SOXX shorted at $541 and more March 2027 low-$400s puts bought, with the hard bounce off recent lows cited as the reason the entry was attractive.
#023
Trading Post
Aug 06, 2026
Trading Post 8/6/2026: Two Shorts, 3 Earnings Reports with 1 Add & 2 Holds
Burry argues that a cluster of AI-infrastructure companies has quietly financed its build-out with off-balance-sheet liabilities — backstops, uncommenced leases and purchase commitments — that are excused today as evidence of future growth rather than counted as obligations. Because those obligations are enormous relative to the businesses carrying them (he cites uncommenced leases running as long as 11 years at 18x a company's entire sales), the apparent scale and momentum of these firms is borrowed rather than earned, and the same disclosure pattern repeats across the group. That makes them, in his phrase, fish in a barrel: large, similar-looking and short-able almost interchangeably, with the added expectation that sheer size will suffocate them. He acts on it by shorting Oracle at $144.63 and Nebius, in larger size, at $211.77 — common stock rather than puts, since implied volatility above 100% makes the options unpriceable. On the long side he does the opposite thing: he uses post-earnings noise in Fiserv, Zoetis and MercadoLibre as an opportunity, treating transitional quarters and credit-portfolio anxiety as reasons to add or hold quality franchises at better prices.
aineocloudairobotfinancefintechhealthsaas
Positions & performance
+5.2%avg return50% win · 8 tracked · L -4.1% · S +33.2%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SNBIS:NASDAQ aineocloud Nebius Group NEWcommon · somewhat larger size than the Oracle short | NASDAQBitget | ISPT | 211.77his fillAug 06 '26 | 108.6 ● | 73.52299.86 | +48.7%APY +308.1% | +48.7% |
| WhySame off-balance-sheet thesis, in larger size: he cites a company with uncommenced leases running as long as 11 years and totalling 18x its entire sales. He looked at puts but declined them — implied volatility above 100% made the option prices nonsensical. |
| LFISV:NASDAQ financefintech Fiserv ADDcommon | NASDAQS&PBitget | ISAPT | 52.0his fillAug 06 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +20.2% |
| WhyHe added on the post-earnings dip to $52 versus his $48 average. He treats the quarter as meaningless — the start of the new CEO's kitchen-sink period — while 84% of revenue still grows and Clover revenue is up 13%, leaving the franchise intact for when growth resumes off a cleaner base. |
| SORCL:NYSE aineocloud Oracle NEWcommon | S&PBitget | ISAPTC | 144.63his fillAug 06 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +17.7% |
| WhyBurry says his ongoing work on Part IV of the Heretic's Guide forced the trade — Oracle is one of the AI-infrastructure 'fish in a barrel' that has gorged on off-balance-sheet liabilities (backstops, uncommenced leases, purchase commitments) that are presented as future growth rather than obligations. |
| LZTS:NYSE health Zoetis HOLDcommon | S&P | IAPTC | 78.0his fillAug 06 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +8.8% |
| WhyStock up ~$4 to $78 against his $83 average; softness in North American pet health is economy-driven and already expected. He values the franchise's market position and high-ROIC R&D/marketing process, likes the new outsider CFO/COO hired from GE HealthCare, and notes the shares trade below IV15 against a non-cheery consensus. |
| LADBE:NASDAQ saas Adobe HOLDcommon · largest position | NASDAQS&PBitget | IPT | 260.24closeAug 06 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | -4.6% |
| WhyNamed as one of the two largest positions in the book at the moment, alongside JD.com. |
| LFLUT Flutter Entertainment ADDcommon · full position | — | AT | 92.91closeAug 05 '26 | 79.15 ● | 72.56229.86 | -26.0%APY -56.6% | -14.8% |
| WhyHe added more after Flutter's report the prior day, taking the holding to a full position. It now sits in the second-largest cohort of the book alongside ZTS and FISV. |
| LJD:NASDAQ airobot JD.com HOLDcommon · largest position | NASDAQBitget | IS | 32.81closeAug 06 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | -16.5% |
| WhyNamed as one of the two largest positions in the book at the moment, alongside Adobe. |
| LMELI:NASDAQ airobot MercadoLibre HOLDcommon | NASDAQBitget | ISAPC | 1782.0his fillAug 06 '26 | 1462.11 ● | 1,4542,342 | -18.0%APY -50.4% | -18.0% |
| WhyDown $140 to $1,782 but still above his $1,611 average; he is comfortable holding and would buy more at $1,500 or less, hoping for $1,300-1,400. Price/sales is plumbing lows while net revenue crossed $10bn (+50% y/y), and he reads the credit-portfolio worry as the reason the price is on offer. |
Full summary 638 words
This Trading Post is an interruption to Burry's larger project — he says he is in the middle of writing Part IV of the Heretic's Guide and describes it as a doozy — and the interruption happened because the research reached a point where he could not keep writing without putting on the trades. The framing is deliberately dismissive: shorting these names, he says, is like shooting fish in a barrel, because they all look similar and probably any single one would do. What he is having trouble getting his head around is not whether the shorts work but why so many companies came to share the same defect at the same time.
The defect is off-balance-sheet liability. He lists the forms it takes — backstops, uncommenced leases, purchase commitments — and argues the fish have gorged on them until they are very fat, very large, and easy to shoot. The specific 'inconvenient truth' he says he came across is uncommenced leases, some running as long as 11 years, amounting to 18x the company's entire sales. He is explicit that this is not an isolated case: the obligation is excused in the present as simply future growth, misinterpreted. His expectation is that scale itself becomes fatal — that it shan't be long before every last one keels over for lack of oxygen.
The positioning that follows is two new shorts, both put on that day. He shorted Oracle (ORCL) at $144.63, and Nebius (NBIS) in somewhat larger size at $211.77. He looked at puts on Nebius and passed: implied volatility is greater than 100% and the prices, in his words, do not make sense — so the expression is common stock.
The rest of the post handles three earnings reports on the long side, and the pattern is that he refuses to judge a company on a transitional quarter. Fiserv is down $2 to $52 against his $48 average; he calls the quarter not merely meaningless but the beginning of the kitchen-sink period for a brand-new CEO, notes that management emphasised underlying strength in the 84% of revenue still growing and Clover's 13% revenue growth, and expects more revenue damage from intentional acts as the CEO reviews costs and underperforming assets — so he added on the dip. Zoetis is up about $4 to $78 against his $83 average, with North American pet demand soft alongside the economy; he likes the new CFO/COO poached from GE HealthCare as an outsider bringing differentiated views, and frames the investment as buying market position and a high-ROIC research and marketing process below IV15 against a non-cheery consensus. MercadoLibre fell $140 to $1,782, still above his $1,611 average; net revenue passed $10 billion for the first time at +50% y/y with $683m operating income at a 6.7% margin, which CFO Martin de los Santos described as a deliberate choice to prioritise long-term engagement and scale over near-term profitability.
The MercadoLibre worry he addresses head-on is the credit portfolio, which grew 75% last quarter. He quotes Fintech President Osvaldo Gimenez explaining that cohorts older than three years are a small part of the book because issuance has accelerated, so the average portfolio simply needs more time to mature, and de los Santos adding that credit-card users are 2-3x more likely to become 'ecosystemic' — buying more on MercadoLibre and bringing more of their financial life into Mercado Pago. Burry reads that tension as the reason the price is available, with price/sales plumbing lows, and says he is very comfortable holding and will buy more at $1,500 or less. Flutter, which reported the day before, he had already added to, making it a full position. He closes with sizing: ADBE and JD are the largest positions, the next cohort down includes ZTS, FLUT and FISV, and MELI is a step below that for now.
Key metrics
- Shorted Oracle (ORCL) at $144.63
- Shorted Nebius (NBIS) at $211.77, in larger size than ORCL
- NBIS puts rejected: implied volatility greater than 100%
- Uncommenced leases as long as 11 years, totalling 18x the company's entire sales
- FISV $52 vs $48 average purchase price; 84% of revenue still growing; Clover revenue +13%
- ZTS $78 vs $83 average purchase price; trading below IV15
- MELI $1,782 vs $1,611 average purchase price; target buy zone $1,500 or less, hoped-for range $1,300-1,400
- MELI net revenue surpassed $10 billion for the first time, +50% y/y
- MELI income from operations $683 million, 6.7% margin
- MELI credit portfolio grew 75% last quarter; credit-card users 2-3x more likely to become 'ecosystemic'
- Position sizing: ADBE and JD largest; next cohort includes ZTS, FLUT, FISV; MELI a step below
Key insights
- Burry opened two fresh AI-infrastructure shorts on the same day — Oracle at $144.63 and Nebius at $211.77 — with Nebius in the larger size, marking the first time the off-balance-sheet thesis has been expressed as live short positions rather than research.
- The analytical hook is uncommenced leases: he flags a company with lease commitments running up to 11 years and totalling 18x its entire annual sales, obligations that sit outside the balance sheet and get read as growth rather than debt.
- He explicitly says this is not a single-name problem — the companies 'all look similar and probably any single one would do' — implying the short list is longer than the two names disclosed, with Part IV of the Heretic's Guide to follow.
- He rejected Nebius puts despite the bearish conviction: implied volatility above 100% made option prices nonsensical, so he took the equity short and the unlimited-loss profile that comes with it.
- Fiserv is a deliberate turnaround bet, not a growth bet: he adds at $52 versus a $48 average and treats the print as the new CEO's kitchen sink, leaning on the 84% of revenue still growing and Clover's +13% as evidence the franchise is intact.
- Zoetis is being held through a mark-to-market loss ($78 vs $83 average) on the argument that soft North American pet demand is cyclical and known, with the shares below IV15 and a new outsider CFO/COO from GE HealthCare as a positive change agent.
- MercadoLibre's selloff to $1,782 is, in his framing, investors mistaking deliberate reinvestment for deteriorating quality — $10bn revenue at +50% y/y and a 6.7% operating margin the company says is a chosen trade-off — with price/sales at lows and a stated add level of $1,500 or below.
- The book's shape is now disclosed: ADBE and JD are the largest positions, followed by a cohort including ZTS, FLUT and FISV, with MELI smaller — a long book of unloved quality franchises funded against short exposure to AI-infrastructure balance sheets.
#022
Trading Post
Aug 05, 2026
Trading Post August 5, 2026: One Buy, One Sale, One Reason
Prediction markets, led by Kalshi at roughly $39.7 billion of annualized volume, are cannibalizing the U.S. online sportsbooks by self-certifying event contracts with the CFTC and claiming a federal exemption from state gambling law — which lets them sidestep state gaming taxes as high as 51%. That regulatory arbitrage, not a structural loss of consumer demand, is what just collapsed Flutter's U.S. EBITDA by 70% and forced a guidance cut and higher leverage. Burry's view is that a tax loophole of this size will not stand: courts, states hunting the revenue, and eventual federal regulation will restore a level playing field, and the incumbents are already building their own prediction-market products rather than fighting. If that happens, the sell-off in the incumbents is a valuation dislocation rather than a terminal-decline signal. He therefore wants the cheapest, highest-quality asset in the space and only that one — Flutter below his intrinsic-value estimate, not DraftKings above it.
consumertravel
Positions & performance
-12.1%avg return0% win · 1 tracked · L -12.1%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFLUT Flutter Entertainment (FanDuel) ADDcommon · more than doubled the position; average cost pulled down into the $98s | — | AT | 90.0his fillAug 05 '26 | 79.15 ● | 72.56229.86 | -26.0%APY -56.6% | -12.1% |
| WhyThe stock fell about $15 to near $90 on a U.S. guidance cut, putting it below his IV15 of $96 for a business he regards as low Castle/high Chapel. He funded the add by selling DKNG, arguing the prediction-market tax loophole that is depressing U.S. sportsbook economics will not survive regulation or taxation. |
| LDKNG:NASDAQ consumertravel DraftKings EXITcommon · entire position | NASDAQBitget | IAT | 21.76closeAug 05 '26 | 18.29 ● | 16.5536.98 | -32.7%APY -66.5% | -15.9%closed |
| WhyHis IV15 for DraftKings is well below the current price, so it offers no margin of safety into an earnings report he expects to miss on the same U.S. weakness Flutter just reported. He sold it outright to concentrate the sector exposure in Flutter, which he calls the only investment he needs in the space. |
Full summary 538 words
This is a short, single-decision Trading Post built around Flutter's results and what they reveal about the U.S. online gambling market. Flutter reported that United States gambling EBITDA fell 70% to $119 million, offset by International EBITDA up 10% (4% organic) to $2.3 billion. Underneath that, U.S. revenue was down 6% and the sportsbook down 15%, while handle grew only 2% and promotional spend rose 140 basis points — the classic signature of an operator paying up to defend share. The company cut estimates on the U.S. weakness, which pushes full-year leverage from 3.7x to 4.3x, away from its 2.5x target, and Peter Jackson is stepping down as CEO after nine years, to be succeeded by Dan Taylor. The stock fell roughly $15 to near $90 and dragged DraftKings down with it ahead of DraftKings' own report the following afternoon, which Burry expects to miss for the same reason.
His diagnosis is that the cause is not consumer fatigue but a regulatory loophole. Prediction markets have, in his words, taken the gambling world by storm, with Kalshi now running $39.7 billion of annualized volume — 87% of it sports, even though the contracts can be written on nearly any real-world event. Kalshi, Polymarket and Crypto.com self-certify these contracts with the Commodity Futures Trading Commission and claim a federal exemption from gambling law, which is normally the states' domain. That exemption lets them escape state gaming taxes that can run as high as 51%, an enormous cost advantage over FanDuel and DraftKings. Notably, the incumbents have stopped fighting prediction markets directly — they are letting the states, which want those taxes, do the fighting — while both DraftKings and FanDuel move to leverage their existing customer bases and mindshare to launch prediction-market products of their own.
On the legal path, he cites prediction markets themselves: a 64% implied chance the Supreme Court agrees to hear a case on the legality, taxation and/or regulation of prediction markets by the end of this year, and a 36% chance it does not. He concedes the issue is neither black and white nor cut and dried, but his base case is that taxation and regulation arrive — enough, at minimum, to restore DraftKings and Flutter to their prior trajectory as online gambling matures. He notes the irony that in a society inserting gambling into every corner of life, two of the worst investments have been two of the largest online gambling operators.
Positioning follows directly from valuation rather than from the narrative. His IV15 for Flutter is $96, above the price the stock traded at after the drop; his IV15 for DraftKings is much lower than DraftKings' current price. So he sold his entire DKNG position and used the proceeds to more than double his Flutter stake, pulling his average cost down into the $98s — close enough to intrinsic value for what he describes as a low Castle/high Chapel business, which makes it a fat pitch and the only investment he needs in the space. He promises a fuller analysis later. The implicit invalidation is the legal one: if the CFTC self-certification exemption survives untaxed and unregulated, the incumbents' U.S. economics do not revert and the Flutter add is a value trap.
Key metrics
- Flutter U.S. gambling EBITDA down 70% to $119 million
- Flutter International EBITDA up 10% (4% organic) to $2.3 billion
- Flutter U.S. revenue down 6%; sportsbook revenue down 15%; handle up only 2%
- Promotional spend up 140 basis points
- Full-year leverage guided up from 3.7x to 4.3x versus a 2.5x target
- FLUT fell ~$15 to near $90
- Kalshi running $39.7 billion annualized volume, 87% of it sports
- State gaming taxes as high as 51%, which prediction markets avoid via CFTC self-certification
- 64% implied odds the Supreme Court agrees to hear a prediction-markets case by year-end (36% it does not)
- Burry's IV15: $96 for FLUT; DKNG IV15 'much lower than today's price'
- New Flutter average cost in the $98s
- CEO Peter Jackson stepping down after 9 years, succeeded by Dan Taylor
Key insights
- Flutter's U.S. EBITDA collapse (down 70% to $119 million) sits alongside International EBITDA up 10% to $2.3 billion — the damage is specifically American, which points at a U.S.-only regulatory cause rather than a global demand problem.
- Handle grew only 2% while promotional spend rose 140 basis points: FanDuel is paying more to hold flat volume, the signature of share loss to a cheaper competing venue.
- The guidance cut pushes leverage from 3.7x to 4.3x against a 2.5x target, so the equity is absorbing balance-sheet risk on top of the earnings miss.
- Kalshi's $39.7 billion annualized volume — 87% sports — is the scale at which prediction markets stop being a curiosity and start showing up in incumbent P&Ls.
- The core asymmetry is tax: CFTC self-certification lets prediction markets claim a federal exemption and avoid state gaming taxes of up to 51%, a cost gap no sportsbook can out-market.
- The incumbents have stopped litigating and are instead building their own prediction markets while letting revenue-hungry states carry the legal fight — a cheaper strategy with the same payoff.
- Prediction markets price a 64% chance the Supreme Court takes a case on their own legality/taxation by year-end, which Burry treats as the timing variable for the thesis.
- The trade is a pure relative-value swap, not a sector call: DKNG sold because its IV15 is far below price, FLUT more than doubled because its IV15 of $96 sits above a ~$90 print.
- He expected DraftKings to miss when it reported the next afternoon — the DKNG exit was timed ahead of that print, not after it.
#021
Trading Post
Aug 04, 2026
Trading Post August 4, 2026: My Options
The S&P 500 has broken to new highs on a rare four-day 5% surge whose only historical analogs cluster around major tech tops, but Burry argues the mechanics of the rally — falling headline volatility forcing vol-targeting funds to lever up and dragging other momentum strategies in — mean new highs will still pull money in. Underneath, the damage done in July has not healed: the Goldman momentum pair trade remains broken and the VIX/VIXEQ ratio at 12-year lows shows calm index volatility masking violent component-level moves. He therefore keeps the structural short on, concentrated in semiconductors and momentum leaders, while managing the options book around price: rolling out duration where volatility is tolerable and taking profits or letting positions lapse where premium is too rich. Working forensically through the Big Five hyperscalers' earnings and filings has already changed his positioning — most visibly by removing his one bullish expression, the Microsoft calls.
aimemoryaineocloudairobotaisemienergygridindexusmag5saas
Positions & performance
+21.4%avg return90% win · 10 tracked · S +21.4%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF HOLDputs and short common | NASDAQBitget | I | 542.21closeAug 04 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +74.0% |
| WhyKeeps both the SOXX put position and the short in the ETF itself: semis and momentum are the market's leadership, and he still thinks a major top — possibly a 1987-type break — is plausible. |
| SAMAT:NASDAQ aisemi Applied Materials HOLDcommon | NASDAQS&PBitget | ISPTC | 546.62closeAug 04 '26 | 349.0 ● | 215.50739.67 | +52.2%APY +198.7% | +36.2% |
| WhyKeeps the Applied Materials short alongside MU and NVDA as expression of the semi-cap downside view; currently held at a profit. |
| SCAT:NYSE energygrid Caterpillar HOLDcommon | S&PBitget | IAPT | 876.54closeAug 04 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +34.9% |
| WhyHolds the Caterpillar short through an excellent earnings report that pushed the stock up substantially on the day; the position is still profitable overall and the multi-year thesis stands. |
| SNVDA:NASDAQ aisemi NVIDIA HOLDputs rolled from December 2026 to June 2027, strikes in the low 100s · a portion of the put position | NASDAQS&PBitget | ISPTC | 211.94closeAug 04 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +26.6% |
| WhyRolled part of the December 2026 NVDA puts out to June 2027 while keeping low-100s strikes — extending duration on the bearish semis/AI-capex view rather than taking the position off. |
| SNVDA:NASDAQ aisemi NVIDIA HOLDcommon | NASDAQS&PBitget | ISPTC | 211.94closeAug 04 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +26.6% |
| WhyMaintains the NVIDIA equity short — the one name in the short book where he says he does not currently hold a profit. |
| SMU:NASDAQ aimemory Micron Technology HOLDcommon | NASDAQS&PBitget | ISAPTC | 892.67closeAug 04 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +19.3% |
| WhyStill short Micron as part of the semiconductor complex; he is holding a profit here and reiterates confidence in the multi-year view while pledging to cut losses if it turns. |
| SQQQ:NASDAQ indexus Invesco QQQ Trust (Nasdaq 100) HOLDputs rolled from January to February 2027, strikes in the high 500s · a set of puts | NASDAQBitget | SC | 723.85closeAug 04 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +8.5% |
| WhyRolled a block of QQQ puts out to February 2027 at a small credit with strikes again in the high 500s — a cheap way to keep index-level downside on while the Nasdaq 100 still trades below its all-time high. |
| SPLTR:NASDAQ saas Palantir Technologies HOLDputs · the other half | NASDAQS&PBitget | ISPTC | 162.66closeAug 04 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +1.9% |
| WhyKeeps the remaining half of the Palantir put short un-rolled, again because option prices are too high; the underlying bearish view is unchanged even after what he calls a flashy earnings print. |
| SPLTR:NASDAQ saas Palantir Technologies HOLDcommon | NASDAQS&PBitget | ISPTC | 162.66closeAug 04 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +1.9% |
| WhyMaintains the equity short in Palantir despite the stock rising on earnings, saying he remains confident in his assessment of its prospects over the next few years but will cut losses if it moves against him. |
| STSLA:NASDAQ airobot Tesla HOLDcommon | NASDAQS&PBitget | ISPTC | 327.35closeAug 04 '26 | 379.8 ● | 297.38498.83 | +8.8%APY +24.4% | -16.0% |
| WhyContinues to hold the Tesla equity short as one of the momentum/retail-favourite names he expects to fare badly over the next few years. |
| SORCL:NYSE aineocloud Oracle COVERJanuary 2027 puts, low 100s strikes · entire position | S&PBitget | ISAPTC | 145.74closeAug 04 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +18.3%closed |
| WhyClosed the Oracle puts for a substantial profit and did not roll them; he says he may re-enter Oracle puts if implied volatility comes back down. |
| LMSFT:NASDAQ mag5 Microsoft EXITcalls · entire position | NASDAQS&PBitget | ISAPT | 492.81closeAug 04 '26 | 519.14 ● | 349.20559.15 | +47.1%APY +164.4% | +5.3%closed |
| WhySold the Microsoft call position completely after working through the hyperscaler earnings forensically; he credits the gain to lucky timing and says his enthusiasm for Microsoft's report did not match the market's. |
| SPLTR:NASDAQ saas Palantir Technologies EXITJanuary 2026 puts · half the put position | NASDAQS&PBitget | ISPTC | 162.66closeAug 04 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | +1.9%closed |
| WhySold the January Palantir puts outright and chose not to roll: elevated implied volatility supported the existing position but made replacement puts too expensive to justify. |
Full summary 514 words
This Trading Post lands while Burry is finishing the fourth installment of his Heretic's Guide, a forensic pass through the Big Five hyperscalers' earnings and filings, and he says that work has already reshaped his portfolio. The market backdrop is a fresh S&P 500 all-time high, with the equal-weight index already there and the Nasdaq 100 still lagging. Citing Jonathan Krinsky at BTIG, he notes the index has surged 5% over four trading days into a new high only three other times: 21 March 2000 — effectively the top of the TMT bubble — 23 April 1999, roughly when the first wave of dot-coms began to crack, and 9 November 2020. His read is that a major top, possibly a 1987-style break, remains possible, but that new highs will still draw money in, so he is not fighting the tape blindly.
What actually matters, he argues, is the SOX and the momentum trade. Both were badly hurt in July and the question is whether they can resume leadership; a six-month, 30-minute SOXX chart shows momentum turning back up. Beneath that, though, the market's plumbing is still strained. The Goldman Sachs Momentum Pair Trade remains damaged after a historic run — and the series, which goes back to the GFC, shows how an extended pair trade eventually resolves — while the VIX/VIXEQ ratio sits at 12-year lows. That combination, he explains, means the pod shops and vol-targeting funds burnt in July are still licking their wounds even as a calm surface index volatility conceals substantial component volatility underneath. It is the same market-structure argument he laid out in his 31 July piece on volatility targeting and pod shops: falling headline vol mechanically forces vol-targeters to lever up and pulls other momentum strategies in behind them.
The trade log follows from that. He sold the Microsoft calls outright — he credits the gain to lucky timing and says his enthusiasm for Microsoft's quarter never matched the market's — removing his one bullish options expression. He rolled part of the NVIDIA December 2026 puts out to June 2027, sticking with low-100s strikes, and rolled a set of QQQ puts from January to February 2027 at a small credit with strikes in the high 500s on both legs. He sold the January Palantir puts without rolling them: implied volatility supported the position he held but made replacements too expensive, and he holds the other half of the Palantir short un-rolled for the same reason. He closed the Oracle January 2027 low-100s puts for a substantial profit and left it there, flagging that he may re-enter if volatility falls.
On the equity side he is standing pat: short SOXX, MU, NVDA, CAT, PLTR, TSLA and AMAT, all profitable except NVDA, and held through strong prints from Caterpillar and Palantir that lifted both stocks sharply on the day. He reiterates confidence in his multi-year assessment of these names but states the invalidation plainly — he will cut losses if they move against him — and repeats his standing caveat that shorting is not for everyone. All long positions are unchanged.
Key metrics
- S&P 500 surged 5% over four trading days to a new all-time high — per Jonathan Krinsky (BTIG) only three prior analogs: 3/21/00, 4/23/99 and 11/9/20
- Equal-weight S&P 500 had already made new highs; the Nasdaq 100 remains some way below its all-time high
- VIX/VIXEQ ratio sits at 12-year series lows
- Goldman Sachs Momentum Pair Trade still damaged after a historic run, with data back to the GFC
- SOXX 30-minute chart over six months shows momentum turning back to the upside
- NVDA puts rolled from December 2026 to June 2027, strikes kept in the low 100s
- QQQ puts rolled from January to February 2027 at a small credit, strikes in the high 500s on both legs
- ORCL January 2027 low-100s puts closed for a substantial profit
- Half the Palantir put position sold; the other half held un-rolled
- Short book (SOXX, MU, NVDA, CAT, PLTR, TSLA, AMAT) profitable in every name except NVDA
Key insights
- The only three prior instances of a 5% four-day surge into a new S&P 500 high were 3/21/00, 4/23/99 and 11/9/20 — two of them near major tech tops, which is why Burry still frames a 1987-type fall as possible.
- He is explicit that new highs are self-reinforcing in the short run: rising prices on falling volatility force vol-targeting funds to lever up and pull other momentum strategies in.
- The VIX/VIXEQ ratio at 12-year series lows is his key tell that calm index volatility is masking large single-stock volatility underneath — the condition that produced July's damage.
- The Goldman Sachs Momentum Pair Trade remains impaired after a historic run; with data back to the GFC, he points to how an extended pair trade eventually unwound as the template.
- Selling the Microsoft calls removes his only bullish options expression and is a direct output of his forensic work on the Big Five hyperscalers' earnings and filings.
- Option premium — not conviction — is now driving his sizing: he rolled NVDA (to June 2027, low-100s strikes) and QQQ (to February 2027, high-500s strikes, at a small credit) but let the Palantir puts go because replacement vol was too expensive.
- The Oracle January 2027 low-100s puts were closed at a substantial profit, with an explicit re-entry condition: lower implied volatility.
- He holds the CAT and PLTR shorts through strong earnings that lifted both stocks materially, and discloses NVDA is the only name in the short book not currently showing a profit — with a stated stop-loss discipline if the shorts run against him.
#020
Essayabridged
Aug 03, 2026
[Abridged] Foundations: Market Structure, Volatility Targeting, Pod Shops & Other Gremlins
The condensed version of the July 31 Foundations essay, written for readers who want the mechanism in under 1,500 words. Modern market behavior — ignoring rallies, panicking 2-4x harder on declines, then forgetting the fear twice as fast — is not human, so Burry attributes it to three mechanical forces acting in sequence. Vol-targeting funds, whose exposure is inversely proportional to prevailing volatility, sell convexly and early. Pod shops, forced by risk limits to unwind at many multiples of their actual losses, supply the extra chaos. Both are spent by the time the index is down 3%, leaving a price-insensitive passive bid to repair the damage. The result is a market that structurally cannot get cheap, ratcheting valuation to the 100th percentile of 145 years, and the only thing that would reset it is a drop deep and long enough to make the passive machine reallocate.
Positions & performance
No single-stock positions stated in this report.
Full summary 589 words
This is the free, abridged treatment of the Foundations essay Burry published on July 31, compressed to what he bills as market understanding and "air superiority" in under 1,500 words. It keeps the skeleton of the argument and drops most of the digressions, the autobiography and several supporting datasets.
He opens on the same statistic: 2011-2026 delivered a 12.09% annualized real total return with a dividend contribution of just 1.76%, the 1st percentile across 145 years. He then moves straight to the volatility study that the previous week's momentum unwind prompted. Glosten, Jagannathan and Runkle's 1993 Journal of Finance paper supplies the four parameters — omega as resting nervousness, beta as memory, alpha as the flinch, gamma as the asymmetric extra fear of loss — and running their formulas on the modern era versus 1985-2000 yields roughly 4x the chaos on a big down day, less next-day memory, and fear dissipating about twice as fast. Lo and MacKinlay's 1988 variance-ratio protocol supplies the second: a random walk returns 1.00, trending prints above, and today's market prints below, walking itself back, with moves inside any two- or three-month window increasingly cancelling out. Trading days are actually less volatile now (13.83% to 13.08%) while close-to-close volatility rose (16.64% to 17.50%), derived entirely outside exchange hours.
The abridged version keeps the punchline about the market's inhuman signature — more upset at small sell-offs than large ones, less forceful selling through the close, and the equal-weighted S&P showing little change, which localizes the bid to the concentrated cap-weighted indices — and states the conclusion directly: it takes a village of inhuman institutions to create a market like this.
The mechanism section is intact and is the reason to read it. Vol-targeting is reduced to one equation, Exposure = Target Volatility / Prevailing Volatility, with the convexity spelled out: vol at 5% implies 200% of assets, vol at 20% implies 50%. His scenario has a 2.5% drawdown lifting realized vol to 20% and forcing a cut from 77% to 50% of assets, while a subsequent -4% day trims only another 8 points. The platforms add the gamma: $400-500 billion leveraged through roughly 1,800 pods into several trillion in assets, over 30% of trading volume, crowded into the same trades. His worked example survives the edit — a $500m pod at 6x gross loses $25m, has half its capital pulled, re-levers at 4x, and unloads $2bn of exposure; fifty such pods would need a day and a half of every share traded in the index to clear. Because pod books run roughly half long and half short, the index can look calm in a week the Morgan Stanley Momentum TMT index falls 53.5%.
The closing sections compress the valuation case and the positioning. Over fifteen years the S&P spent only 1.96% of sessions more than 20% below a prior high against a 17.7% median since 1928, with no faster recovery among 1,215 comparable-drawdown blocks. Mean CAPE of 29.57 is the 100th percentile of 145 years of Shiller data, and his own estimate of 44.8-45.7x clears the 1999 record of 44.20. He points readers to AP SBC, D'AI of the Triffids, Heretic's Guide Part III, the March Market Structure piece and the July 23-24 posts. One notable difference from the full version: here he frames the reset condition as a 25% loss maintained over two years rather than the 400-day figure. Positioning is one sentence — long undervalued, out-of-favor securities, short highly valued momentum favorites — with the usual caveat that most people should not short.
Key metrics
- 2011-2026 delivered 12.09% annualized real total return with dividend contribution of only 1.76% — 1st percentile across 145 years
- A big down day today triggers ~4x the chaos of 1985-2000, with ~2x faster decay of residual fear (GJR 1993)
- S&P intraday volatility 13.83% (1985-2000) vs 13.08% (2011-today); close-to-close rose 16.64% to 17.50%, entirely outside exchange hours
- Vol-targeting rule: Exposure = Target Volatility / Prevailing Volatility — vol at 5% implies 200% of assets, vol at 20% implies 50%
- A 2.5% drawdown lifting realized vol to 20% forces a cut from 77% to 50% of assets; a further -4% day trims only 8 more points (33% to 25%)
- Pod shops: $400-500 billion leveraged through ~1,800 pods into several trillion in assets, controlling over 30% of trading volume
- A $500m pod at 6x ($3bn book) losing $25m has half its capital pulled, re-levers at 4x, and dumps $2bn of exposure; 50 such pods need 1.5 days of all index volume to clear
- Morgan Stanley Momentum TMT index fell 53.5% in a week the S&P looked calm
- S&P spent only 1.96% of sessions more than 20% below a prior high over 15 years vs a 17.7% median since 1928; none of 1,215 comparable-drawdown blocks recovered faster
- Mean 15-year CAPE of 29.57 is the 100th percentile of 145 years of Shiller data; his S&P 500 CAPE estimate of 44.8-45.7x clears the 1999 record of 44.20. Stated reset condition here: a 25% loss maintained over 2 years
Key insights
- This is the free/abridged companion to the July 31 Foundations essay, aimed at delivering the mechanism in under 1,500 words; the analytical spine survives but the supporting datasets and cross-references are thinned.
- The vol-targeting equation is presented as the single most portable takeaway: Exposure = Target Volatility / Prevailing Volatility, with 1/x convexity meaning exposure doubles to 200% at 5% vol and halves to 50% at 20% vol.
- The front-loading point explains the market's odd signature — a 2.5% drawdown forces a cut from 77% to 50% of assets while a further -4% day trims only 8 more points, so the chaos is worse at small declines than large ones.
- The pod amplification example is retained in substance: a $500m pod at 6x turns a $25m loss into $2bn of forced exposure reduction, and 50 such pods would need 1.5 days of total index volume to exit.
- Because pod books are roughly half long and half short, index-level damage is muted while single stocks are destroyed — the Morgan Stanley Momentum TMT index fell 53.5% in a week the S&P looked calm.
- The reset condition differs between versions: the abridged post says a 25% loss maintained over two years, while the full July 31 essay specifies a 25% drop held more than 400 days on the 1987 template.
- Valuation conclusion is unchanged — 15-year mean CAPE of 29.57 at the 100th percentile of 145 years, and an adjusted S&P CAPE of 44.8-45.7x above the 1999 record of 44.20.
- Positioning is stated in one line with no new names: long undervalued out-of-favor securities, short highly valued momentum favorites, plus his standing warning that most people should not short.
#019
Essay
Jul 31, 2026
Foundations: Market Structure & What The Heck Just Happened
The modern market's behavior cannot be explained by human psychology, so it must be explained by machinery. Running GJR (1993) and Lo-MacKinlay (1988) on 1985-2000 versus 2011-2026, Burry finds a market that ignores rallies entirely, panics with roughly 4x historical intensity on big declines, then forgets that fear twice as fast — and that reverts rather than trends day to day. He attributes the pattern to a village of mechanical actors: vol-targeting funds whose exposure is inversely proportional to prevailing volatility, so they dump convexly and early; pod shops whose risk limits force unwinds at 40-200x their actual losses; and a price-insensitive passive bid that shows up on every decline and repairs the damage. The consequence is not that these players make stocks expensive — none has a view on price — but that they disable the process by which markets used to get cheap, so multiples ratchet up and never reset. He is not calling a top; he is naming the condition for one, which is a drawdown deep and long enough to make the passive machine reallocate.
Positions & performance
No single-stock positions stated in this report.
Full summary 600 words
Burry opens with the statistic that frames everything after it: over 2011-2026 the S&P 500 delivered a 12.09% annualized real total return, but dividends contributed only 1.76% — the 1st percentile across 145 years of 15-year blocks. Companies replaced the dividend — the only non-reversible component of shareholder return, he argues — with buybacks and RSU withholding: a cash transfer to employees invisible to earnings and to standard free-cash-flow math that often increases share count. That throws the whole return onto earnings growth, and therefore onto earnings quality, where his prior work lives.
The body is a volatility study prompted by July's momentum unwind and snapback. He disclaims ever using the Greeks, then borrows them from Glosten, Jagannathan and Runkle's 1993 Journal of Finance paper — omega as resting nervousness, beta as memory, alpha as the flinch, gamma as the asymmetric extra fear of loss that Fischer Black described in 1976. Run on 1985-2000 versus today, a big down day now triggers roughly 4x the chaos, with less next-day memory and fear draining about twice as fast. Lo and MacKinlay's 1988 variance-ratio protocol supplies the second finding: a random walk returns 1.00, trending prints above, mean reversion below. The old market printed 1.02; today it prints 0.881, z = -2.56 across 3,774 sessions.
Supporting facts sharpen it. Intraday volatility fell (13.83% to 13.08%) while close-to-close rose (16.64% to 17.50%), so the entire increase happened outside exchange hours. The S&P now closes at 0.154 of its daily range against roughly 0.09 in every decade since the 1980s, and 2%+ down days close in the bottom quarter of range only 81% of the time in the 2020s versus about 90% historically — evidence of something bidding the lows, and against forced deleveraging, which would sell through the close. The equal-weighted S&P shows no such change, localizing the bid to the cap-weighted indices. No human, he concludes, gets 4x as upset, buys the dip anyway, and recovers twice as fast. It takes a village of institutions.
He names the village. Vol-targeting funds — risk parity, CTAs, insurer annuity hedges, bank VaR desks — run Exposure = Target Volatility / Prevailing Volatility, a convex 1/x function. A 2.5% drawdown lifting realized vol to 20% forces a cut from 77% to 50% of assets, while a further -4% day trims only 8 more points — the violence is front-loaded. He sizes the pool near $500 billion against a $60 trillion market, but stresses it arrives at once in S&P futures. The platforms add the gamma: roughly $428 billion leveraged through 1,400-1,800 pods, over 30% of volume, crowded into one TMT-long/momentum-short book. His centerpiece example: a $500m pod at 6x gross loses 5% of capital on a 0.83% book move, is halved, re-levers at 4x, and turns a $25m loss into $2bn of forced turnover — 80x. Fifty such pods would need a day and a half of all Nasdaq 100 volume, where 99.77% of shares sit still.
Positioning follows. He is not calling a top — shocks now get repaired faster than ever — but names the condition: a 25% drop held more than 400 days, the 1987 template, enough to clear the pod shops and perhaps push the passive machine toward 6% bonds. Until then multiples ratchet: a 15-year mean CAPE of 29.57 is the 100th percentile of Shiller's 145 years, and his adjusted S&P CAPE of 44.8-45.7x clears the 1999 record of 44.20. He says watch long bonds, citing the 10-year at 4.7347%. His book mirrors the pod shops': long out-of-favor names that have based, short the AI narrative and the memory cycle.
Key metrics
- S&P 500 delivered 12.09% annualized real total return 2011-2026 with dividend contribution of only 1.76% — 1st percentile across 145 years of 15-year blocks
- A big down day today triggers ~4x the chaos of 1985-2000, but residual fear dissipates ~2x as fast (GJR 1993 framework)
- Two-day Lo-MacKinlay variance ratio is 0.881 (z = -2.56, 3,774 sessions) vs 1.02 in 1985-2000 — the market now walks itself back rather than trending
- Intraday S&P volatility fell from 13.83% to 13.08% while close-to-close rose from 16.64% to 17.50% — the entire increase occurred outside exchange hours
- Vol-targeting capital ~$500 billion against a $60 trillion US market — but it arrives all at once in S&P 500 futures, where the marginal dollar sets price
- Pod shops control ~$428 billion of investor capital (Pensions & Investments, April) leveraged through 1,400-1,800 pods, and can be over 30% of daily trading volume
- A $500m pod at 6x gross loses 5% of capital on a 0.83% adverse move; halved and re-levered at 4x, a $25m loss forces $2bn of turnover — 80x. Nasdaq 100 trades only ~0.23% of shares outstanding a day, so 99.77% sits still
- S&P spent just 1.96% of sessions more than 20% below a prior high over 15 years vs a 17.7% median since 1928; 1987's -33.5% took 414 sessions to a new high vs 103 for 2020's -33.9%
- Mean 15-year CAPE of 29.57 is the 100th percentile of 145 years of Shiller data; his adjusted S&P 500 CAPE of 44.8-45.7x clears the 1999 record of 44.20
- 10-year Treasury at 4.7347%, an 18-month high; his stated reset condition is a 25% index drop maintained more than 400 days
Key insights
- The dividend contribution to the S&P's 2011-2026 real return was 1.76%, the 1st percentile of 145 years — buybacks and RSU withholding replaced the only non-reversible component of shareholder return with an employee cash transfer invisible to earnings and standard FCF.
- GJR (1993) applied to 1985-2000 vs today shows ~4x the chaos on a big down day but ~2x faster decay of residual fear, while Lo-MacKinlay's variance ratio fell from 1.02 (random walk) to 0.881 with z = -2.56 — the market now mean-reverts rather than trends.
- The entire rise in close-to-close volatility (16.64% to 17.50%) occurred outside exchange hours while intraday volatility actually fell, meaning risk has migrated to the overnight gap.
- Evidence points away from forced deleveraging: 2%+ down days close in the bottom quarter of range only 81% of the time in the 2020s vs ~90% historically, and the effect vanishes in the equal-weighted S&P — so the mechanical bid is concentrated in cap-weighted indices.
- Vol-targeting's convex 1/x exposure rule front-loads the selling: a 2.5% drawdown forces a cut from 77% to 50% of assets while a subsequent -4% day trims only 8 more points, which is why the market is more chaotic at 2-3% declines than at larger ones.
- The pod-shop amplification math is the essay's sharpest number: a $500m pod at 6x turns a $25m loss (0.83% of book) into $2bn of forced turnover — 80x — and fifty such pods out of ~1,800 would need 1.5 days of all Nasdaq 100 volume to exit.
- Repair speed has quadrupled: 1987's -33.5% took 414 sessions to a new high, 2020's -33.9% took 103. Over 15 years the S&P spent just 1.96% of sessions more than 20% underwater vs a 17.7% historical median, and none of 1,215 comparable-drawdown blocks recovered faster.
- He declines to call a top and instead names the condition for a real reset: a 25% drop held more than 400 days, which would clear the pod shops and vol traders and possibly push the passive bid toward 6% ten-year bonds.
#018
Trading Post
Jul 30, 2026
Trading Post July 30, 2026: A Sizable Batch of Trades, But Only Two Kinds of Charts
The one-day collapse and immediate snapback in the Momentum Pair trade was a historic reversal — larger, he says, than the analogous episode 26 years ago — and the market's reflex has been to put the trade straight back on. Burry reads the rebound as tired rather than durable, and uses it to press both sides of his book. On the long side he keeps buying out-of-favor names that have spent a long time falling and are now basing, which he interprets as shares turning over into stronger hands and therefore de-risked for long-term holders. On the short side he adds to the semiconductor and AI complex through Micron, SOXX, and dated puts on Nvidia and the Nasdaq 100. The structural argument behind the trades is deferred to a forthcoming piece on hyperscaler earnings quality.
aimemoryaisemiconsumershoppingconsumertravelhealthindexus
Positions & performance
+13.2%avg return71% win · 7 tracked · L -7.0% · S +28.4%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDcommon | NASDAQBitget | I | 506.0his fillJul 30 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +72.1% |
| WhyBasket short on semis at roughly $506, sized up alongside the Micron short to press the semiconductor leg of the AI trade rather than relying on single names. |
| SNVDA:NASDAQ aisemi Nvidia ADDDec 18, 2026 puts, low $100s strike range | NASDAQS&PBitget | ISPTC | 195.04closeJul 30 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +20.2% |
| WhyCore short against the AI narrative, expressed in dated puts rather than stock. Added on the same day he judged the momentum snapback to be a tired knee-jerk reaction. |
| SMU:NASDAQ aimemory Micron Technology ADDcommon | NASDAQS&PBitget | ISAPTC | 880.0his fillJul 30 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +18.2% |
| WhyAdds to the memory-cycle short at roughly $880 — the cyclical, highly valued momentum favorite he repeatedly names as the other side of the pod shops' crowded long book. |
| LZTS:NYSE health Zoetis ADDcommon · full position | S&P | IAPTC | 76.0his fillJul 30 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +11.6% |
| WhySame setup as DraftKings and lululemon — a base/consolidation following a long decline, with shares turning over into stronger hands. The add takes it to a full position. |
| SQQQ:NASDAQ indexus Invesco QQQ Trust (Nasdaq 100 ETF) ADDJan 15, 2027 puts, high $500s strike range | NASDAQBitget | SC | 683.55closeJul 30 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +3.1% |
| WhyIndex-level expression of his bearish view on the AI/momentum complex, added into the sharp rebound in the Momentum Pair trade that he judges to be running out of legs. |
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon · full position | NASDAQS&P | ISAPT | 118.0his fillJul 30 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -10.7% |
| WhyContinues the position he opened with his June 24 LULU deep dive; the stock is basing after a long fall and he views it as de-risked at $118 for long-term holders. Now a full position. |
| LDKNG:NASDAQ consumertravel DraftKings ADDcommon · with Flutter, a larger position | NASDAQBitget | IAT | 23.4his fillJul 30 '26 | 18.29 ● | 16.5536.98 | -32.7%APY -66.5% | -21.8% |
| WhyBasing and consolidating after a long fall, which he reads as turnover of shares into stronger hands. He considers the buy de-risked at this price for long-term holders, and sizes it alongside Flutter as one of his larger positions. |
Full summary 443 words
This is a short, dense trade log posted the day after a violent unwind and same-day reversal in the Momentum Pair trade. Burry opens by flagging what is coming rather than what just happened: he is preparing a post on Microsoft, Meta and the other hyperscalers, pending the earnings and filings landing that day and the next, and says there is "a ton of material" with "lots of accounting maneuvers and poor earnings quality on display." In the interest of time he skips the analysis and runs straight through the trades.
Every action in the post is a buy or an addition; there are no sells. On the long side he added to DraftKings around $23.40, noting that together with Flutter it constitutes one of his larger positions. He added to Zoetis around $76 and to lululemon around $118, both now full positions. He groups all three under a single pattern: a basing or consolidation phase following a long fall. His reading of that pattern is that it represents turnover of shares into stronger hands, and he states plainly that he believes these three buys are "rather de-risked at these prices," at least for long-term holders.
The short side is expressed mostly through options and baskets rather than single-stock borrow. He added to QQQ puts expiring January 15, 2027 in the high $500s strike range, and to Nvidia puts expiring December 18, 2026 in the low $100s strike range. He shorted more Micron at roughly $880 and more of the semiconductor ETF SOXX at roughly $506. The pairing is deliberate: cheap, washed-out consumer and healthcare names against the most highly valued momentum favorites in semis and AI.
On market structure he is brief but pointed. The knee-jerk reaction to the previous day's Paired Momentum unwind has been to put the trade right back on. He calls the move a historic reversal, "even more so than what happened 26 years ago," and concedes that with so many oversold and overbought signals firing, a reversal was to be expected. But his verdict on the trade's durability is one line: "the legs, they look tired." He includes charts of the extreme moves in the popular long-momentum and short-momentum baskets, and jokes that there are more than two shapes of charts in the market — just not many more.
He closes by saying he is working on several articles and is not sure which will finish first, with a teaser that "we are dealing with a giant." The post carries no invalidation condition or price target; it is positioning ahead of the hyperscaler earnings work and ahead of the market-structure essay he would publish the following day.
Key metrics
- DKNG added around $23.40
- ZTS added around $76, now a full position
- LULU added around $118, now a full position
- QQQ Jan 15, 2027 puts added in the high $500s strike range
- NVDA Dec 18, 2026 puts added in the low $100s strike range
- MU shorted more at ~$880
- SOXX shorted more at ~$506
- All buys and additions in this post — no sells
Key insights
- Every trade in the post is a buy or an add — Burry explicitly notes there are no sells, which is unusual for a Trading Post covering a violent two-day market move.
- DKNG at ~$23.40 and FLUT together form one of his larger positions, making sports betting a meaningful long-side allocation rather than a starter.
- ZTS at ~$76 and LULU at ~$118 both reached full-position size on this day; LULU extends the thesis from his June 24 deep dive on the company's management.
- His long-side screen is explicitly technical, not just fundamental: a long fall followed by basing/consolidation, which he interprets as share turnover into stronger hands and therefore a de-risked entry.
- The AI/semis short is expressed four ways at once — MU common at ~$880, SOXX at ~$506, NVDA Dec 18 2026 puts in the low $100s strikes, and QQQ Jan 15 2027 puts in the high $500s strikes — spreading the bet across single name, basket, and index.
- He read the same-day snapback in the Momentum Pair trade as a knee-jerk, calling the reversal more extreme than the analogous 1999-2000 episode and judging the trade's remaining legs "tired."
- The post pre-announces a hyperscaler earnings-quality piece on Microsoft and Meta, flagging "accounting maneuvers and poor earnings quality" ahead of the filings — a signal of where the next short-side work was headed.
#017
Trading Post
Jul 24, 2026
Trading Post July 24, 2026 - 4 Shorts, 3 Longs
On a busy trading day Burry bought three longs and expanded four shorts, all expressions of theses he says only work if he stays patient. The longs are a paired bet that regulated sportsbooks Flutter and DraftKings are mispriced by fear of prediction markets, plus an add to managed care in Molina. The shorts extend his central claim that AI demand is not real end demand: per the 2026 BIS annual report, much and possibly most of Nvidia's current and future revenue is financed off-balance-sheet in a circular arrangement, so he added to NVDA and to the broader SOXX complex, plus Micron and Caterpillar. He remains short Tesla, which he says shrinks on its own, QQQ through puts, and Palantir - whose successful protest forcing the DIA to abandon its own fully operational intelligence system he reads as a company that must litigate to keep its customer.
aimemoryairobotaisemiconsumertravelenergygridhealthindexussaas
Positions & performance
+8.7%avg return60% win · 10 tracked · L -7.9% · S +15.8%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF ADDcommon; plus puts · large position together with puts | NASDAQBitget | I | 535.83his fillJul 24 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +73.7% |
| WhyAdded at $535.83 to express the circular-financing thesis across the whole semiconductor complex rather than a single name. Combined with his puts he calls it a large position, making it one of the primary vehicles for the AI short. |
| SCAT:NYSE energygrid Caterpillar ADDcommon | S&PBitget | IAPT | 893.49his fillJul 24 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +36.1% |
| WhyHe sold short more Caterpillar at $893.49, adding to an existing short rather than initiating. No new rationale is stated in this post - it is presented alongside the AI-capex-linked shorts as part of the same patient, already-researched thesis. |
| SNVDA:NASDAQ aisemi Nvidia ADDcommon; continues to hold puts in good size · puts held in good size | NASDAQS&PBitget | ISPTC | 210.28his fillJul 24 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +26.0% |
| WhyHe believes much and possibly most of Nvidia's current and future demand is not driven by end customers or end demand but is financed, off-balance-sheet and not lit. Per the 2026 BIS annual report, future revenues are majority financed in a circular arrangement, which is the core of the short. |
| SMU:NASDAQ aimemory Micron Technology ADDcommon | NASDAQS&PBitget | ISAPTC | 933.86his fillJul 24 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +22.9% |
| WhyHe sold short more Micron at $933.86, adding to the semiconductor short complex alongside NVDA and SOXX on the same day. No separate rationale is stated here, but it sits directly in the group of names exposed to his circular-financing thesis on AI demand. |
| LMOH:NYSE health Molina Healthcare ADDcommon | — | ITC | 197.02his fillJul 24 '26 | 233.09 ● | 121.06244.89 | +18.3%APY +69.7% | +18.3% |
| WhyHe added to an existing managed-care long at $197.02, one of only three buys on a day otherwise spent expanding the short book. No fresh rationale is given in this post; it is framed under his closing point that these are long-held, well-researched theses that require patience to size up. |
| SQQQ:NASDAQ indexus Invesco QQQ Trust HOLDputs | NASDAQBitget | SC | 684.23closeJul 24 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +3.2% |
| WhyHe remains short the Nasdaq-100 via puts, the index-level hedge that sits behind his single-name AI shorts. No change or price is stated on the day. |
| LDKNG:NASDAQ consumertravel DraftKings ADDcommon · with FLUT one of his larger positions | NASDAQBitget | IAT | 23.07his fillJul 24 '26 | 18.29 ● | 16.5536.98 | -32.7%APY -66.5% | -20.7% |
| WhyThe second leg of the same play against prediction markets, bought at $23.07. He explicitly sizes FLUT and DKNG as a combined position, so the bet is on the regulated sportsbook duopoly rather than on either single name. |
| STSLA:NASDAQ airobot Tesla HOLDcommon | NASDAQS&PBitget | ISPTC | 313.03closeJul 24 '26 | 379.8 ● | 297.38498.83 | +8.8%APY +24.4% | -21.3% |
| WhyHe has not covered the Tesla short and states it gets smaller all on its own - i.e. the position is shrinking as the stock falls, so he is letting the winner run rather than adding or covering. |
| LFLUT Flutter Entertainment ADDcommon · a decent amount; with DKNG one of his larger positions | — | AT | 100.72his fillJul 24 '26 | 79.15 ● | 72.56229.86 | -26.0%APY -56.6% | -21.4% |
| WhyBought as a play against prediction markets - the incumbent sportsbooks are being priced as if prediction-market platforms take their business, and he is taking the other side. Added together with DraftKings, the pair makes up one of his larger positions. |
| SPLTR:NASDAQ saas Palantir Technologies HOLDcommon | NASDAQS&PBitget | ISPTC | 122.92closeJul 24 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | -29.8% |
| WhyHe remains short and reads the day's Bloomberg story as a warning sign rather than a win: Palantir's protest forced the DIA to withdraw a bid to improve a military intelligence system that had already reached full operational capability in March 2026. He argues a company that has to litigate to keep its customer will, like every other enemy of the United States, lose the war and should get used to fighting for its food. |
Full summary 598 words
This is a no-frills Trading Post covering a heavy day of activity, followed by a short essay on patience and a note on Palantir. Burry bought a number of stocks and sold short a number of stocks, and simply walks through them.
On the long side, he bought a decent amount of Flutter (FLUT) at $100.72 and DraftKings (DKNG) at $23.07. He explicitly frames the two together as a play against prediction markets, and says that added together FLUT and DKNG make one of his larger positions - a bet that the regulated sportsbook incumbents are being marked down on prediction-market disruption fears. He also added to his Molina Healthcare long at $197.02, without further comment.
On the short side he sold short more Caterpillar (CAT) at $893.49 and more Micron (MU) at $933.86, both adds to existing positions stated without separate rationale. The reasoning he does give attaches to Nvidia: he added to his NVDA short at $210.28 and continues to hold puts in good size, because he believes much of current and future demand is not driven by end customers or end demand. Much and possibly most is financed, off-balance-sheet and not lit, and future revenues are majority financed in a circular arrangement per the 2026 BIS annual report - the same evidence he unpacked in the previous day's Trading Post. He extended the same view across the sector by adding to his iShares Semiconductor ETF (SOXX) short at $535.83, which together with his puts he calls a large position. Three shorts he is holding without action: he has not covered Tesla (TSLA), noting it gets smaller all on its own; he remains short the Invesco QQQ Trust (QQQ) through puts; and he remains short Palantir (PLTR).
The middle of the post is a short meditation on patience. None of this works, he writes, if he is not patient with his well-researched theses. Patience is the single most difficult skill to master in investing and also the most impactful, because the more patience one has the larger a position can be. He warns against confusing patience with foolhardiness - a hardy fool is just a hardy fool - and cites Warren Buffett ruing his own impatience over an investment, telling shareholders they would have been better off if he and Charlie had just gone to the movies.
The Palantir note is the substantive news of the day, drawn from a Bloomberg story by Katrina Manson: a US government agency withdrew a bid to improve a military intelligence system central to lethal targeting after a protest from Palantir, complicating the Defense Intelligence Agency's effort to enhance an AI-enabled system used for worldwide military operations including targeting. Palantir's dozen-page May 18 protest, cited in a June 3 GAO decision notice reviewed by Bloomberg, argued the DIA was 'impermissibly seeking a developmental solution when a commercial solution is already available.' Burry notes the MARS program had advanced significantly since a damning 2020 GAO assessment and, per a DIA spokesperson, reached 'full operational capability' in March 2026. He calls it a disturbing little battle won by Palantir - the military forced to buy from Palantir despite already having a perfectly good system - and concludes that like every other enemy of the United States, Palantir will lose the war and should start getting used to having to fight for its food.
He closes with work in progress - Temple & Webster, Software Part III, MELI, Hong Kong Part III - and a recommendation of the New York Times Machine archive, searchable back into the 1800s, for perspective and new ideas.
Key metrics
- Flutter (FLUT) bought at $100.72
- DraftKings (DKNG) bought at $23.07
- Molina Healthcare (MOH) added at $197.02
- Caterpillar (CAT) sold short at $893.49
- Micron (MU) sold short at $933.86
- Nvidia (NVDA) short added at $210.28
- iShares Semiconductor ETF (SOXX) short added at $535.83
- 2026 BIS annual report: future AI revenues majority financed in a circular arrangement
- DIA's MARS program reached 'full operational capability' in March 2026, per a DIA spokesperson
- Palantir's dozen-page May 18 protest, and a June 3 GAO decision notice reviewed by Bloomberg, preceded the withdrawn DIA bid
- Palantir argued the DIA was 'impermissibly seeking a developmental solution when a commercial solution is already available'
- A damning 2020 GAO assessment of the MARS program preceded its improvement
Key insights
- FLUT and DKNG are a single paired bet, not two ideas: bought at $100.72 and $23.07 respectively and sized together as one of his larger positions, explicitly as a play against prediction markets.
- The Nvidia short rests entirely on demand quality, not valuation: he argues much and possibly most of current and future demand is financed, off-balance-sheet and not lit, with future revenues majority financed circularly per the 2026 BIS annual report.
- SOXX at $535.83 plus puts is his sector-level expression of the same thesis - a large position that hedges the risk of being wrong on Nvidia specifically while right on AI financing generally.
- Micron at $933.86 and Caterpillar at $893.49 were adds to existing shorts with no new stated rationale, indicating conviction adds on strength rather than new ideas.
- He is not covering Tesla because the position 'gets smaller all on its own' - a statement that the short is working and being allowed to decay naturally rather than harvested.
- QQQ puts remain the index-level overlay behind the single-name AI shorts, so the book is layered rather than concentrated in one expression.
- Palantir's protest forced the DIA to withdraw a bid to improve a targeting-related intelligence system that had reached 'full operational capability' in March 2026 - Burry reads a company litigating to block a customer's in-house alternative as a sign of a business that will have to fight for its food.
- His stated framework for sizing is patience: the more patience one has, the larger a position can be - which is how he justifies carrying large, long-duration shorts against a momentum market.
- Pipeline disclosed: Temple & Webster, Software Part III, MELI and Hong Kong Part III are the write-ups in progress.
#016
Short Thoughts
Jul 24, 2026source ↗
Short Thoughts July 24, 2026 - Offshore Insurers, Meet the Hyperscalers
Private equity firms have bought life insurers and stuffed them with opaque private credit, using affiliated offshore reinsurers in Bermuda to keep the resulting leverage off balance sheet while ratings stay pristine. Because state guaranty funds cover policyholder losses and those assessments are creditable against state premium taxes, the ultimate loss-absorber is the taxpayer, and no one in the chain - policyholders, surviving insurers, regulators - has an incentive to monitor. The assets being fed into that machine are increasingly structured securities backed by data center and chip leases, which is where the private credit shell game meets the hyperscalers' own off-balance-sheet financing. The trigger is rates: private credit and PE debt is overwhelmingly floating rate, and with the 10-year at 4.68% the can-kicking gets more expensive by the month. When credit withdraws funding from the data center buildout - now itself a growing share of US GDP - the contagion runs from private credit through insurers to the real economy.
mag5
Positions & performance
Full summary 600 words
This is a deliberately short post built around one document: 'Private Credit's State Backstop: How Private Equity Socializes Risk Through Insurers,' a 65-page academic paper by Andrew Granato and Pranjal Drall. Burry establishes their credentials - Granato an assistant professor at the University of Texas School of Law with a Stanford economics B.A. and both a Yale J.D. and a Yale Ph.D. in financial economics; Drall a Yale doctoral fellow pursuing both degrees - because the paper formalizes a subject he has been pushing since The Heretic's Guide to AI's Stars Part III (May 22) and building toward in the unpublished Part IV.
The argument runs through long quotations. PE firms have acquired large life insurers and loaded their balance sheets with private credit assets that are opaque and hard for regulators to value. When such an insurer becomes insolvent, state guaranty funds assess surviving insurers to cover the shortfall - and in most states those outlays are fully creditable against state premium taxes, so the taxpayer is the terminal loss-bearer. Burry starts with the endgame from page 30: a distressed insurer approaching receivership has every incentive to expand rapidly, selling more policies and hiring more brokers because premiums are cash now against obligations due later, effectively borrowing from lenders who will not check the borrower's distress. Meanwhile surviving insurers, whose assessments depend on recent premium volume, pull back sales precisely when the guaranty fund needs the assessment base most, so the bailout grows in the final days.
On the liability side comes 'shadow insurance': ceding policyholder liabilities to affiliated reinsurers under lighter regulation, whose balance sheets need not be disclosed on the cedent's, making it extremely difficult to judge whether reinsurer assets support the liabilities. The motive is structural. The biggest PE firms went public between 2007 and 2014, creating permanent pressure to grow AUM; a life insurer supplies a stable, fee-generating asset base that never has to be liquidated to pay outside limited partners, unlike a traditional fund's decade-long cycle. McKinsey calls this a 'virtuous flywheel.' Burry's version: when the asset manager wants an underperforming loan out of an affiliated private credit fund, the insurer becomes the convenient buyer at generous prices - and empirical evidence shows PE-owned insurers pay more for identical assets when the seller is an affiliate.
His own contributions are the visuals the paper lacks. He shows 16.6x leverage at the captive offshore insurer Athene Annuity Re - which is then leveraged again - and says he has seen mid double-digit leverage of 35x and more, versus the 30x that killed Long Term Capital Management in 1998. He charts the duration mismatch between the assets and liabilities, the same mechanism behind the 2023 failures of Silicon Valley Bank, First Republic and Signature Bank. He compiles regulatory data showing PE sponsors dominate Bermuda's captive insurance landscape, and traces private credit and insurance money through a recent Meta data center financing deal. Nikkei Asia's report this week on the hyperscalers' massive off-balance-sheet leverage is the other half of the join: software loans, real estate loans, data center loans and chip loans all find a home in Bermuda under the invisible shell, while ratings never move because the NAIC's post-2009 treatment lets below-investment-grade ABS qualify for minimum-risk capital if held at a steep enough discount.
The reckoning comes from rates. The 10-year closed at 4.68%, and with floating rates the rule across private credit and PE-land, higher-for-longer is the catalyst that ends the rabid can-kicking. No trades are stated here - this is thesis groundwork for the credit and AI-infrastructure short book rather than a Trading Post.
Key metrics
- Granato & Drall paper 'Private Credit's State Backstop: How Private Equity Socializes Risk Through Insurers' - 65 pages, 1.38MB PDF, zero visuals
- 16.6x leverage at the captive offshore insurer Athene Annuity Re, which is then leveraged again
- Burry says he has seen mid double-digit leverage in the system - 35x and more
- Long Term Capital Management failed in 1998 with sure-fire bets at 30x leverage
- 10-year Treasury closed at 4.68% the day of the post
- 10-year Treasury yield chart shown over the last 5 years
- The biggest PE firms went public in a wave between 2007 and 2014, creating pressure to grow AUM consistently
- First wave of PE entry into life insurance was 2009-2014; PE-owned insurers increased private-label ABS holdings by two-thirds of the industry average
- After 2009 the NAIC permitted below-investment-grade ABS to qualify for minimum-risk treatment if held at a steep enough discount
- 2024 data: PE-owned insurers with bond portfolios far more concentrated in asset-backed and structured securities than the industry still carry very-high-quality ratings
- An ordinary PE fund must raise, deploy, harvest and return capital within about a decade - the constraint insurance permanent capital removes
- Duration mismatch took down Silicon Valley Bank, First Republic and Signature Bank in 2023
Key insights
- Guaranty funds plus premium-tax credits destroy monitoring incentives at every level: policyholders do not monitor insurers, and insurers do not monitor each other, because assessments are creditable against state taxes.
- A failing PE-owned life insurer has an incentive to grow fastest right before receivership - selling policies and hiring brokers for immediate cash - which makes the eventual taxpayer bill larger, not smaller.
- Burry shows 16.6x leverage at Athene Annuity Re that is then re-levered, and says he has seen 35x and more in the system, against the 30x that destroyed LTCM in 1998.
- 'Shadow insurance' - ceding liabilities to affiliated offshore reinsurers whose balance sheets are not disclosed on the cedent's - is the mechanism that makes the true location of risk unknowable to outsiders.
- The 2007-2014 wave of PE IPOs is the root cause: public shareholders reward stable fee income, and a life insurer is permanent capital that never has to be liquidated to pay LPs.
- Affiliate transactions are the flywheel's dirty secret - PE-owned insurers demonstrably pay higher prices for identical assets when the seller is an affiliate, letting underperforming loans be dumped from private credit funds onto insurance balance sheets.
- The assets increasingly come off data center and chip leases, which is how a private credit accident becomes an AI capex accident and, given the buildout's share of US growth, a GDP accident.
- Post-2009 NAIC rules let below-investment-grade ABS carry minimum-risk treatment if held at a steep enough discount, so PE-owned insurers hold far more structured product than the industry and still show very-high-quality ratings.
- The catalyst Burry names is higher-for-longer rates against overwhelmingly floating-rate private credit - the 10-year at 4.68% is what has driven the recent egregious can-kicking.
Trading Post July 23, 2026 - 3 Stock Buys & the BIS Weighs in on AI's Rampant Circular Financing
The Momentum Pair Trade run by multi-strategy pod shops - long the highest-momentum cohort, short the worst - has reached crowding extremes, which mechanically leaves genuinely cheap, good businesses trading at the wrong price. Burry is therefore taking the other side: buying deep value, principally in Hong Kong and Australia, which he frames as shorting the pod shops by proxy. On the other side of his book, the BIS annual report now provides official confirmation that the AI complex is being funded circularly: on announced multi-year commitments, circular financing accounts for nearly 100% of chip makers' forward revenue and roughly two-thirds of the revenue and spending of the AI labs and hyperscalers. If forward AI revenue is largely vendor-financed rather than end-demand driven, corporate credit reprices on the first disappointment and the entire trifecta - labs, hyperscalers, chip makers - is exposed. Nvidia's rebuttal that its strategic investments are a 'minimum portion of revenue' is backward-looking and therefore beside the point.
aisemi
Positions & performance
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| L0700 HK Tencent Holdings ADDcommon | — | — | 448.6his fillJul 23 '26 | — | — | — | n/a |
| WhyHe is buying the cheap, good businesses on the losing side of the crowded Momentum Pair Trade - effectively shorting the pod shops by going long deep value - and Hong Kong is where that discount is widest. Tencent is the name he introduced in his February 26 post Hong Kong Stocks: Structure and Strategy as an all-time great compounder, and the volume indicators he described in the May 25 Short Thoughts support the add. |
| L3690 HK Meituan ADDcommon | — | — | 87.6his fillJul 23 '26 | — | — | — | n/a |
| WhyThe second of his three buys, a Hong Kong name he says he deliberately hid inside the May 25, 2026 Short Thoughts post rather than flagging outright. Same logic as Tencent: deep-value Hong Kong compounders being discarded by the negative-momentum leg of the pod-shop pair trade. |
| LTPW AU Temple & Webster ADDcommon · a few more shares | — | — | 5.0his fillJul 23 '26 | — | — | — | n/a |
| WhyHe picked up a few more shares in Australia at 5.00 AUD while finishing a full write-up on the name, which he says will not be long in coming. It fits the same bucket of cheap good businesses getting the short end of the momentum stick. |
Full summary 556 words
Burry's July 23, 2026 Trading Post has two halves: three buys, and an update to his AI circular-financing work using the Bank for International Settlements annual report. He opens on market structure. The globally popular Momentum Pair Trade - long the cohort of most positive momentum stocks while mercilessly shorting the most negative momentum stocks - has been the winning trade for the multi-strategy or 'pod shop' hedge fund model, which has been vacuuming up the assets earmarked for alternative baskets. He reproduces the chart he used in the July 8 Trading Post showing how extreme that crowded trade has become, and says it is still true and valid. His conclusion is that the other side of that trade is where the opportunity sits: he is vacuuming up cheap good businesses that are getting the short end of that stick, which he describes as, in a sense, shorting the pod shops by going long deep value.
Positioning follows directly from that. He added to his Tencent (0700 HK) long at HK$448.60, pointing back to his February 26 post Hong Kong Stocks: Structure and Strategy, which introduced Tencent as an all-time great compounder and argued Hong Kong stocks are in the dumps and have been for some time. He added to Meituan (3690 HK) at HK$87.60, a name he says he hid inside the May 25, 2026 Short Thoughts post - the same post in which he explains the volume indicators he uses, and which he applies to the Tencent chart shown here. In Australia he picked up a few more shares of Temple & Webster (TPW AU) at 5.00 AUD, with a full write-up on the company promised shortly.
The second half updates his July 8, 2026 Short Thoughts on NVDA, Neos, Hyperscalers, Jevons Paradox and Compression to reflect the BIS annual report released on June 28. He singles out Graph 13, 'Corporate credit is vulnerable to repricing on AI disappointments.' Its right-hand panel, Graph C, shows circular financing in the amount of many hundreds of billions of dollars accounting for the majority of both future spending and future revenue of the AI labs, hyperscalers and chip makers - what he calls the entire AI trifecta.
He then turns that against Nvidia's own defense. Back in November 2025 Nvidia rebutted circular-financing accusations by saying the strategic investments it made into its customers' businesses 'represent a minimum portion of revenue' already booked. Burry flags 'already booked' as the tricky part: it is a statement about the past. What the BIS shows on a forward basis, using announced multi-year commitments, is that circular financing is nearly 100% of chip makers' revenue - implicating a large part of Jensen Huang's GTC claim of 'visibility' to $1 trillion - and roughly two-thirds of the revenues and spending of the rest of the trifecta.
He closes by calling the coordinated effort to create the AI bubble stunning, and directs readers back to The Heretic's Guide to AI's Stars Part III: Tracepalooza & the Bezzle (May 22), the source of the two visuals he reuses here, which covers AI demand, captive offshore insurer financing, and the concept of compression. He highly recommends reading Part III in full. The report states no new short trades; the AI short book is context, while the actionable trades on the day are the three deep-value longs.
Key metrics
- Tencent (0700 HK) added at HK$448.60 per share
- Meituan (3690 HK) added at HK$87.60 per share
- Temple & Webster (TPW AU) bought at 5.00 AUD per share
- BIS annual report released June 28, 2026; Graph 13: 'Corporate credit is vulnerable to repricing on AI disappointments'
- Circular financing running to many hundreds of billions of dollars per BIS Graph C
- Circular financing is nearly 100% of chip makers' future revenue on announced multi-year commitments
- Circular financing approximates 2/3 of the revenues and spending of the rest of the AI trifecta (labs, hyperscalers)
- Jensen Huang's GTC claim of 'visibility' to $1 trillion
- Nvidia's November 2025 defense: strategic investments in customers 'represent a minimum portion of revenue' already booked
- Momentum Pair Trade crowding chart carried over from the July 8, 2026 Trading Post - 'still true and valid'
Key insights
- The Momentum Pair Trade is the crowded engine of the pod-shop multi-strategy complex; Burry's deep-value buying is an explicit, structural bet against that positioning rather than a stock-by-stock view.
- Hong Kong is his preferred venue for that bet - he added to both Tencent (HK$448.60) and Meituan (HK$87.60) on the same day, extending the thesis from his February 26 Hong Kong Stocks: Structure and Strategy post.
- Meituan was deliberately buried in the May 25, 2026 Short Thoughts post rather than announced - a reminder that his Short Thoughts pieces contain unflagged position disclosures.
- The BIS annual report's Graph 13 is his key new evidence: circular financing accounts for the majority of both future spending and future revenue across AI labs, hyperscalers and chip makers.
- Nvidia's November 2025 defense that strategic investments are a 'minimum portion of revenue' refers only to revenue already booked; the BIS forward view puts circular financing at nearly 100% of chip makers' revenue.
- Roughly two-thirds of the revenue and spending of the AI labs and hyperscalers is circular on BIS numbers, which is what makes corporate credit vulnerable to repricing on the first AI disappointment.
- Jensen Huang's $1 trillion 'visibility' claim at GTC is, on this analysis, substantially a restatement of announced multi-year circular commitments rather than independent end demand.
- Temple & Webster (TPW AU) at 5.00 AUD is an accumulating position ahead of a forthcoming write-up - a signal of a name he intends to size up.
Trading Post July 17, 2026: Plus Netflix & What's Up with the VIX
The most crowded trade on record is unwinding faster than it ever has, and the VIX is not showing it because the violence is inside the index rather than at the index level. The Morgan Stanley TMT momentum pair peaked June 22 at twice its October 2002 record, but where 2002 was earned on the short side after the tech bust, today it is built on the long side, so the reversal is a transfer of capital from the era winners into the names momentum had driven to unfathomable depths. That is exactly the cohort Burry owns, which is why PayPal, lululemon, Fiserv and Sprouts have been fuel for the short leg of the pair and are now the beneficiaries. Suppressed index volatility is a gift on the other side of the book, letting him add long-dated NVIDIA puts for pocket change while trimming an Oracle put position that got too large. Netflix, meanwhile, fails his evergreen test and is a Chapel, not a Castle, so even after an earnings sell-off it is Just Outside rather than a Fat Pitch.
aineocloudaisemiconsumerhealthconsumermediaconsumershoppingconsumertravelfinancefintech
Positions & performance
+0.8%avg return67% win · 3 tracked · L -26.8% · S +14.6%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SNVDA:NASDAQ aisemi NVIDIA ADDDec 2026 puts struck in the low $100s | NASDAQS&PBitget | ISPTC | 202.81closeJul 17 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +23.3% |
| WhyHe bought more December 2026 puts struck in the low $100s for what he calls pocket change because NVIDIA remains expensive with significant margin and earnings risk over the next five years, and with stock volatility minimal and the VIX holding in the teens the optionality is unusually cheap. |
| SORCL:NYSE aineocloud Oracle TRIMJan 2027 puts struck in the low-mid $100s · half | S&PBitget | ISAPTC | 126.41closeJul 17 '26 | 119.02 ● | 109.21250.25 | +5.8%APY +18.4% | +5.8% |
| WhyHe sold half the January 2027 puts struck in the low-mid $100s because the position had become too big after working well, and he does not fall in love with options trades that go his way. He retained the other half to maintain short exposure. |
| LDKNG:NASDAQ consumertravel DraftKings ADDcommon | NASDAQBitget | IAT | 25.0his fillJul 17 '26 | 18.29 ● | 16.5536.98 | -32.7%APY -66.5% | -26.8% |
| WhyAdded in the $25s with no change to the basic thesis he laid out in the July 8 Trading Post, where he sketched the valuation and plotted DraftKings and Flutter on the All Map; he intends to write both up in more detail. |
Full summary 593 words
This is a Trading Post, the position-by-position trade log, with a Netflix valuation note attached and a longer market-structure section on what the VIX is failing to measure.
Netflix is the setup. The stock closed at $74 after peaking at $134.15 on June 30, 2025, and traded down dramatically after reporting earnings the prior evening. He flags the unusual content accounting, including three separate changes to how Netflix amortizes production expense, then applies his evergreen test instead: does the company produce content that lasts, watchable on repeat across generations? Disney and Pixar do, and Warner Brothers does with Harry Potter and a few others. Netflix, he argues, does not. His example is Suits, a 2011 show that became a 2023 streaming hit whose economics were captured by NBCUniversal rather than Netflix. His line is that Disney produces wine and Netflix produces milk. So Netflix is Chapel rather than Castle, and on his Chapel numbers $67.75 is about 1.6 times intrinsic value, which makes the stock Just Outside with the Fat Pitch in the low $50s. He points readers to his earlier piece D AI of the Triffids: Office Software Triage for the methodology.
The trades are three. He bought more December 2026 NVIDIA puts struck in the low $100s for pocket change, reasoning that NVIDIA remains expensive with significant margin and earnings risk over the next five years, and that with stock volatility minimal and the VIX holding in, the position could be bumped up cheaply. He sold half of his January 2027 Oracle puts struck in the low-mid $100s because the position had become too big after working, and he does not fall in love with options trades that go well; he kept half to maintain short exposure. And he added to DraftKings in the $25s with no change to the basic thesis, referring back to the July 8 Trading Post where he described the thesis and valuation and plotted DraftKings and Flutter on the All Map.
The closing section explains why none of this looks like stress on the surface. The VIX sits at 17.87 and appears calm, but VIXEQ, which measures expected volatility of S&P 500 constituents, has been running to highs, putting the VIX/VIXEQ ratio at data-set lows. The index is hiding more internal volatility than at any point since 2014, the signature of rotation. The Morgan Stanley TMT momentum pair, long the era winners and short its losers, peaked on June 22 at a level of 500 against the prior record of 246 from October 2002; the series spans the dot-com bust, 9/11, the GFC, Volmageddon and COVID. Crucially the 2002 record was earned on the short side, when former tech winners had fallen so far they became the momentum losers, while today the record is built on the long side.
Since June 22 the pair has fallen about 40% in eighteen sessions, the fastest unwind in nearly three decades, decomposing into 18 points of longs falling and 29 points of shorts rallying. In every prior unwind of even 25% the VIX traded between 23 and 28; in this one it never left the teens. The names doing the rallying on the short leg are the interminably declining stocks he writes about and owns, PayPal, lululemon, Fiserv and Sprouts among them. That is how the positioning follows: the rotation pays his cheap longs while suppressed index volatility funds long-dated bearish optionality on the AI winners. The view breaks if volatility normalizes upward and takes away the cheap puts, or if the momentum long leg simply resumes its climb.
Key metrics
- Netflix closed at $74 after peaking at $134.15 on June 30, 2025
- On his Chapel numbers $67.75 is about 1.6 times Netflix intrinsic value, putting the Fat Pitch in the low $50s and the stock Just Outside
- Netflix has changed the amortization of production expense three times
- Added to DraftKings in the $25s
- Bought more NVDA December 2026 puts struck in the low $100s for pocket change
- Sold half of the ORCL January 2027 puts struck in the low-mid $100s, retaining half
- VIX at 17.87 while VIXEQ, the measure of future volatility of S&P 500 constituents, runs to highs; the VIX/VIXEQ ratio is at data-set lows, the most hidden index volatility since 2014
- The Morgan Stanley TMT momentum pair peaked June 22, 2026 at a level of 500 versus the prior record of 246 set in October 2002
- The momentum pair has fallen about 40% in eighteen sessions, the fastest unwind in the series nearly three-decade history
- The unwind decomposes into 18 points of longs falling and 29 points of shorts rallying
- In every prior unwind of even 25% the VIX traded between 23 and 28; through this one it never left the teens
Key insights
- The VIX at 17.87 is misleading: VIXEQ is at highs and the VIX/VIXEQ ratio is at data-set lows, meaning the index conceals more constituent-level volatility than at any point since 2014.
- The Morgan Stanley TMT momentum pair peaked June 22, 2026 at 500 against the prior record of 246 in October 2002, and this time the extreme was built on the long leg rather than the short leg, which makes the unwind a transfer into the most hated names rather than a broad drawdown.
- The pair has fallen about 40% in eighteen sessions, the fastest unwind in the series nearly thirty-year history, and it decomposes into 18 points of longs falling versus 29 points of shorts rallying, so the shorts covering did most of the damage.
- Prior momentum unwinds of even 25% ran with the VIX at 23-28; this one has never left the teens, which is precisely why long-dated puts remain cheap enough to add.
- He is using suppressed volatility to increase, not decrease, bearish AI exposure: more December 2026 NVIDIA puts struck in the low $100s bought for pocket change on margin and earnings risk he sees over the next five years.
- Risk management overrode a winning trade in Oracle: he cut the January 2027 puts in half purely because the position had grown too large, keeping half so the short thesis stays live.
- Netflix fails his evergreen test, so it is a Chapel rather than a Castle, and at $67.75 it trades at about 1.6 times intrinsic value with the Fat Pitch only in the low $50s despite the earnings crash from a $134.15 peak.
- The Suits example is the crux of the Netflix argument: a 2011 show became a 2023 streaming hit but the economics accrued to NBCUniversal, evidence Netflix rents attention rather than owning durable content value.
- DraftKings was added in the $25s with the thesis unchanged from the July 8 Trading Post, with detailed write-ups of DraftKings and Flutter still to come.
#013
Short Thoughts
Jul 15, 2026source ↗
Short Thoughts July 15, 2026: PYPL / IBM / HCA & More
Two mispricings landed on the same day and both are the market misreading cyclicality and control value as impairment. The Stripe/Advent $60.50 bid for PayPal is 1.21x an IV15 that contains no control premium, so it is an opening bid rather than a clearing price when his own true intrinsic range is IV8-IV10, or $75 to $115. IBM fell 26% on a preannouncement driven by IBM Z, historically its most volatile segment, which swung from +15% to -7% growth, yet the market erased $67 billion of value over a $700 million shortfall in a segment that is 8% of revenue. That reaction is existential fear that AI cloud buildout and tools like Anthropic COBOL modernization will finally break the mainframe annuity, and it is testable against a specific slate of July and August earnings. He will not chase IBM at $218.29 against a $120 IV15, but he will keep buying the cheap, momentum-shorted operating businesses he already owns.
aineocloudconsumerhealthconsumershoppingcybersecurityfinancefintechhealthmag5saas
Positions & performance
+14.6%avg return80% win · 5 tracked · L +14.6%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LHCA:NYSE health HCA Healthcare ADDcommon | S&P | IAPT | 362.0his fillJul 15 '26 | 511.78 ● | 353.99556.52 | +41.4%APY +174.8% | +41.4% |
| WhyHCA fell $28 to $362 on an earnings preannouncement he considers routine news-cycle noise with revenue and earnings in the same zone as before. He calls HCA a ruthlessly efficient provider and serial acquirer earning remarkable returns on capital, and notes that buying pullbacks to 10-12x owners earnings has always worked. |
| LFISV:NASDAQ financefintech Fiserv ADDcommon | NASDAQS&PBitget | ISAPT | 49.75his fillJul 15 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +25.7% |
| WhyAdded alongside HCA, SFM and ZTS at target buy prices set by his own valuation work rather than by any growth-versus-value label; Fiserv is one of the interminably declining payments names he views as cheap quality. |
| LPYPL:NASDAQ financefintech PayPal Holdings HOLDcommon | NASDAQS&PBitget | ISPT | 47.37his fillJul 15 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | +23.9% |
| WhyStripe and Advent bid $60.50 versus the $47.37 close, only 1.21x his IV15, and IV15 already excludes any control premium. His IV10 is $75-$80 and IV8 is $110-$115, so a proper control bid should be near $100 or above; he says he is not selling and expects the bid to rise. |
| LZTS:NYSE health Zoetis ADDcommon | S&P | IAPTC | 74.0his fillJul 15 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +14.6% |
| WhyFourth of the four adds on the day, bought at a preset target price after patience waiting for the Fat Pitch; plotted with the other holdings on his All Map. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADDcommon | NASDAQ | IAPT | 75.0his fillJul 15 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -32.7% |
| WhyAdded in the same Trading Postette batch of four buys; SFM is one of the beaten-down names he owns that the momentum machine has driven down, and it sits in his Fat Pitch buy zone on the All Map. |
Full summary 600 words
A multi-topic note on three events of July 15, 2026: a private-equity bid for PayPal, IBM's largest one-day fall ever, and an HCA preannouncement, plus a Trading Postette and a closing essay on Ben Graham.
On PayPal, Stripe and Advent offered $60.50 per share against a $47.37 close. Burry calls it too low. The bid is 1.21x his IV15, and IV15 is what a minority investor would pay, with no control premium embedded. His true intrinsic value runs at IV8 to IV10, where IV10 is $75-$80 and IV8 is $110-$115, so a control bid should land near $100. With control of the cash flows and the businesses come many levers to raise value, so 21% over market is not nearly enough. He calls PayPal one of the cheapest quality businesses in the portfolio and is not selling.
IBM went from $312 on July 7 to $213 on the morning of July 14, down 26% at the lows and the worst day since at least 1968. He walks the history: a 50x-earnings Go-Go leader that unbundled software in 1968 to head off a 13-year DOJ case dropped as without merit in 1982; a 41.5% crash; a 58% decline in 1973-74 with no single day worse than 10%; 23% on Black Monday 1987; and 13.1% in February 2026 when Anthropic released a COBOL modernization tool.
The cause of the plunge is IBM Z. Software is $30 billion growing near 10% at 84% gross margin, consulting $21 billion at a 12% segment margin, and IBM Z is customer-owned mainframe hardware at 59% gross margin running 70% of the world transactional workflows. Clients diverted Z spend into commodity servers, storage and memory now forecast to be in shortage, and were distracted by the Mythos cybersecurity scramble. Z has always been the most volatile segment, swinging from 15% growth in Q1 to -7% in Q2, yet $67 billion of market cap vanished over a $700 million shortfall plus 5% software growth guidance, erasing 25% of the market cap of a company where Z is 8% of revenue. His falsification test is earnings season: SAP and IBM on July 22, Wipro July 17, Infosys and SK Hynix July 23, ServiceNow and Microsoft near July 29, Cognizant and Capgemini into early August, with Accenture the cleanest comp, in September. If nobody corroborates the diversion story, IBM becomes a show-me story at a single-digit multiple, roughly 50% below current.
He still did not buy IBM. A delta-E of 0.87 from stock comp at 2.6% of revenue cuts $10.8 billion of GAAP income to about $9.3 billion of owners earnings at roughly 12% ROIC and 21-22x, against an IV15 of $120, the midpoint of Castle $126 and Chapel $115, versus a $218.29 close. It is high Chapel rather than Castle: consulting and software face frontier models, as the Starbucks replacement story showed, while the COBOL moat and switching costs remain formidable and pension underfunding is only $2.3 billion. He contrasts IBM booking labor in cost of revenue with Palantir routing Forward Deployed Engineers into R&D and sales for an 80%-plus optical gross margin.
HCA, down $28 to $362 on its own preannouncement, he dismisses as the usual news cycle that periodically puts a ruthlessly efficient serial acquirer on sale; pullbacks to 10-12x owners earnings have always been buyable. He then added to four positions: HCA at $362, Sprouts at $75, Fiserv at $49.75 and Zoetis at $74. It closes with Graham recommending Computing-Tabulating-Recording in 1916 and being told its book value was water; his own process sets target buy prices regardless of label, then waits for the Fat Pitch.
Key metrics
- Stripe/Advent bid $60.50 per share for PayPal versus a $47.37 close, a 21% premium
- Bid is 1.21x his IV15 for PYPL; IV10 is $75-$80 and IV8 is $110-$115
- IBM $312 on July 7 to $213 on the morning of July 14; down 26% at the lows, the worst day since at least 1968
- IBM prior big drops: 41.5% crash after the Go-Go era, 58% in 1973-74, 23% on Black Monday 1987, 13.1% in February 2026 on the Anthropic COBOL news
- IBM software is a $30 billion business growing around 10% at 84% gross margin; consulting is $21 billion at 28% gross margin and only a 12% segment margin
- IBM Z carries 59% gross margins and runs 70% of the world transactional workflows; Z grew 15% in Q1 and -7% in Q2
- About $67 billion of market cap lost on a $700 million IBM Z revenue shortfall and 5% software growth; Z is 8% of revenue but 25% of market cap disappeared
- IBM delta-E of 0.87 on SBC of 2.6% of revenue, cutting $10.8 billion GAAP income to about $9.3 billion owners earnings; ROIC about 12%
- IBM IV15 of $120 (midpoint of Castle $126 and Chapel $115) versus a $218.29 close, roughly 21-22x owners earnings; net pension underfunding only $2.3 billion
- IBM gross margin fell 56% in 1990 to 36% in 1999, then rose from about 47% pre-COVID to 60% recently
- HCA down $28 to $362 on its preannouncement; pullbacks to 10-12x owners earnings have historically been buyable
- Adds on the day: HCA at $362, SFM at $75, FISV at $49.75, ZTS at $74
Key insights
- A takeover bid is not a valuation ceiling: at $60.50 versus a $47.37 close the PayPal offer is 1.21x IV15, and since IV15 embeds no control premium, a control transaction should clear near $100 against an IV8 of $110-$115.
- IBM lost about $67 billion of market cap over a $700 million shortfall in IBM Z, a segment that is only 8% of revenue but accounted for 25% of the market cap destroyed.
- IBM Z has always been the most cyclical segment, swinging from 15% growth in Q1 to -7% in Q2, so the preannouncement was probably not even a top-10 IBM preannouncement historically.
- The IBM thesis is externally testable in weeks: SAP on July 22, Wipro July 17, Infosys and SK Hynix July 23, ServiceNow and Microsoft near July 29, with Accenture the cleanest comp in September; corroboration of the spend-diversion story makes the sell-off noise, silence makes IBM a single-digit-PE show-me story roughly 50% lower.
- Even after a 26% crash IBM is not cheap on his math: about 21-22x owners earnings and $218.29 against an IV15 of $120, which is why he explicitly did not buy.
- Stock comp matters more than headlines suggest: IBM has been a pure diluter since suspending buybacks after the 2019 Red Hat deal, and a delta-E of 0.87 cuts $10.8 billion GAAP income to roughly $9.3 billion of owners earnings.
- IBM books consulting labor in cost of revenue for a 28% gross margin while Palantir routes Forward Deployed Engineers into R&D and sales for an 80%-plus optical gross margin, the accounting difference Burry says invalidates the software-company claim.
- HCA pullbacks to 10-12x owners earnings have always been buyable, and the business is largely insulated from macro, which is why a $28 preannouncement drop to $362 was an add rather than a warning.
- The genuine tail risk in IBM is not the Z cycle but an AI-driven COBOL offramp attacking a 60-year-old transaction-processing annuity, a scenario he finds hard to imagine but calls truly scary.
Short & Thankful: 300
A milestone note rather than an investment piece. Cassandra Unchained crossed 300,000 subscribers 231 days after launch, and Burry uses the occasion to thank the community rather than to argue a market view. The notable substance is the reach and composition of the audience: 346,680 followers, subscribers across all 50 states and 212 countries, and a majority - 52% - based outside the United States. He points out the growth chart shows no parabola, and says he finds that comforting.
Positions & performance
No single-stock positions stated in this report.
Full summary 268 words
This is a short community post, not research. On Sunday, July 12, 2026, less than eight months into Cassandra Unchained - 231 days, to be exact - Burry marks the moment the publication crossed 300,000 subscribers. He breaks the numbers down: 346,680 followers and 300,044 subscribers.
He describes the venture as new and intense, and says he is stunned and humbled by the community that has formed around it, singling out the Chats as lively and educational at any hour and the Discussions under every post as erudite and challenging. On the growth chart itself he makes a brief but characteristic observation: there is no parabola, and that is okay, and even comforting - a nod to the parabolic price action he spends most of his other posts warning about.
The audience composition is the one genuinely informative datapoint. Subscribers come from all 50 states and 212 countries, and 52% of them are based outside the United States. Burry says this does not surprise him, because when readers write to him it is rarely someone in the US. He closes by noting it is the middle of summer in the States, hot and beautiful, wishing readers a good Sunday, and expressing gratitude for the mutual respect he sees in every Chat and Discussion. The post also carries the standard Cassandra Unchained description: that the publication is now Dr. Michael Burry's sole focus, giving readers a front row seat to his analytical efforts and projections for stocks, markets and bubbles, often with an eye to history and its timeless patterns. No positions, tickers, prices or market views are stated.
Key metrics
- 300,044 subscribers as of Sunday, July 12, 2026
- 346,680 followers
- 231 days since Cassandra Unchained launched, just under eight months
- Subscribers in all 50 states and 212 countries
- 52% of subscribers are from outside the United States
Key insights
- Cassandra Unchained reached 300,044 subscribers in 231 days, under eight months from launch, alongside 346,680 followers.
- The audience is majority non-US: 52% of subscribers sit outside the United States, spread across 212 countries and all 50 states.
- Burry notes the subscriber growth chart shows no parabola and calls that comforting - a deliberate echo of the parabolic-blowoff language he uses in his market work.
- The publication is described as Burry's sole focus, positioning Cassandra Unchained as the primary vehicle for his analytical output rather than a side project.
- No investment content: the post states no positions, prices or tickers, and should be treated as a community milestone rather than a research signal.
#011
Housing / GSEabridged
Jul 12, 2026source ↗
[Abridged Version] Fannie, Freddie, Trump & the Housing Crisis: A CUSIP by CUSIP Examination of the Health of the Toxic Twins' and Their Most Risky Business
Fannie and Freddie common is a bet on one Treasury decision. The Senior Preferred Stock liquidation preference is soaking up all the retained earnings, so the government gets no real cash today and shareholders get no claim on the capital being built. Deem the SPS repaid and the equity is worth 3-4x now and 6-7x in time even after warrant dilution; affirm it and the stock goes to the low single digits. In the meantime Trump's push for more mortgage credit is driving the Twins to expand aggressively into a weakening multifamily market, while their own filings show rising write-offs, systemic origination fraud and a repurchase-enforcement function that recovers 38 cents on the dollar at Fannie and 3 cents at Freddie. That decay is the argument for acting now: uplist and clarify the SPS so private capital and private managers can take over before the credit cycle makes the exit story harder to tell.
financefintech
Positions & performance
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFNMA Fannie Mae REITcommon | NASDAQ | — | —Jul 12 '26 | — | — | — | n/a |
| WhyBurry stays constructive because the Treasury is currently receiving no meaningful cash from the Twins - it all accretes to the SPS liquidation preference while the capital stays inside the enterprises. If the SPS is deemed repaid he sees the shares rising 3-4x initially and 6-7x in time, assuming warrant exercise and dilution, and he expects action before the warrants expire in September 2028, just ahead of the next presidential election. |
| LFMCC Freddie Mac REITcommon | NASDAQ | — | —Jul 12 '26 | — | — | — | n/a |
| WhySame SPS-driven upside case, with the added near-term catalyst of the Executive Order 14393 report due around the July 11 weekend, which bulls hope recommends capping the Enterprise Regulatory Capital Framework at 2.5%. Freddie's specific blemish is enforcement: it has forced repurchase on only 3% of loans submitted for fraud versus Fannie's 38%, which Burry treats as evidence for privatization rather than a reason to sell. |
Also discussedWD:NYSE financefintech
Full summary 605 words
This is the free, abridged version of Burry's CUSIP-level examination of Fannie Mae (FNMA) and Freddie Mac (FMCC), published two days after the paid piece. He begins by correcting a widespread misconception: the Twins are not currently sending profits back to the US government in any meaningful cash sense. Everything accrues to the Senior Preferred Stock liquidation preference while the capital stays inside the enterprises. That makes the equity a bet on one administrative decision. If Treasury affirms the SPS LP, the shares fall to the low single digits or lower. If it is deemed repaid, Burry sees them rising 3-4x initially and 6-7x in time, explicitly assuming the Treasury warrants are exercised and the dilution taken. Those warrants expire in September 2028, just before the next presidential election, and he expects action before then as Trump works through donors and unfinished business in a lame-duck year.
The nearest catalyst is Executive Order 14393, Promoting Access to Mortgage Credit, signed March 13, 2026, which gave the FHFA director 120 days to report on the efficiency of national housing finance markets - due right around the July 11 weekend. Bulls hope it recommends capping the Enterprise Regulatory Capital Framework at 2.5%, a dramatic cut in required capital. Burry damps expectations: there is no indication ERCF clarity arrives with the report, it may never be public, and it may already sit with the Administration. The SPS decision rests with Treasury and the President.
The body compresses his credit work. Trump's November 24, 2025 FHFA order raised 2026 combined multifamily purchase caps to $176 billion. Against that, he argues the housing shortage is a household-formation problem, not a building problem: a record 1 million apartments were under construction in July 2023 versus 901,000 in 1973, the pipeline never broke 500,000 units from 1975 to 2015, and the US holds a record 435 units per 1,000 people and 14 million more units than households. Persons per household fell from 3.39 to 2.53, so the marginal renter is a single earner with thinner income - a fragile cash-flow buffer beneath multifamily debt. Were one fifth of solo dwellers to pair up, he says every mainstream estimate would show the shortage eliminated.
The evidence comes from the mid-May ABS-15G data dump, the Twins' 10-Qs and Freddie's April 30 performance reports on six MBS programs. Freddie's multifamily volume rose 24% to $12.8 billion in 1Q 2026; Fannie's average loan went to $26 million from $18 million with full interest-only at 68% from 54% and LTV at 63% from 61%. Watchlist ratios run 13.42% to 23.58%. Multifamily serious delinquency is up 48% since December 2023, write-offs hit $243 million versus $61 million, liquidation losses 18.8% versus 10.4%, and provisions $277 million versus $24 million. Walker & Dunlop's 2025 10-K calls commercial real estate fraud systemic and no longer anecdotal, directly contradicting Fannie's claim, reported by The Real Deal on February 2, that its 193 fraud investigations in 2024 are now trending down. The two findings that matter: Fannie resolves 38% of repurchase demands against Freddie's 3%, showing no prioritisation of chasing fraud, and Freddie carries $15.5 million across 44 loans against an entity it lists as Unavailable.
Positioning is unchanged and constructive. The surfacing multifamily losses are not material next to roughly $29 billion of combined earnings, but their presence and trend change the urgency. An uplisting and SPS clarification require no conservatorship termination, and he closes by asking Trump to uplist and deem the SPS repaid before housing stress becomes the story. The thesis breaks if the SPS is affirmed, or if credit deterioration accelerates far enough to consume the earnings cushion.
Key metrics
- If the SPS liquidation preference is deemed repaid, Burry sees the shares rising 3-4x from current levels initially and 6-7x in time, assuming warrant exercise and dilution; if affirmed, they fall to the low single digits or lower
- The Treasury warrants expire in September 2028, just before the next presidential election, and Burry believes there will be action before they lapse
- Executive Order 14393, Promoting Access to Mortgage Credit, was signed March 13, 2026 and gave the FHFA director 120 days to report - due around the July 11 weekend; bulls expect a recommendation to cap the ERCF at 2.5%
- Trump's November 24, 2025 FHFA order raised 2026 combined multifamily loan-purchase caps to $176 billion
- Freddie Mac's multifamily business rose 24% year-over-year to $12.8 billion in 1Q 2026
- Fannie Mae's average new loan jumped to $26 million from $18 million, full interest-only share to 68% from 54%, and original LTV to 63% from 61% - all signs of easing credit standards
- Multifamily serious delinquency is up 48% since December 2023; Fannie's April 29 10-Q showed multifamily write-offs of $243 million versus $61 million, liquidation losses of 18.8% versus 10.4%, and a $277 million credit provision versus $24 million
- Freddie's six CMBS programs carry watchlist ratios from 13.42% (K-Deal) to 23.58% (ML-Deal affordable); a DSCR below 1.10 fixed-rate or 1.00 floating lands a loan on the servicer watchlist
- Fannie forces successful repurchase on 38% of loans submitted versus only 3% at Freddie; one Freddie ABS-15G/A shows $15.5 million across 44 loans owed by an entity listed as identity Unavailable
- US housing stock is a record 435 units per 1,000 people with 14 million more units than households, while persons per household fell from 3.39 to 2.53
- Fannie said multifamily fraud tips prompted 193 investigations in 2024, up from 14 in 2022, and claims the numbers are now trending down - contradicted by Walker & Dunlop's 2025 10-K
- Combined GSE earnings of roughly $29 billion dwarf the multifamily losses now surfacing
Key insights
- The single most-corrected misconception in the piece: Treasury receives no significant cash from the Twins today. Every dollar of retained earnings accretes to the SPS liquidation preference while the capital physically stays with Fannie and Freddie.
- Burry's upside math is explicit and dilution-adjusted - 3-4x initially and 6-7x in time if the SPS is deemed repaid, assuming the warrants are exercised. The downside if it is affirmed is the low single digits or lower.
- The near-term catalyst is the Executive Order 14393 report, due around July 11, 2026 under a 120-day clock from the March 13 signing. Bulls want an ERCF cap at 2.5%, but Burry warns the report may not be public and may not address the ERCF at all.
- Timing is anchored to the September 2028 warrant expiry landing just before the next presidential election, which Burry expects will pull the GSEs back into Trump's focus during his lame-duck year.
- The Twins responded to Trump's $176 billion cap order by easing structure, not price: Fannie's full interest-only share rose to 68% from 54% and average loan size to $26 million from $18 million, expanding into a market where multifamily serious delinquency is already up 48% since December 2023.
- Walker & Dunlop's 10-K language - that CRE fraud risk is systemic and no longer anecdotal - is a public-company filing contradicting Fannie's own February assurance that fraud tips are trending down.
- The enforcement gap is the abridged piece's headline finding: Fannie forces repurchase on 38% of submitted loans while Freddie manages 3%, evidence that neither institution prioritises chasing fraud - and a reason to assume similar slack in less urgent functions.
- Burry's ask requires no conservatorship termination and no IPO. Uplisting and an SPS announcement are administrative acts, so the January 2025 side letter's procedural sequence is not the binding constraint on the catalysts that matter.
Washington Goes To Fannie Mae & Freddie Mac: The Ugly, the Bad & the Good that Investors Need to Know
The entire value of Fannie and Freddie common equity hinges on one administrative decision: whether Treasury affirms the Senior Preferred Stock liquidation preference, which grows in line with retained earnings, or deems it repaid. Affirmation sends the stocks to the low single digits; repayment hands shareholders the retained capital and the shares run 3-4x. Meanwhile Trump's FHFA order to expand multifamily lending is pushing the Twins to ease credit standards into a deteriorating multifamily market, where a supply overshoot and a fragmenting, single-earner household base are eroding rents and coverage. Burry's forensic read of the May 2026 ABS-15G filings, Fannie's 10-Q and Walker & Dunlop's 10-K shows real fraud, rising write-offs and a fraud-enforcement function that barely functions - three cockroaches, and there are never just three. But none of it is material against roughly $29 billion of combined earnings, so the bear evidence actually strengthens the bull case: it argues for urgency, for uplisting, for cancelling the SPS and for putting the Twins back under private management before the credit cycle makes the exit story harder to tell.
financefintech
Positions & performance
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LFNMA Fannie Mae REITcommon | NASDAQ | — | —Jul 10 '26 | — | — | — | n/a |
| WhyDespite a first-quarter jump in multifamily write-offs to $243 million from $61 million and a credit provision of $277 million versus $24 million, Burry judges the deterioration immaterial against roughly $29 billion of combined GSE earnings, so the capital-building bull thesis proceeds. The value driver remains the Senior Preferred Stock liquidation preference: if it is deemed repaid the shares run 3-4x, and the only catalysts needed are an uplisting and an SPS announcement, both administrative acts available before the September 2028 warrant expiry. |
| LFMCC Freddie Mac REITcommon | NASDAQ | — | —Jul 10 '26 | — | — | — | n/a |
| WhySame SPS-driven setup as Fannie, but with a sharper governance discount: Freddie has forced repurchase on only 3% of loans it submitted for fraud versus Fannie's 38%, and its ML-Deal affordable program carries a 23.58% servicer watchlist ratio at a 1.15x average DSCR. Burry reads that institutional slack as an argument FOR privatization rather than against the equity, since the single-family guarantee franchise that drives earnings remains structurally sound. |
Also discussedWD:NYSE financefintech
Full summary 624 words
This is the third instalment of Burry's GSE work, following December 8, 2025's Fannie & Freddie, Toxic Twins No More and March 26, 2026's The Toxic Twins Recurrence, prompted by a mid-May 2026 data dump. He strips the story to one variable. Investors obsess over an IPO or an exchange listing, but the dominant driver of the common's value is the Senior Preferred Stock liquidation preference, which grows in line with retained earnings. Today the Twins keep their earnings and the enterprises get stronger, yet common holders have no claim on that capital. If Treasury affirms the SPS LP the shares fall straight to the low single digits; if it is deemed paid off the stocks run perhaps 3-4 times today's price. He expects SPS clarification to precede any IPO, since the IPO cannot be marketed without it.
The macro spine is a contrarian read on the housing shortage. Trump's November 24, 2025 FHFA order raised 2026 combined multifamily purchase caps to $176 billion from $146 billion and directed the GSEs to hit those targets even into a shrinking market. Burry argues the market is shrinking for good reason: over 1 million apartments were under construction in July 2023 versus 901,000 in 1973, against a pipeline that never broke 500,000 units from 1975 to 2015. The US now has a record 435 units per 1,000 people and 14 million more units than households. What America has is an income-per-unit shortage, because persons per household fell from 3.39 to 2.53 and solo living drove about half of net household growth since 1960. The marginal renter is a single earner with fragile income - exactly the tenant base behind fixed-rate multifamily loans.
The evidence is forensic. Freddie's 1Q 2026 multifamily volume rose 24% to $12.8 billion with the pipeline up 41%, but the mix moved away from the affordable mandate: targeted-affordable fell from $3.7 billion to $3.0 billion and small-balance collapsed from $1.3 billion to $0.3 billion. Fannie eased harder - average loan size $26 million from $18 million, full interest-only 68% from 54%, LTV 63% from 61%. Freddie's published table shows CMBS watchlist ratios of 13.42% to 23.58% with essentially zero realized losses, which Burry tests against the 526-unit Solaris complex in Austin: a $50 million March 2024 loan now re-auctioning below $10 million. Walker & Dunlop's 2025 10-K discloses a $221.6 million repurchase requirement from borrower fraud including concealed flip transactions, and flatly contradicts Fannie's February claim, reported by The Real Deal, that fraud tips are trending down. Fannie's April 29 10-Q shows write-offs of $243 million versus $61 million and liquidation losses of 18.8% versus 10.4%.
The ABS-15G analysis yields two findings. Against covered pools of $2.235 trillion at Fannie and $669 billion at Freddie, demand rates are under a tenth of a percent and originator rejections are literally zero - every demand is deemed legitimate. But Fannie resolves only 37.9% and Freddie just 3%, with homebuilder captives at 45-67% versus non-banks at 20-40%, and $15.5 million across 44 loans owed by an entity Freddie lists as Unavailable. To Burry this is institutional decay from 17 years outside market discipline, consistent with administrative expense down 25% year-over-year.
Positioning follows: he stays long both names. None of the deterioration is material against roughly $29 billion of combined earnings, and the single-family guarantee franchise remains sound, so the bear case makes the bull case - it raises the urgency of getting private capital and management in. All that is required is an uplisting and an SPS announcement, neither of which triggers the January 2025 side letter's procedural sequence, and the warrants expire September 2028. The view breaks if Treasury affirms the SPS liquidation preference, or if credit deterioration escalates enough to threaten the capital build.
Key metrics
- Trump's November 24, 2025 FHFA order raised 2026 combined multifamily loan-purchase caps to $176 billion from $146 billion, roughly a 20% increase, with the 50% mission-driven affordable minimum unchanged
- Freddie Mac multifamily volume rose 24% year-over-year to $12.8 billion in 1Q 2026 with the forward rate-lock pipeline up 41%, while targeted-affordable volume FELL from $3.7 billion to $3.0 billion and small-balance lending collapsed from $1.3 billion to $0.3 billion
- Fannie Mae's average new multifamily loan jumped to $26 million from $18 million, full interest-only share to 68% from 54%, and original LTV to 63% from 61%, with DSCR flat at 1.7x
- Fannie multifamily write-offs rose to $243 million from $61 million; liquidated loan losses 18.8% versus 10.4%; total credit provision $277 million versus $24 million in 1Q 2025
- Multifamily serious delinquency is up 48% since December 2023, and the new serious delinquencies are primarily FIXED-rate conventional loans, an income-loss signal rather than a rate-reset signal
- Freddie's six CMBS programs show watchlist ratios from 13.42% (K-Deal) to 23.58% (ML-Deal affordable, average DSCR 1.15x), yet near-zero delinquency and zero realized losses
- Walker & Dunlop's 2025 10-K disclosed a $221.6 million repurchase/indemnification requirement from fraudulent loans plus $40.9 million of related expense, with $194.6 million originated by employees no longer at the company
- Fannie's ABS-15G/A filings cover a $2.235 trillion portfolio including $1.13 trillion of single-family RMBS; Freddie discloses against $669 billion; repurchase demand rates for both are under a tenth of one percent of covered pool value
- Fannie's repurchase resolution rate is 37.9% versus Freddie's 3%; rejections by originators are zero for both; withdrawals total roughly $319 million or about 28% of combined demands
- Pulte Mortgage shows a $6.4 million demand from Fannie resolved 52.8% versus a $7.4 million demand from Freddie resolved 3.4%; homebuilder captives resolve at 45-67% while non-banks resolve at 20-40%
- Freddie discloses $15.5 million of repurchase demands across 44 loans against a counterparty whose identity is listed as Unavailable, its third-largest single entry after United Wholesale Mortgage and Rocket Mortgage
- US housing stock is at a record 435 units per 1,000 people (one unit per 2.3 people) with 14 million more units than households, while persons per household fell from 3.39 to 2.53
- Fannie's 1Q 2026 10-Q shows administrative expense down 25% and salary expense down 24% year-over-year at a $4.3 trillion-asset guarantor; combined GSE earnings were approximately $29 billion in 2024
- Single-family REO liquidated at 103% of unpaid principal in 1Q 2026 versus 146% a year earlier; Freddie's foreclosed 526-unit Solaris complex in Austin, financed with a $50 million loan in March 2024, is being re-auctioned at under $10 million, an 80%-plus implied loss
Key insights
- The SPS liquidation preference, not the IPO or an uplisting, is the binary that sets the equity's value: affirmation means low single digits, deemed repayment means roughly 3-4x. Burry expects clarification to come BEFORE any IPO because the deal cannot be marketed without it.
- Zero out of every repurchase demand across both GSEs was rejected by an originator as unfounded - the fraud is not disputed, only the timing and amount of payment. That makes the 37.9% (Fannie) and 3% (Freddie) resolution rates a statement about GSE effort, not about the merits.
- Freddie's own published table shows up to 23.58% of the ML-Deal affordable program on the servicer watchlist with roughly 0.1% delinquency and zero realized losses. The Austin Solaris property - a $50 million loan re-auctioning under $10 million - shows losses are coming that the table simply has not posted yet.
- The most telling detail in Fannie's disclosures is that new serious delinquencies are primarily FIXED-rate multifamily loans. Fixed-rate loans fail when borrower income falls, not when rates reset, which points at vacancy and rent weakness rather than a refinancing wall.
- The GSEs answered Trump's cap order by easing structure rather than coverage: Fannie's full interest-only share went to 68% from 54% and average loan size to $26 million from $18 million, while DSCR stayed pinned at 1.7x. Loans written today will not reveal their quality for a year or two.
- Walker & Dunlop's $221.6 million repurchase demand is entirely separate from its DUS first-loss risk-sharing, and Burry notes neither Bloomberg nor the Wall Street Journal covered it. It also directly contradicts Fannie's public line that multifamily fraud is trending down.
- Fannie cut administrative expense 25% and salary expense 24% year-over-year while running fraud detection on a $4.3 trillion book. Fewer people chasing fraud is a plausible mechanical explanation for Fannie's own claim that fraud tips fell from 193 investigations in 2024 to lower levels.
- The housing shortage is a household-formation artefact, not a construction shortfall: at 435 units per 1,000 people the US has 14 million more units than households, and if roughly one fifth of solo dwellers paired up the shortage would vanish on every mainstream estimate.
- Burry's conclusion inverts the usual short logic - the evidence of decay is his strongest argument for privatization. Two press releases (uplist, cancel the SPS) would unlock private capital and management without triggering the January 2025 side letter's RFI, FSOC briefing and Treasury proposal sequence.
Trading Post July 8, 2026: IV15 (and IV10, IV12, IV18, IV20..), Two New Buys, One Addition to a Position
Burry uses this Trading Post to restate that a low multiple is not the same thing as value, and that his buy discipline runs off IV15 - the price at which he expects 15% compounded annual returns over 15 years - derived from a multi-stage DCF on conservatively adjusted owners' earnings rather than a Graham-and-Dodd style ratio. Against that framework he opens two new positions in online gaming, Flutter at about $107 and DraftKings in the low $26s, sized 60/40 as a single full position. Both stocks have been hit by prediction markets, which structure event contracts as CFTC swaps and therefore trade in all 50 states while avoiding state gaming taxes and riding third-party infrastructure like Robinhood and Webull; Burry's core call is that this is a regulatory loophole adjacent to a heavily taxed industry and the political climate will not tolerate it, so the volume hit is temporary. Flutter is a scaled, high-return business whose debt load should force paydown and buybacks after past capital misallocation, while DraftKings is an inflecting operator where the value lies in the transition he expects shortly. He separately added to JD.com at $27.58, already a top-three position, on the view that the unwinding of the AI/memory mania in South Korea and Japan will benefit Hong Kong and Chinese equities.
airobotconsumertravelsaas
Positions & performance
-19.8%avg return0% win · 3 tracked · L -19.8%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LJD:NASDAQ airobot JD.com ADDcommon · Added to an existing top-three position | NASDAQBitget | IS | 27.58his fillJul 08 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | -0.6% |
| WhyAlready one of his top three positions; he bought more because he believes the unwinding of the AI/memory mania in South Korea and Japan will rotate money into and benefit Hong Kong and Chinese stocks. |
| LFLUT Flutter Entertainment NEWcommon · Roughly 60% of a full position (FLUT and DKNG together are one full position, about 60/40 Flutter/DraftKings) | — | AT | 107.0his fillJul 08 '26 | 79.15 ● | 72.56229.86 | -26.0%APY -56.6% | -26.0% |
| WhyWorld's largest online betting and gaming company, owner of FanDuel, the #1 U.S. sportsbook, plus Paddy Power and Betfair and a footprint across Europe, India and Brazil. Burry calls it a fundamentally very good operating business with terrific scale that has been hurt by past capital misallocation, and his valuation assumes the debt level now forces a focus on debt paydown and buybacks. Volume is swelling at these lows, which he likes to see, and the stock has been pressured by the prediction-market volume threat he expects regulation to close. |
| LDKNG:NASDAQ consumertravel DraftKings NEWcommon (Class A) · Roughly 40% of a full position (FLUT and DKNG together are one full position, about 60/40 Flutter/DraftKings) | NASDAQBitget | IAT | 27.17closeJul 08 '26 | 18.29 ● | 16.5536.98 | -32.7%APY -66.5% | -32.7% |
| WhyBought in the low $26s. U.S.-only online sports and iGaming with huge brand mindshare, a great core business with very high returns on investment, and an inflecting company where Burry says the value sits in the operating transition he foresees in the near future. The selloff came from prediction markets taking volume via a CFTC swaps loophole that avoids state gaming taxes, which he believes the political climate will not tolerate for long. |
Full summary 711 words
The July 8, 2026 Trading Post opens with methodology before it gets to trades. Burry restates the great tension in his valuation process: a low multiple is not necessarily a value. The Graham and Dodd roots are there, but his process since the 1990s has not hinged on a 1940 version of value. IV15 - the price at which he expects compounded annual returns of 15% over 15 years - is his benchmark buy price target, though a higher quality business can be a fat pitch above that mark and a lower quality business only well below it. IV15 is not a simple PE ratio. It comes from a discounted cash flow analysis that pulls from business quality and applies conservative adjustments to owners' earnings for stock-based compensation, bedeviled accounting and other factors. The DCF generally runs three stages, adds a fourth for inflecting growth companies, and adjusts for special cases including serial acquirers.
He points readers to where the method is laid out: D'AI of the Triffids: Office Software Triage (May 15) for the full treatment, and its ~1,500-word abridged version (May 17), which was the first post for which he wrote a shorter companion - he now writes abridged versions of everything, with the full versions carrying roughly 11,000 more words and additional visuals. He expanded the methodology again in Software and Payments Stocks Part II: Productivity Tools and Cybersecurity (May 28), where he added Synopsys (SNPS), Cadence Design Systems (CDNS), JFrog (FROG) and Samsara (IOT) to bring the Cassandra Unchained software and payments universe to 50 names, the SW50. He notes he sets alerts at IV10, IV12, IV15, IV18 and IV20 for every stock, that baseline intrinsic value typically sits between IV8 and IV10 depending on business quality, and that only below baseline intrinsic value are buybacks genuinely accretive to intrinsic value per share - above it, buybacks pull in shares while diluting intrinsic value per share, which he calls the depressing nuance of tech companies repurchasing stock at high prices to offset part of stock-based compensation. The All Map, a baseball-field visual, sorts companies into Fat Pitches, Just Outside and The Out Field by IV15 multiple, AICT Tier and Composite Score.
The trades: he bought Flutter Entertainment (FLUT) at about $107 and DraftKings Class A (DKNG) in the low $26s. Together the two are a full position, roughly 60/40 Flutter to DraftKings, and he may make each a full position later. Flutter is the world's largest online betting and gaming company, running FanDuel - the number one U.S. sportsbook - alongside Paddy Power and Betfair in the UK and operations across Europe, India and Brazil; it is an older business under a newer name that has only traded in the U.S. for two and a half years, so investors may not know it even though most Americans recognize FanDuel. He notes volume swelling at these lows on the chart, which he likes to see. DraftKings is U.S.-only online sports and iGaming built around a brand with huge mindshare.
The threat that created the entry point is prediction markets, which structure their products as swaps under the CFTC. As securities, event contracts are automatically available in all 50 states while avoiding state gaming taxes, and they carry no infrastructure of their own, riding on Robinhood, Webull and crypto.com. That volume hit has hurt both stocks. Burry believes the political climate will not tolerate a loophole adjacent to a heavily regulated and taxed industry, and that prediction markets will in time be subsumed into regulation and taxation; both Flutter and DraftKings have also launched their own prediction-market efforts. Both have expanded through acquisition; for Flutter his valuation assumes the focus is now debt paydown and buybacks, which the debt level should force, and he flags past capital misallocation against a fundamentally very good operating business with terrific scale. DraftKings he frames as inflecting, with the value in the operating transition he foresees in the near future. Both, he says, are great businesses at the core with very high returns on investment.
Finally, he bought more JD at $27.58, already one of his top three positions, on the view that the unwinding of the AI and memory mania in South Korea and Japan will benefit Hong Kong and Chinese stocks.
Key metrics
- Bought Flutter Entertainment (FLUT) at ~$107
- Bought DraftKings Class A (DKNG) in the low $26s
- FLUT and DKNG together are one full position, roughly 60/40 Flutter/DraftKings
- Bought more JD at $27.58 - one of his top three positions
- IV15 is the price at which he expects 15% compounded annual returns over 15 years, and is his benchmark buy price target
- Baseline intrinsic value typically sits between IV8 and IV10 depending on business quality; buybacks are only accretive below that level
- He sets alerts at IV10, IV12, IV15, IV18 and IV20 for every stock he analyzes
- The DCF behind IV15 runs three stages, adding a fourth for inflecting growth companies, with adjustments for serial acquirers and other special cases
- The Cassandra Unchained software and payments universe is now 50 stocks (the SW50) after adding SNPS, CDNS, FROG and IOT
- The software and payments work is a six-part series covering 46 stocks originally; abridged posts run ~1,500 words versus ~11,000 more words in the full versions
- The All Map splits companies into Fat Pitches, Just Outside and The Out Field by IV15 multiple, AICT Tier and Composite Score
Key insights
- IV15 is a buy price target, not a valuation - the price at which Burry expects 15% compounded annual returns over 15 years, built from a three-stage DCF (four stages for inflecting growth companies) on owners' earnings adjusted down for stock-based compensation and accounting distortion.
- Baseline intrinsic value sits between IV8 and IV10 depending on business quality, and buybacks are only accretive to intrinsic value per share below that level - above it, repurchases reduce shares while diluting intrinsic value, which he calls the depressing nuance of tech buybacks funded to offset SBC.
- He runs a laddered alert system at IV10, IV12, IV15, IV18 and IV20 on every name he analyzes, so entries are pre-priced rather than reactive.
- Flutter at ~$107 and DraftKings in the low $26s were bought as a single full position split roughly 60/40, with the option to size each up to a full position later.
- The entry thesis on both gaming names is regulatory mean reversion: prediction markets structure event contracts as CFTC swaps, giving them all 50 states without state gaming taxes and no infrastructure cost (they ride Robinhood, Webull, crypto.com), and Burry expects that loophole to be closed by regulation and taxation.
- Flutter's appeal is scale plus forced discipline - FanDuel is the #1 U.S. sportsbook and the debt level should compel debt paydown and buybacks, which his valuation assumes after a history of capital misallocation.
- DraftKings is an inflection bet specifically: the value is in the operating transition he expects in the near future, not in the current numbers.
- He added to JD at $27.58, already a top-three position, on a rotation call - that unwinding of the AI/memory mania in South Korea and Japan will push money into Hong Kong and Chinese equities.
- The Cassandra Unchained software and payments universe has grown to 50 names (the SW50) after adding SNPS, CDNS, FROG and IOT, and every full post now has a ~1,500-word abridged companion.
#008
Short Thoughts
Jul 08, 2026source ↗
Short Thoughts July 8, 2026 - NVDA, Neos, Hyperscalers, Jevons Paradox, and Compression
Burry argues NVIDIA's revenue is only mostly real because the AI buildout has a physical bottleneck at the frontier, not a demand engine behind it. Air Street Capital's data shows roughly 94% of announced Grace Blackwell capacity is unbuilt and only about 25% is deployed or in installation, with chips parked in Construction-in-Progress producing revenue for NVIDIA while costing buyers no depreciation at all. Old A100s and H100s rent well precisely because the new chips cannot be plugged in yet - that is evidence of the bottleneck, not of non-obsolescence - and the new data centers are being built separate from the old ones, so there is no economic upgrade path. Either the backlog gets plugged in, ending the shortage and starting a long digestion of chips aging fast, or it sits in CIP growing obsolete and creates an original-sin depreciation schedule when finally placed in service; both outcomes compress NVIDIA's recurrence rate, scarcity premium and margin. Meanwhile circular financing (an xAI SPV partly funded with NVIDIA's own off-balance-sheet equity) sustains marginal demand above market-rate financing and parks the credit risk in PE and private credit funds marking to model, so chip obsolescence threatens retirement collateral one structure away from a writedown. The trilemma has no third door: Nvidia needs permanent demand growth while hyperscalers need the buildout to end in 3-4 years, and the bulls need both to be true at once.
aineocloudaisemimag5
Positions & performance
Full summary 598 words
This Short Thoughts note is Burry's direct rebuttal to an anonymous missive a subscriber forwarded him, and it extends the depreciation argument he began in Unicorns and Cockroaches (November 26, 2025) and The Supply-Side Gluttony Recurrence (December 11, 2025). He opens by narrowing the claim: rentability was never his point and depreciation is not a bet on when a chip stops working. It is the recovery of capitalized sunk cost over a defined window of earnings at the frontier, set against terminal value. A chip can rent and still depreciate very fast economically. He never said A100s would stop functioning.
His proof that depreciation is a lever rather than a measurement is a paired accounting example. In early 2025 Amazon shortened a subset of server lives to five years from six, explicitly citing the pace of development, adding $677 million of expense across nine months. In the same quarter Meta extended its useful server lives to 5.5 years, reducing the depreciation hit to net income by $2.9 billion. Same hardware, opposite conclusions from two of the biggest. He adds Satya Nadella on the Dwarkesh Patel podcast - he did not want to get stuck with four or five years of depreciation on one generation - and Jensen Huang at GTC saying that when Blackwell ships in volume, you could not give Hoppers away. The biggest buyer and the biggest seller both concede the frontier moves faster than the schedules in use.
The forwarded missive tracks arguments CoreWeave CEO Michael Intrator has pushed at him for months across CNBC and Cramer appearances: five-year contracts, A100s fully booked, H100s coming off contract re-leased at 95% of the original price, customers willing to rent six to seven years. Burry says Intrator refuted himself at Fortune Brainstorm AI when he reframed circular financing as logistics against a physical bottleneck driven by policy, infrastructure and energy. Burry agrees with the bottleneck and inverts the conclusion: the constraint is deploying frontier chips, not redeploying old ones, so old chips are getting work they would never otherwise have received. Pointing at A100 rentals is pointing at the severity of the bottleneck.
The hard evidence is an Air Street Capital State of AI Report Compute Index study, reported by Analytics India Magazine: 100,128 Grace Blackwell GPUs deployed as of July 1 against an announced pipeline of roughly 1.66 million, with another 308,640 in installation. Hopper clusters remain the largest deployed systems; A100s cluster in research, led by Meta's 16,000-GPU Research SuperCluster. Installation-phase chips sit in Construction-in-Progress: NVIDIA booked the revenue, the buyer pays no depreciation. Deployed plus CIP is about 25%, leaving 75% as press release, and HUMAIN and Stargate Abilene are over 60% of that announced pipeline - sovereign projects that must clear financing, politics and construction. Against Huang's $1 trillion of visibility through 2027, Burry takes 94% unbuilt.
Positioning follows from the trilemma rather than from a stated trade. He is bearish NVIDIA and the hyperscalers: circular financing (an xAI SPV, some cash NVIDIA's own equity held off balance sheet) props up marginal demand, and hundreds or thousands of PE and private credit funds hold the resulting debt at mark-to-model - he points readers to The Unicus Investor's work on BlackRock TCP Capital. If Jevons paradox wins, hyperscalers borrow into a future that may not include them and become the new CLECs, writing down repeatedly. If compression wins, spending falls and NVIDIA suffers worse than a merely cyclical outcome. The bulls need a third door - permanent demand, spending that ends on schedule, and 5-6 year schedules proven correct. There is no third door.
Key metrics
- Amazon shortened a subset of server lives to 5 years from 6 in early 2025, citing pace of development - $677 million of additional expense across 9 months
- Meta extended useful server lives to 5.5 years in the same quarter, reducing the depreciation hit to net income by $2.9 billion
- 100,128 Grace Blackwell GPUs (GB200/GB300) deployed as of July 1, 2026 versus an announced pipeline of ~1.66 million - less than 5% operational
- Another 308,640 GPUs are in the installation phase; deployed plus installation is about 25% of the total
- 94% of announced Grace Blackwell capacity is unbuilt by Burry's calculation (Air Street Capital reported more than 95%)
- HUMAIN and Stargate Abilene are over 60% of the announced Grace Blackwell pipeline
- Meta Research SuperCluster, at 16,000 A100 GPUs, leads early A100 deployments concentrated in research compute
- Jensen Huang claims $1 trillion of cumulative Blackwell plus Rubin revenue through 2027 - not a big leap from the annual run rate already in evidence
- CoreWeave's Michael Intrator: contracts run five years, H100s coming off contract re-leased at 95% of the original price, customers willing to rent GPUs six to seven years
- Vera Rubin entered production at GTC 2026 in March and succeeds Grace Blackwell in the second half of 2026
- Hyperscaler free cash flow is falling toward zero even as reported profit holds, aided by long depreciation schedules
Key insights
- Burry's sharpest evidence that depreciation is an accounting lever, not a physical measurement: Amazon shortened server lives to 5 years (a $677 million expense hit over 9 months) in the same quarter Meta extended them to 5.5 years (a $2.9 billion benefit to net income), on the same hardware.
- Construction-in-Progress is the hidden mechanic - chips delivered, paid for and capitalized but not yet installed generate revenue for NVIDIA while the buyer records no depreciation, so obsolescence accrues silently and creates an original-sin schedule when the assets finally go into service.
- Only 100,128 of roughly 1.66 million announced Grace Blackwell GPUs were deployed as of July 1, 2026, with 308,640 more installing - about 25% deployed or installing and 94% of the announced capacity unbuilt.
- Concentration risk in the pipeline is extreme: HUMAIN and Stargate Abilene are over 60% of announced Grace Blackwell capacity, and both are essentially sovereign projects gated on financing, politics and construction rather than on demand.
- The bull case that old GPUs still rent (Intrator's H100s re-leased at 95% of original price) is, in Burry's reading, evidence of the frontier bottleneck rather than evidence against obsolescence - old chips get work only because new chips cannot be plugged in.
- New chips cannot go into old data centers, so there is essentially no economic upgrade path for existing facilities; the trillions being spent build separate infrastructure that will itself age the same way.
- Both resolutions are bad for NVIDIA: plugging the backlog in ends the shortage and starts a long digestion phase, while leaving it in CIP grows obsolescence - either way the recurrence rate of revenue, the scarcity premium and margins fall.
- Circular financing (an xAI SPV partly capitalized with NVIDIA's own off-balance-sheet equity) sustains marginal demand above what market-rate financing would support, and the resulting debt sits in PE and private credit funds at mark-to-model with no observable price until something forces one.
- The hyperscaler trilemma: NVIDIA needs permanent demand growth while hyperscalers need the buildout largely over in 3-4 years to protect their multiples - free cash flow is already falling toward zero while long depreciation schedules keep reported profit intact.
- Burry's CLEC analogy is the specific downside shape - laying capacity into a glut financed with too much debt, ending in multiple writedowns rather than one.
#007
Community
Jul 04, 2026
July 2026 Charity - Seva Foundation, Transforming Lives by Restoring Sight
Burry names Seva Foundation his July 2026 Charity of the Month because it is, in his words, one of the most efficient and effective charities he has come across. The arithmetic is the argument: most global blindness is cataract-driven and curable in a 20-minute surgery costing about $50 in materials, so marginal dollars convert almost directly into restored sight. Seva does not simply wire money from a distance; it builds self-sustaining local eye-care systems with partners in more than 20 countries, which multiplies reach per dollar. The organization is currently spending more than it takes in ($12.6 million of program expenses against $10.3 million of revenue), drawing down reserves to keep expanding, which is exactly why incremental funding matters now. Cassandra Unchained is donating 5% of July gross revenues.
Positions & performance
No single-stock positions stated in this report.
Full summary 430 words
This is a short, non-investment post: Michael Burry's July 2026 Charity of the Month selection for Cassandra Unchained. He names the Seva Foundation and calls it one of the most efficient and effective charities he has ever come across, pledging 5% of the publication's gross revenues during July.
The case he makes is arithmetic. Ninety percent of the world's blind population lives in low income countries, and ninety percent of that blindness is either preventable or curable. The leading cause is cataracts, which can be corrected in a roughly 20-minute surgery using about $50 worth of materials. What distinguishes Seva, in his telling, is that it does not simply provide funds from a distance: across more than 20 countries it works with partners to build self-sustaining local eye-care systems, which dramatically broadens reach at remarkably low cost. Its four main activities are Vision Centers (permanent local infrastructure), Technology (retinal cameras, handheld screening devices, ophthalmological tools), Eye Care for Children (screening, early detection and long-term follow-up), and Training and Job Creation (hiring and training ophthalmologists, support staff and community workers).
He stresses the pediatric angle: 90 million children live with vision impairment globally, and because 80% of learning is visual, poor sight limits education, causes social isolation, raises injury risk and creates barriers to employment. Last year Seva's programs reached 7 million people across 20 countries in South and Southeast Asia, sub-Saharan Africa and Latin America, plus the American Indian Sight Initiative in the United States. Against that, 2024-2025 revenues were only $10.3 million versus $12.6 million of program expenses, with the gap funded by drawing down program reserves in order to keep expanding services. Its $29 million of net assets are mostly fixed: buildings, equipment and technology installed over more than four decades.
On credibility, Burry notes Seva is headquartered in Berkeley, California, holds top ratings from Charity Navigator and CharityWatch, and is a Nicholas Kristof Holiday Impact Prize winner. He closes with the founding story: in 1977 Dr. Larry Brilliant, a World Health Organization physician, finished the work to eradicate smallpox; in 1978 he, his wife Girija, WHO physician Dr. Nicole Grasset, Harvard professor Ram Dass and the Grateful Dead's Wavy Gravy (Hugh Romney, Jr.) met in a Michigan conference room and resolved to eliminate avoidable blindness. Steve Jobs was an early donor. They named it Seva, Sanskrit for selfless service. Since 1978 the organization has helped 72 million people see. Seva is a 501(c)3 and contributions are tax deductible to the fullest extent allowed by law.
There are no tickers, prices, trades or market views in this post.
Key metrics
- 90% of the world's blind population lives in low income countries
- 90% of that blindness is preventable or curable
- Cataracts are the leading cause; a ~20 minute surgery using about $50 worth of materials
- Seva works in more than 20 countries building self-sustaining local eye-care systems
- 90 million children globally live with vision impairment; 80% of learning is visual
- Seva's sight programs reached 7 million people last year across South/Southeast Asia, sub-Saharan Africa and Latin America
- 2024-2025 revenues of $10.3 million against $12.6 million in program expenses (drawing down program reserves)
- $29 million in net assets, mostly fixed assets: buildings, equipment, installed technology
- 72 million people helped to see since 1978
- Cassandra Unchained will donate 5% of gross revenues during July
Key insights
- Burry's charity screen is explicitly a cost-per-outcome screen: cataract blindness reversed by a 20-minute procedure using ~$50 of materials is about as high a return on a charitable dollar as exists.
- Seva's model is capacity-building, not grant-making - it builds local eye-care systems with partners across 20+ countries, which is why its reach (7 million people last year) is so large relative to a ~$10 million revenue base.
- The organization is running a deliberate deficit: $12.6 million of program spending against $10.3 million of 2024-2025 revenue, funded by drawing down reserves, which is the specific reason marginal donations matter right now.
- Most of Seva's $29 million of net assets are fixed - buildings, equipment and installed technology accumulated over four decades - so the balance sheet is deployed infrastructure rather than an endowment buffer.
- The pediatric program is framed as an economic argument as much as a medical one: 90 million children have vision impairment and 80% of learning is visual, so untreated sight loss compounds into lost education and employment.
- Third-party validation is part of the pitch: top ratings from Charity Navigator and CharityWatch, plus a Nicholas Kristof Holiday Impact Prize.
- Cassandra Unchained is donating 5% of July gross revenues, tying subscriber spending directly to the program - a recurring feature of the publication's monthly charity posts.
- The founding lineage - smallpox eradicator Dr. Larry Brilliant, Dr. Nicole Grasset, Ram Dass and Wavy Gravy in 1978, with Steve Jobs as an early donor - has produced 72 million people helped to see.
#006
Trading Post
Jul 02, 2026
Trading Post July 2, 2026: A New Short and Five Additions to Long Positions
Micron is the only pure-play DRAM stock on a major U.S. exchange, so in good cycles it gets pumped more than it should and in bad ones dumped more than it should, and 42 years of data show a business that destroys capital in roughly a third of all quarters and burns cash in half of them. Right now the stock is more extended above its 200-day moving average than at any point in its history, including the dot-com peak, at 14.2x sales against a 2.2x median, after a 700%-plus one-year gain fed by scarcity, social proof, authority and commitment biases rather than by economics. HBM is simply the latest product declared precious in a long series of capacity-constrained cycles, and Micron is no longer the lead steer since South Korean capex plans dictate how much it must spend. He shorts the stock directly, since puts are expensive, and simultaneously funds the other side by adding to oversold, cash-generative long positions and to the GSEs, expecting capital to rotate out of the chip complex as the AI supply-side glut asserts itself.
aimemoryconsumerhealthfinancefintechhealth
Positions & performance
+7.8%avg return75% win · 4 tracked · L -0.2% · S +31.5%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SMU:NASDAQ aimemory Micron Technology NEWcommon | NASDAQS&PBitget | ISAPTC | 1051.87his fillJul 01 '26 | 720.04 ● | 192.591,255 | +31.5%APY +105.4% | +31.5% |
| WhyMicron is the most extended above its 200-day moving average in its 42-year history, more than at the dot-com peak, and history says a greater-than-100% extension makes a 30% drop within 12 months nearly certain; at 14.2x price/sales versus a 2.2x median, on a business with 4% median ROIC against a cost of capital above 10%, he shorted the stock directly because the puts seemed expensive. |
| LPYPL:NASDAQ financefintech PayPal Holdings ADDcommon | NASDAQS&PBitget | ISPT | 45.31his fillJul 02 '26 | 58.7 ● | 38.4663.93 | +29.6%APY +98.3% | +29.6% |
| WhyHe added on the same long-term framework he uses for shorts, run in reverse: PayPal's price/sales and its position relative to the 200-day moving average put it at the depressed end of its multi-decade range, the mirror image of Micron's extension. |
| LZTS:NYSE health Zoetis ADDcommon | S&P | IAPTC | 74.8his fillJul 02 '26 | 84.83 ● | 68.77132.49 | +13.4%APY +39.5% | +13.4% |
| WhyZoetis is especially oversold, much like Sprouts was, and he expects the fundamentals of these two stellar companies to drive them forward and to pick up the shine as money rotates out of the AI theme. |
| LSFM:NASDAQ consumerhealth Sprouts Farmers Market ADDcommon | NASDAQ | IAPT | 89.33his fillJul 02 '26 | 50.5 ● | 47.3594.89 | -43.5%APY -77.9% | -43.5% |
| WhyHe originally bought Sprouts personally at a very low price but in small size and trimmed it as it ran; the pullback lets him finally build a real position in a name he considers a stellar company. |
| LFNMA Fannie Mae ADDcommon | NASDAQ | — | —Jul 02 '26 | — | — | — | n/a |
| WhyAdded ahead of his third GSE post; the logarithmic long-term charts show where Fannie used to trade across vastly different circumstances, and he is playing the conservatorship-exit outcome. |
| LFMCC Freddie Mac ADDcommon | NASDAQ | — | —Jul 02 '26 | — | — | — | n/a |
| WhyHe owns both GSEs because in some scenarios Freddie Mac trumps Fannie Mae and he is not sure how the resolution will play out, so he sizes into both rather than picking one. |
Full summary 734 words
This trading post pairs one new short with five additions to existing longs, and Burry frames it with a line he repeats about the chip cycle: this time may be different, but not nearly different enough.
The short is Micron. He works from a chart of Micron's price from 1984 to the present with price/sales and price/tangible book in the upper panel, and the extension of the share price over its 200-day moving average in the lower panel. That extension measure, though volatile, has never been as stretched as it is in 2026, not even at the dot-com peak. Over 42 years Micron has had 34 drawdowns greater than 30%, which works out to a 30% swing a pinch too close to once a year. The fundamentals underneath are, in his words, terrible: a median ROIC of 4% and median ROE of 7% against a cost of capital well above 10%; ROA, ROIC and ROE negative in 32%, 31% and 34% of quarters respectively going back over three decades, meaning one quarter in every three Micron destroys capital; and free cash flow negative in 48% of quarters. He dismisses the sell-side focus on EBITDA margins as ridiculous for a business this dedicated to incinerating cash on big machines.
The behavioral layer matters as much as the arithmetic. Because Micron is the only pure DRAM play on a major U.S. exchange, it gets pumped too hard in good times and dumped too hard in bad ones. HBM is simply the newest product deemed precious in a long series of capacity-constrained cycles. He argues the six-month rally in all things chips, after three years of AI hype, is less a recognition of an actual shortage than the familiar late-cycle behavior of chip stocks rallying near bubble peaks. Micron's 700%-plus one-year gain, he says, exposes investors to a smorgasbord of Cialdini's psychological shortcuts: scarcity bias (sold out through 2026, the only memory pure-play), social proof (wall-to-wall coverage of AI gains), authority bias (52 of 57 analysts rate it buy with no sells, and President Trump endorsed the stock that day), and commitment bias (three years of AI gains finally reaching chip stocks). He connects this to his earlier essay Foundations: The Psychology of Investing in the Information Age, and situates the trade inside the AI thesis he has built over eight months starting with The Cardinal Sign of a Bubble: Supply-Side Gluttony, plus Unicorns and Cockroaches and The Heretic's Guide to AI's Stars Part III. He also notes Micron is no longer the lead steer in memory: South Korean capital expenditure plans basically dictate how much Micron must spend.
On evidence, his tables list every greater-than-30% decline with its peak, trough, drop size and 200-day extension. The key finding is that on the relatively few occasions when the extension exceeded 100%, a 30% fall within 12 months was nearly certain, an admittedly small sample at the extremes. The peak-to-trough crash has ranged widely but runs 2.6 months at the 25th percentile and 7.4 months at the 75th. Valuation seals it at 14.2x price/sales versus a 2.2x median, and he explicitly rejects P/E as a measure for cyclicals given the volatility of earnings. He shorted Micron at $1,051.87 the prior day, choosing the stock directly because the puts seemed expensive, and says he will look to add puts should the stock settle down and volatility come in.
The long side is the mirror image of the same framework. He added to PayPal at $45.31, showing PayPal's long-term price/sales and 200-day relative chart in the same format as Micron's. He added to Sprouts Farmers Market at $89.33, a name he had bought personally at a very low price but in small size and trimmed as it ran, so the fall lets him build a real position. He added to Zoetis at $74.80, calling it especially oversold much like Sprouts was, and arguing that the fundamentals of these two stellar companies will drive them forward and that they are the types of stocks that would pick up the shine from the AI theme. Finally he added to Fannie Mae at $6.15 and Freddie Mac at $5.69, noting the logarithmic long-term charts show where these used to trade across vastly differing circumstances, and that he owns both because in some scenarios Freddie trumps Fannie and he is not sure how it will play out. His third GSE post is coming soon.
Key metrics
- Micron shorted at $1,051.87
- Micron has had 34 drawdowns greater than 30% over the last 42 years
- Micron 200-day moving average extension is the most extreme in its history, exceeding the dot-com peak
- When 200DMA price extension has exceeded 100%, a 30% fall within 12 months is nearly certain
- Peak-to-trough crash duration: 2.6 months at the 25th percentile, 7.4 months at the 75th percentile
- Micron median ROIC of 4% and median ROE of 7% over 42 years, against a cost of capital well above 10%
- Micron ROA, ROIC and ROE are negative in 32%, 31% and 34% of quarters respectively over three decades
- Micron free cash flow is negative in 48% of quarters
- Micron price/sales of 14.2x versus a median of 2.2x
- Micron stock is up more than 700% over the last year
- 52 of 57 Wall Street analysts rate Micron a buy, with no sells; President Trump endorsed the stock
- Added PYPL at $45.31, SFM at $89.33, ZTS at $74.80, FNMA at $6.15, FMCC at $5.69
Key insights
- Micron's extension above its 200-day moving average is the most extreme in its 42-year history, exceeding the dot-com peak, and when that extension has topped 100% a 30% decline within 12 months has been nearly certain.
- The frequency argument is the base rate: 34 drawdowns greater than 30% in 42 years means Micron falls 30% slightly more often than once a year.
- Over 42 years Micron's median ROIC is 4% and median ROE 7% against a cost of capital above 10%, with returns negative in roughly a third of quarters and free cash flow negative in 48% of them.
- At 14.2x price/sales versus a 2.2x median, the multiple is the tell; he refuses to use P/E on cyclicals because earnings volatility makes it meaningless.
- He shorted the stock rather than buying puts because implied volatility made the options expensive, and he intends to convert to puts if the stock settles and vol comes in.
- He maps the bull case to Cialdini's biases: scarcity (sold out through 2026, only U.S. memory pure-play), social proof, authority (52 of 57 buy ratings, no sells, plus a Trump endorsement), and commitment.
- Micron has lost pricing agency: South Korean capex plans dictate how much Micron must spend, so it is no longer the lead steer in memory.
- The long book is the same screen inverted: PYPL at $45.31, SFM at $89.33 and ZTS at $74.80 are oversold, cash-generative businesses he expects to pick up the shine as money leaves the AI theme.
- He owns both GSEs, adding FNMA at $6.15 and FMCC at $5.69, precisely because in some resolution scenarios Freddie Mac trumps Fannie Mae and he will not pre-judge the outcome.
#005
Trading Post
Jun 30, 2026
Trading Post June 30, 2026: A Little Window-Dressing, the SOX, and Caterpillar
Quarter-end window dressing pushes funds to mark up their winners, and on June 30 that bid lands squarely on the semiconductor complex, which is already at its most extreme extension above the 200-day moving average since 2000. Burry argues the SOX is not a Nvidia problem but an index-level valuation problem: price/sales above 16x that barely moves when Nvidia is removed, making SOXX a rare and unusually clean expression of overvaluation. The proximate cause of the rally, big capital spending announced out of Korea, is to him the beginning of the end rather than evidence of demand, because supply-side gluttony is what ends these cycles. So he sells into the artificial quarter-end strength across the complex and adds duration to the options book at cheap volatility, betting on the historically typical 30%-plus SOX drawdown.
airobotaisemienergygridindexus
Positions & performance
+42.1%avg return100% win · 7 tracked · S +42.1%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF NEWcommon | NASDAQBitget | I | 642.8his fillJun 30 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +78.1% |
| WhyHe shorted the ETF outright on top of the puts because the SOXX is in his view a pure form of overvaluation in an index, rarely seen and never so easily recognized, with the index price/sales above 16x and barely dented by removing Nvidia. |
| SSOXX:NASDAQ aisemi iShares Semiconductor ETF REITputs rolled from Jan 2027 low-mid $300s strikes to March 2027 low-mid $400s strikes | NASDAQBitget | I | 640.76closeJun 30 '26 | 140.99 ● | 86.81218.65 | +78.0%APY +349.6% | +78.0% |
| WhyRolling out to March 2027 and up to low-mid $400s strikes bought him duration at relatively cheap volatility, positioning the put book closer to the money while the SOX sits at its most extreme extension above the 200-day moving average since 2000. |
| SAMAT:NASDAQ aisemi Applied Materials NEWcommon | NASDAQS&PBitget | ISPTC | 729.4his fillJun 30 '26 | 349.0 ● | 215.50739.67 | +52.2%APY +198.7% | +52.2% |
| WhyAdded as part of the semi-cap leg of the short, sold into a rally he attributes to newly announced big spending out of Korea, which he reads as the beginning of the end rather than a durable demand signal. |
| SCAT:NYSE energygrid Caterpillar NEWcommon | S&PBitget | IAPT | 1060.98his fillJun 30 '26 | 570.75 ● | 537.081,073 | +46.2%APY +169.2% | +46.2% |
| WhyCaterpillar jumped out at him on the same 200-day-moving-average extension chart he used for the SOX; he has never shorted it before and has only ever made money long, so the setup rather than the business is what triggered the short. |
| SNVDA:NASDAQ aisemi Nvidia NEWcommon | NASDAQS&PBitget | ISPTC | 198.09his fillJun 30 '26 | 155.6 ● | 148.80243.37 | +21.4%APY +66.0% | +21.5% |
| WhyShorted alongside the semiconductor complex into quarter-end strength; he stresses that the overvaluation is not just Nvidia, since removing it barely dents the SOX price/sales ratio of over 16x. |
| SQQQ:NASDAQ indexus Invesco QQQ Trust (Nasdaq 100 ETF) HOLDJanuary puts, high $500s strikes | NASDAQBitget | SC | 736.4closeJun 30 '26 | 662.32 ● | 555.60762.86 | +10.1%APY +28.4% | +10.1% |
| WhyHe calls the high $500s January QQQ puts perfect as they stand and deliberately did nothing with them, unlike the SOXX puts which needed more tenor and higher strikes. |
| STSLA:NASDAQ airobot Tesla NEWcommon | NASDAQS&PBitget | ISPTC | 416.22his fillJun 30 '26 | 379.8 ● | 297.38498.83 | +8.8%APY +24.4% | +8.8% |
| WhyHe shorted Tesla explicitly because he was happy it jumped back to the $416 level, using the quarter-end window-dressing bid as the entry point; it is not part of the SOXX and is a separate expression of the same froth. |
Full summary 549 words
This is a quarter-end trading post, and Burry opens with why the date matters. The ends of quarters are a time for window dressing, when funds around the world try to make their top positions look like the widely-acclaimed winners and sell down unpopular stocks that have recently underperformed. He believes June 30 is the most significant of these dates each year, probably because the mid-year statistics and portfolios anchor the fund-raising push into year end. He concedes he cannot prove this, only that it is what he has seen across 25 years of running money for others, during which he deliberately kept a light hand on such days so as never to be accused of window dressing himself. This June 30 is different: with his book positioned as it is, window dressing is not even possible, and he is instead trading against the day's trends.
The first action is an options roll. He moved his SOXX puts from January 2027 to March 2027 and lifted the strikes from the low-mid $300s to the low-mid $400s, gaining duration at what he describes as relatively cheap volatility. The case for pressing is technical and valuation-based. The Philadelphia Semiconductor Index sits at a historically high extension above its 200-day moving average, a level not seen since 2000; he notes that the SOX has reached the current roughly +65% extension only once before, in that same year. Bloomberg's price/sales history for the index does not run back through the 2000 peak to 1994, so he cannot make the direct comparison, but the current ratio is over 16x, and crucially, removing Nvidia from the index barely dents it. That is the core of the argument: this is not a single-stock story. He also shows the historical record of SOX drawdowns, observing that extension above the 200-day is not even a requirement for a 30% or greater decline, especially once a long bear market is in place.
His QQQ position needed no work. He calls the high $500s January puts perfect and left them untouched.
The surprise name is Caterpillar, which he says jumped out at him when he ran the same extension chart. He has never shorted Caterpillar before and notes it has always done great for him on the long side, which makes the entry a purely situational one. He shorted CAT at $1,060.98. He then shorted Nvidia at $198.09, SOXX itself at $642.80, Applied Materials at $729.40, and Tesla at $416.22, adding that he was happy Tesla jumped back to that level. He points out that not all of these names are part of the SOXX, and that SOXX itself is a pure form of overvaluation in an index, a form that is rarely seen and never so easily recognized as such.
The macro hook closes the post. He attributes the day's rally to big spending announced out of Korea, and interprets that not as demand confirmation but as the beginning of the end, consistent with his running supply-side gluttony framing of the AI capital expenditure cycle. His conclusion is that it is only a matter of time. The view would be invalidated if the SOX could sustain a 16x-plus price/sales multiple and its 2000-like extension above trend, or if the Korean capex wave translated into durable rather than glut-forming demand.
Key metrics
- Philadelphia Semiconductor Index (SOX) at a historically high extension above its 200-day moving average, a level not seen since 2000
- SOX has only reached the current +65% extension once before, in 2000
- SOX price/sales ratio over 16x, and removing Nvidia barely dents it
- Bloomberg price/sales data for the SOX does not extend back through the 2000 peak to 1994
- SOXX puts rolled from January 2027 to March 2027, strikes moved from low-mid $300s to low-mid $400s
- QQQ January puts held at high $500s strikes
- Shorted Caterpillar (CAT) at $1,060.98
- Shorted Nvidia (NVDA) at $198.09
- Shorted SOXX at $642.80
- Shorted Applied Materials (AMAT) at $729.40
- Shorted Tesla (TSLA) at $416.22
- Extension above the 200-day moving average is not required for a 30% or greater SOX drawdown, especially in a long bear market
Key insights
- He treats June 30 as the most distorted mark-up day of the year because mid-year portfolios anchor the second-half fundraising push, and he is deliberately selling into that artificial bid.
- The SOX is at an extension above its 200-day moving average last seen in 2000, roughly +65%, a level reached only once before in the index's history.
- The overvaluation is index-wide, not a Nvidia artifact: SOX price/sales is over 16x and removing Nvidia barely changes it, which is why he shorted the SOXX ETF directly at $642.80.
- Rolling the SOXX puts from January 2027 to March 2027 while raising strikes from the low-mid $300s to the low-mid $400s bought duration at cheap implied volatility rather than adding notional risk.
- Caterpillar at $1,060.98 is his first-ever short in a name that has only ever made him money long, and it was flagged purely by the 200-day extension screen, not by a fundamental change.
- History says a 30% or greater SOX drawdown does not require an extreme 200-day extension at all, so the current extension is an amplifier rather than a precondition.
- He is treating the Korean capital spending announcement as a top signal, the beginning of the end, in line with his supply-side gluttony thesis on AI capex.
- His QQQ high $500s January puts were left untouched, showing he distinguishes between the broad index hedge, which is already positioned correctly, and the semiconductor expression, which he actively pressed.
#004
Essay
Jun 29, 2026
Foundations: My 1989
Burry traces his analytical style back to 1987-1992, when he stress-tested IBM disk drives to destruction, built PCs from cash-market parts, and ran a BBS-era gaming and stock-tracking setup. The formative lesson was watching IBM itself: an 18-layer bureaucracy on a vibrant 190-acre San Jose campus that was a ghost town by 1998 and is now a residential development. That something so large, important and seemingly permanent could disappear that fast is what shaped his reading of the 1998-2000 bubble from Sand Hill Road, and it is the same lens he applies now. He closes with his September 2000 WorldCom notes, where the disqualifying fact was that after adding back acquisition costs the company was nowhere near earning its cost of capital, and he draws the direct parallel to today: the spending is capex rather than acquisitions, but the question is still what ROIC it will ever produce.
consumershopping
Positions & performance
Full summary 664 words
This is a memoir entry in Burry's Foundations series, prompted by an interview of Marc Andreessen conducted by his partner Ben Horowitz about the origins of Mosaic and Netscape. Listening to it stirred Burry's own memories of the pre-World-Wide-Web years, and the piece covers 1987 through the late 1990s, sitting alongside his earlier Foundations: My Summer of 1994 post.
He starts in high school, where he founded the Santa Teresa High School Computer Science Club and learned Turbo Pascal in a text-only MS-DOS environment. An IBM Thomas J. Watson Scholarship covered his UCLA tuition and guaranteed four summers inside IBM R&D at the Cottle Road and Almaden Research Center sites, the birthplace of the hard disk drive. His actual job was destructive testing: injecting contaminant into full head disk assemblies, recording when and where failures occurred, then disassembling the 14-inch golden ferric-oxide disks to find physical evidence. He earned $7.425 an hour against the $2.85 he had made pumping gas, and worked nights, weekends and holidays specifically for the time-and-a-half and double-time. His bench had an IBM 3270 terminal on VM/CMS; he logged data on graph paper and analyzed it in APL on a special clunky keyboard, learning matrix math a year before he formally took the course. He also absorbed how IBM worked: 18 levels of management, a 10,000-person, 190-acre campus run by a third-level manager, and CEO Akers addressing employees by closed-circuit television. At one point his failure results were literally flown to Akers while white-haired managers watched over his shoulder.
With his summer cash he built PCs, hunting newspaper ads and buying parts for cash from sellers at San Jose industrial parks and loading docks, where a 20MB drive that retailed for $400-$500 could be had for $250. Assembly was anything but plug and play: IRQ conflicts, DRAM that had to match the motherboard and the 286 or 386 CPU, VLSI and OPTi chipsets, DIP switches, jumpers, parity errors, and memory optimization via autoexec.bat and config.sys. The point of all that speed was the Wildcat bulletin board system on the 7th floor of Sproul Hall at UCLA, where he and other students played a non-live space war strategy game against players scattered across northern Europe, and where he also tracked the stock and futures positions he had bought with leftover summer earnings. By 1991-1992 he had a 60MB drive and more DRAM, but networking had become easy and therefore less interesting, and pre-med took over.
The last act runs through Vanderbilt Medical School: Mosaic in his first year, Netscape by the second, and, after his father died in his third year leaving him about $50K, the launch of sealpoint.com, branded The Market Wizard, built first in HTML, then HoTMetaL, then CorelDraw. Having just read Roger Lowenstein's Buffett biography, he published his portfolio and stock discussions, started the Value Investing thread on techstocks.com in 1996, rebranded to valuestocks.net, wrote for MSN Money, and won Forbes Best of the Web awards in 1999 and 2000 before founding Scion Capital. He reproduces his short January 2000 Ross Stores (ROST) write-up, which he links to his recent lululemon athletica (LULU) piece.
There are no trades in this post. The investment content is in the two closing observations. Returning to Silicon Valley in 1998 for medical residency, he found the IBM Cottle campus a ghost town; it was later sold to Hitachi, scraped, and redeveloped as mixed residential and retail. That disappearance of something so huge and vibrant heavily informed how he read the 1998-2000 bubble while working on Sand Hill Road. He also unearths a notepad file dated September 12, 2000 containing his first look at WorldCom, scribbled while reading form 10s, in which he was a skeptic from the start because after adding back all acquisition costs the company was not near earning its cost of capital. He calls that a timeless theme for doomed companies: today the spending is massive capex rather than acquisitions, but the question remains what the ROIC will be.
Key metrics
- IBM internship wage of $7.425/hour versus $2.85/hour as a high school gas station attendant
- IBM Cottle Road campus: 10,000 people on 190 acres, run by a 3rd-level manager with 15 management levels above him (18 levels total)
- 14-inch ferric-oxide-coated golden disks, each roughly a pound and as thick as 4-5 vinyl records
- 20MB hard drive bought for about $250 in San Jose cash markets versus $400-$500 retail
- College-era PCs built around 286 and 386 Intel CPUs with VLSI and OPTi chipsets; 60MB drive by 1991-1992
- Inherited about $50K after his father died during his third year of medical school, which funded sealpoint.com
- 1996: started the Value Investing thread on techstocks.com, which became Silicon Investor
- Forbes Best of the Web awards for stock market sites in 1999 and 2000 (valuestocks.net)
- September 12, 2000 notepad file records his first look at WorldCom
- January 2000 Ross Stores (ROST) write-up reproduced from the valuestocks.net site
Key insights
- Burry's destructive-testing job at IBM was literally about finding where and when the most expensive hardware fails, a framing that maps onto how he approaches companies today.
- Watching the 190-acre, 10,000-person IBM Cottle campus go from vibrant to ghost town within a decade is what convinced him that scale and importance offer no protection, and it shaped his interpretation of the 1998-2000 bubble.
- IBM's 18 layers of management, where a 10,000-person campus was run by a third-level manager, is his shorthand for bureaucratic decay inside an apparently dominant franchise.
- His first WorldCom note, dated September 12, 2000, rejected the company on one test: after adding back all acquisition costs, it was nowhere near earning its cost of capital.
- He explicitly generalizes the WorldCom test to the present, noting that acquisitions are less common now than massive spending, but the ROIC question is identical, which links this essay to his AI capex thesis.
- His research method has never been formal: open a notepad file and start scribbling while reading the filings.
- The Ross Stores (ROST) write-up from January 2000 is reproduced here because the same argument feeds his current lululemon athletica (LULU) piece on bad management in retail.
- His investing platform was built from a medical student's side project, from sealpoint.com to valuestocks.net to Forbes Best of the Web in 1999 and 2000, before Scion Capital.
#003
Single Stockabridged
Jun 26, 2026
[Abridged] lululemon athletica (LULU): An Ode to Bad Management & The Politics of Skin-Tight Leggings
lululemon shareholders have suffered a long series of management own goals - the 2011 John Galt bags, the 2013 sheer leggings and Chip Wilson's comments, the 2024 seam-and-transparency repeat, fumbled color and size inventory, and a Japanese war drum at a Great Wall yoga event in its best growth market. Those self-inflicted wounds now sit on top of Trump tariffs costing 280 basis points of gross margin and the de minimis repeal that forces US e-commerce to ship from Canada, with no permanent CEO to fix any of it until September 8. Burry argues this is a near-exact rerun of the 2013-2018 episode, which ended with gross margins back above 58% within a year of a new CEO and the stock quadrupling, and that the market is again extrapolating temporary damage forever. With tangible book per share doubled to ~$40, the stock at ~2.5x tangible book and under 10x earnings - uniquely among US clothing retailers - and a P/IV15 of 0.72x, value becomes its own catalyst, aided by a $1 billion buyback against a ~$12 billion market cap. The risk is entirely execution: the incoming CEO is unproven and the market has punished her appointment.
consumershoppingfinancefintechhealthmag5saas
Positions & performance
-15.4%avg return0% win · 1 tracked · L -15.4%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LLULU:NASDAQ consumershopping lululemon athletica REITcommon | NASDAQS&P | ISAPT | 124.65his fillJun 26 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -15.4% |
| WhyThis abridged version restates the long: he added to LULU the day before the original piece, putting his average cost at $124.65 against a share price near $105. The reiterated case is that IV15 is $155.81 (P/IV15 of 0.72x, an IV18 worth roughly an 18% CAGR over 15 years), that LULU is the only US clothing retailer under both 3x tangible book and 10x earnings, and that the tariff, de minimis, product and CEO-vacuum problems are temporary and mostly self-inflicted. |
Full summary 605 words
This is the free, abridged version of the June 24 deep dive on lululemon, condensing the same argument into a tighter form. Burry frames LULU as a company whose shareholders have suffered through an extraordinary run of own goals: the 2011 'Who is John Galt' shopping bags that offended its core base of liberal 25-45 year-old women; the 2013 discovery that the flagship yoga pants were transparent, followed by Chip Wilson telling Forbes 'you don't want certain customers coming in'; gross margins cratering from nearly 56% to ~48%; and a Trump-era law on overseas operations that cost another 250 basis points. That five-year malaise ended when Calvin McDonald became CEO, gross margins returned to all-time highs above 58% within a year, and the stock reached $160 by September 2018 - quadrupling the 2014 low.
He argues today's turmoil mirrors the 2010s, and calls that a good thing. The current troubles began in 2024 with new leggings featuring an unflattering seam and transparency when stretched, compounded by fumbled color and size inventory and a management vacuum awaiting a new CEO at summer's end. Tariffs are hitting gross margins by 280 basis points, and the removal of the US de minimis exemption is worse in the short run because LULU must ship all e-commerce from Canada. On May 27 the company settled founder Chip Wilson's proxy fight, seating two of his directors.
The historical analogy is Ross Stores, which he bought at ~7x earnings in October 2000 when he opened Scion Capital and everyone else was gaming the next TMT rally; ROST compounded at 20.76% for over 25 years, double the S&P's 10% and ahead of Microsoft's 12%. He overlays the charts and notes their similarity up to that red line, arguing LULU's position next to 'stocks for bubbleheads' is the same setup. China remains the bright spot but is dimming, with constant-currency growth slowing to 13% from 19%, not helped by the Japanese war drum used at a Great Wall yoga event with 2,000 guests - another own goal followed by a halfway apology.
On the brand, he dismisses ALO and Vuori as incremental threats benefiting from base effects and influencer spend, noting Vuori skews male and lifestyle rather than performance and that at least 95% of LULU customers shop neither; LULU leads on retention and dominates its customers' athleisure spend. He adds the political overlay: LULU's college-educated, $100k-plus, urban female 25-45 customer is per Pew the most left-leaning cohort in retail, the Federal Reserve finds the partisan confidence gap exceeds income, age and education combined, and Democratic sentiment sits near 32 against 87 for Republicans - so if even 150-250bp of comp weakness is political, competitive pessimism is misplaced.
The valuation section drives the positioning. Tangible book per share has doubled from $20 to ~$40 even after buybacks well above book; at $105 the stock is ~2.5x tangible book, lowest since 1Q 2009, and it is the only US clothing retailer below both 3x tangible book and 10x earnings. With over $2 billion of pre-tariff operating income, $1.5 billion net cash, 108 million shares (down from 145 million) and $1 billion of buyback left against a ~$12 billion market cap, he sees a spring-loaded franchise. IV15 is $155.81 for a P/IV15 of 0.72x, putting LULU alongside Adobe at about IV18 and an expected ~18% CAGR; he added the prior day at an average price of $124.65. The caveats: not all IV18s are fat pitches, only 2 of 32 analysts are positive, and the stock is down 31% since Heidi O'Neill - criticized for Nike's DTC execution - was named CEO, arriving only on September 8.
Key metrics
- IV15 of $155.81 per share; P/IV15 = 0.72x, placing LULU at about IV18 alongside Adobe with a roughly 18% CAGR expectation over 15 years
- Average cost after the latest add: $124.65
- Tangible book value per share doubled from ~$20 to ~$40 in 3+ years despite buybacks well above tangible book; at $105 the stock is ~2.5x tangible book, the lowest multiple since 1Q 2009
- Only US clothing retailer of any market capitalization trading below 3x tangible book and below 10x earnings
- Tariffs are a 280bp negative impact on gross margins; the de minimis exemption removal is worse near term because e-commerce must ship from Canada
- China constant-currency growth slowed to 13% from 19% last year, with management guiding to a pickup for the rest of the year
- Gross margins in the 2013 episode fell from nearly 56% to ~48%, then recovered above 58% within a year of Calvin McDonald taking over; the stock hit $160 by September 2018
- Ross Stores compounded at 20.76% a year for over 25 years from October 2000, double the S&P 500's 10% and ahead of Microsoft's 12%
- Balance sheet and capitalization: over $2 billion of pre-tariff operating income, $1.5 billion net cash, 108 million shares out versus 145 million a decade ago, ~$12 billion market cap and $1 billion of buyback remaining
- Michigan consumer sentiment: Democrats near historic lows around 32 versus Republicans at 87; possibly 150-250bp of comp weakness is politically driven
- Only 2 of 32 covering analysts rate LULU a buy; the stock is down 31% since Heidi O'Neill was chosen as CEO, and she does not arrive until September 8
- LULU has traded 16.6x its shares outstanding since the December 29, 2023 peak at $516.39
- Fiserv is described as only an IV12 despite a long decline; Veeva is a fat pitch but not yet an IV15
Key insights
- The abridged case rests on pattern recognition: 2011, 2013 and 2024 are the same failure mode, and the 2013 version resolved with margins above 58% within a year of a new CEO and a quadrupling stock.
- At ~$105 LULU trades at ~2.5x tangible book, the lowest since 1Q 2009, and is the sole US clothing retailer under both 3x tangible book and 10x earnings.
- Tangible book value per share doubled from $20 to ~$40 in three-plus years despite buybacks executed well above tangible book - evidence the underlying returns on capital are intact.
- The de minimis repeal, not the 280bp tariff hit, is the more damaging near-term margin problem because all US e-commerce now ships from Canada.
- China is still the growth engine but decelerating - constant-currency growth of 13% versus 19% last year - and the Great Wall war-drum misstep hurt in the very market the thesis needs.
- Roughly 150-250bp of comp weakness may be political rather than competitive, given Democratic consumer sentiment at ~32 versus 87 for Republicans and LULU's uniquely left-leaning customer base.
- He is explicit that not all IV18s or IV20s are fat pitches, contrasting LULU with Fiserv (only an IV12 after a long fall) and Veeva (a fat pitch but not yet IV15).
- Turnover of 16.6x shares outstanding since the December 2023 peak at $516.39 means high-basis holders are gone, so a rally would face little overhead resistance.
- The governance angle cuts both ways: Elliott owns ~5% and Chip Wilson 8%, yet the board picked Heidi O'Neill over both of their preferences, which is why expectations - and the share price, down 31% since her selection - are already depressed.
#002
Trading Post
Jun 25, 2026
Trading Post June 25, 2026: Hong Kong, Adds, One Sale to Add, One New Position and One Short Cover
The chip mania is draining capital out of Hong Kong and into South Korea and Japan, as momentum chasers and dedicated Asian funds reallocate, which is visible in the divergence between the Hang Seng and the TOPIX and KOSPI. That flow has broken the charts of China's leading companies and pushed them back near their lows, and it may continue - but it is technical pressure, not a fundamental verdict on the businesses. Burry therefore uses the weakness to add to Chinese names (selling Alibaba only to book a tax loss and rotate into JD, with Tencent and Meituan next), while continuing to build his discounted US positions in Fiserv and Adobe and taking cheap long-dated Microsoft optionality instead of paying current prices for the stock. On the short side he takes half his Palantir risk off into the decline but keeps the bearish exposure through puts.
aichina-aiairobotfinancefintechmag5saas
Positions & performance
+29.1%avg return100% win · 4 tracked · L +29.1%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LMSFT:NASDAQ mag5 Microsoft NEWDecember 15, 2028 LEAP calls, strikes in the low $700s | NASDAQS&PBitget | ISAPT | 352.83closeJun 25 '26 | 519.14 ● | 349.20559.15 | +47.1%APY +164.4% | +47.1% |
| WhyA new position taken in long-dated call options rather than stock. He says he considers $350 the good place to buy Microsoft common, but bought the December 2028 LEAPs because at low-$700s strikes he found them inexpensive - cheap long-dated optionality instead of paying up for shares at current levels. |
| LFISV:NASDAQ financefintech Fiserv ADDcommon | NASDAQS&PBitget | ISAPT | 47.55his fillJun 25 '26 | 62.51 ● | 43.8770.41 | +31.5%APY +99.1% | +31.5% |
| WhyAdded to the existing Fiserv long at $47.55, continuing to build in a name he has described as having fallen a very long way to reach his fat-pitch zone. |
| LADBE:NASDAQ saas Adobe ADDcommon | NASDAQS&PBitget | IPT | 195.11his fillJun 25 '26 | 248.37 ● | 190.12362.71 | +27.3%APY +83.6% | +27.3% |
| WhyAdded to Adobe at $195.11, one of the two stocks he currently scores at about IV18 - his deepest discount-to-intrinsic-value bucket, implying roughly an 18% long-run compounding rate. |
| LJD:NASDAQ airobot JD.com ADDcommon | NASDAQBitget | IS | 24.79his fillJun 25 '26 | 27.41 ● | 24.5134.30 | +10.6%APY +28.8% | +10.6% |
| WhyFunded by the Alibaba tax-loss sale. Hong Kong stocks are being sold hard as the semiconductor narrative pulls capital out of Hong Kong and into South Korea and Japan, pushing China's leading companies back toward their lows; he treats that as technical flow pressure from momentum chasers and reallocating Asian funds, not a fundamental deterioration. |
| LBABA:NYSE aichina-ai Alibaba Group EXITcommon | Bitget | ISAP | 95.07closeJun 25 '26 | 90.92 ● | 90.57181.10 | -4.4%APY -10.6% | -4.4%closed |
| WhySold Alibaba specifically to harvest a tax loss and recycle the proceeds into more JD, noting the leading Chinese names generally track each other in the short run. He states plainly that he expects to buy Alibaba back, so this is a tax trade rather than a change of view. |
| SPLTR:NASDAQ saas Palantir Technologies COVERcommon · half of the short | NASDAQS&PBitget | ISPTC | 107.15his fillJun 25 '26 | 159.5 ● | 106.37212.94 | -48.9%APY -81.5% | -48.9%closed |
| WhyCovered half of the Palantir common short at $107.15, taking risk down after the decline while explicitly keeping the bearish exposure on through puts he continues to hold. |
| L700 HK Tencent Holdings HOLDcommon | — | C | —Jun 25 '26 | — | — | — | n/a |
| WhyHolds and says he expects to buy more alongside Meituan, since the Chinese leaders track each other short term and the current drawdown is driven by capital rotating from Hang Seng into the Japanese TOPIX and Korean KOSPI on the chip mania rather than by company fundamentals. |
| L3690 HK Meituan HOLDcommon | — | — | —Jun 25 '26 | — | — | — | n/a |
| WhySame rationale as Tencent: he expects to buy more as the Hong Kong tape breaks down on technical, flow-driven selling tied to the global semiconductor mania rather than on deteriorating business performance. |
Full summary 268 words
This is a short position-by-position trading log rather than an argued essay. Burry frames the day around Hong Kong, where stocks are falling hard because the semiconductor narrative is pulling capital out of Hong Kong and into South Korea and Japan. Momentum chasers are doing part of it, but he also points to dedicated Asian equity funds that can easily reallocate, and says the effect is plain in the charts of the Hang Seng versus the Japanese TOPIX and the Korean KOSPI. The result is that China's leading companies have been pushed back down near their lows and their charts are broken, which he concedes can continue. His key judgment is that this is a feature of the global semiconductor mania and therefore largely technical pressure, not fundamental.
Acting on that, he sold Alibaba purely to harvest a tax loss and used the proceeds to buy more JD, adding at $24.79, and says he expects to buy Alibaba back and to buy more Meituan and Tencent as well, since these names generally track each other in the short run. Outside Asia he opened a new position in Microsoft December 15, 2028 LEAP calls at strikes in the low $700s, explaining that he regards $350 as the good level to buy Microsoft common but found the long-dated calls inexpensive. He added to Fiserv at $47.55 and to Adobe at $195.11, both names he has flagged as deeply discounted to his intrinsic value framework. On the short book, he covered half of his Palantir short at $107.15 while continuing to hold puts, taking down risk without abandoning the bearish position.
Key metrics
- Added JD at $24.79
- Added Fiserv (FISV) at $47.55
- Added Adobe (ADBE) at $195.11
- Covered half the Palantir (PLTR) short at $107.15, continues to hold puts
- Bought Microsoft December 15, 2028 LEAP calls with strikes in the low $700s
- States $350 is a good level to buy Microsoft common stock
- Sold Alibaba for a tax loss to fund the JD add, expects to buy Alibaba back
Key insights
- He reads the Hong Kong sell-off as a flow problem, not a fundamentals problem: the chip narrative is siphoning Asian allocations from the Hang Seng into the TOPIX and KOSPI, and he is willing to buy into broken charts on that basis.
- The Alibaba sale is a tax-loss trade, not a thesis change - he states outright that he expects to buy it back, and rotated the proceeds into JD at $24.79 because the Chinese leaders track each other short term.
- Tencent and Meituan are flagged as the next adds, meaning the China exposure is being concentrated rather than reduced into the drawdown.
- He prefers cheap long-dated optionality to paying up: he wants Microsoft common at $350 but expressed the view through December 2028 LEAP calls struck in the low $700s because they were inexpensive.
- Fiserv at $47.55 and Adobe at $195.11 are continued averaging down in his most discounted US names, consistent with Adobe being one of only two stocks he scores at about IV18.
- The Palantir short is being managed, not closed: half the common short covered at $107.15 with puts retained, keeping convex downside exposure while banking realized profit.
#001
Single Stock
Jun 24, 2026
lululemon athletica (LULU): An Ode to Bad Management & The Politics of Skin-Tight Leggings
lululemon is a structurally excellent, cash-generative franchise that has been repeatedly damaged by its own management - the 2011 John Galt bags, the 2013 sheer-pants episode, the 2024 Breezethrough leggings, botched color and size inventory, a Japanese war drum at a Great Wall yoga event - layered on top of exogenous hits from Trump II tariffs and the removal of the de minimis exemption. Because there is no permanent CEO until September 8, none of this can be fixed yet, so the market has extrapolated temporary and self-inflicted problems into permanent brand decline. Burry argues competition from ALO and Vuori is incremental rather than existential (at least 95% of LULU customers are not their customers) and that part of the US comp weakness is political rather than competitive, given LULU's uniquely left-leaning customer cohort and record-low Democratic consumer sentiment. Meanwhile tangible book value per share has doubled to ~$40 and the stock has fallen to ~2.5x tangible book and under 10x earnings, the only US clothing retailer on both measures, at an IV15 discount he scores as an IV18. Value therefore becomes its own catalyst, with a new CEO, activist and founder board representation, and a $1 billion buyback as accelerants; the weak spot is that the entire thesis depends on execution by an unproven incoming CEO.
consumershoppingfinancefintechhealthmag5saas
Positions & performance
-15.4%avg return0% win · 1 tracked · L -15.4%
| Position | Tags | Mentions | Entry | Last | 52w range | Since 1st | Return |
|---|
| LLULU:NASDAQ consumershopping lululemon athletica ADDcommon | NASDAQS&P | ISAPT | 124.65his fillJun 24 '26 | 105.4 ● | 90.38225.98 | -15.4%APY -34.3% | -15.4% |
| WhyHe added to the position the day before publication, taking his average cost to $124.65 with the stock recently touching $104-105. His case: LULU trades at roughly 2.5x tangible book (lowest since 1Q 2009) and under 10x earnings, the only US clothing retailer meeting both tests, while his IV15 estimate is $155.81 (P/IV15 of 0.72x, an IV18) implying about an 18% CAGR over 15 years. The damage - Trump II tariffs, de minimis removal, product and inventory fumbles, a CEO vacuum until September 8 - he judges temporary and largely self-inflicted, with a $1 billion remaining buyback against a ~$12 billion market cap doing real work down here. |
Full summary 627 words
Burry opens by cataloguing lululemon's history of self-inflicted crises, because his argument is that today's collapse rhymes with an earlier one that ended in a quadrupling of the stock. Crisis one was 2011, when founder Chip Wilson, an Ayn Rand devotee, printed 'Who is John Galt?' on tens of thousands of the company's shopping bags and offended the progressive, mostly female 25-45 year-old core customer; Burry had his wife buy one and framed it. Crisis two was 2013, when the flagship yoga pants proved transparent and Wilson blamed women's bodies on Bloomberg TV and told Forbes 'you don't want certain customers coming in.' Gross margins fell from nearly 56% to ~48%, an overseas tax law under Trump's first term cost another 250 basis points, the CEO carousel spun, and the stock languished five years - until Calvin McDonald arrived, margins returned above 58%, the stock broke the 2013 high in March 2018 and hit $160 by September 2018.
Today's version is nearly identical. The 2024 Breezethrough leggings had an unflattering seam and stretched sheer and were pulled in July 2024; color and size inventory was fumbled; four earnings calls have failed to address it; and there is no permanent CEO until Heidi O'Neill arrives from Nike on September 8. Chip Wilson's proxy fight was settled May 27 with two of his directors on the board. On the June 4 call, interim co-CEO Meghan Frank detailed a 330 basis point product margin decline, 280bp of it tariffs, and Burry stresses that the de minimis repeal is worse in the short run because e-commerce must now ship from Canada. Q1 revenue was $2.5 billion, comps -2%, North America comps -6%, with China Mainland up 30% (13% comps, 8 points of it a Chinese New Year shift) even after lululemon used a Japanese war drum at a Great Wall yoga event and had to half-apologize. Guidance calls for flat to -1% full-year revenue, a 410bp Q2 gross margin decline and 500bp of SG&A deleverage.
The historical analogy is Ross Stores. Burry overlays ROST's chart into October 2000 - when he opened Scion Capital and bought it at ~7x earnings while everyone chased TMT - onto LULU's chart today, calling them doppelgangers. ROST went from $1.82 to $227.42, a 20.76% CAGR over 25 years, double the S&P's ~10% and ahead of Microsoft's 12%, with store count up only 5.6x; the rest was margin expansion, multiple expansion and buybacks. He also cites Abercrombie & Fitch, American Eagle and Urban Outfitters as proof that buying beaten-down branded apparel works.
On the brand, he dismisses ALO and Vuori as incremental threats - Vuori skews male and lifestyle, ALO leans on expensive influencer marketing, and at least 95% of LULU customers shop neither - and cites Lawrence and Lozanski's 2014 Canadian Review of Sociology paper on lululemon folding empowerment into consumerism. He adds a political overlay: Pew says LULU's cohort is the most left-leaning in retail, the Federal Reserve finds the partisan confidence gap exceeds income, age and education combined, and Michigan sentiment is ~32 for Democrats versus 87 for Republicans, so 150-250bp of comp weakness may be sentiment, not share loss.
Positioning follows the valuation: tangible book per share has doubled to ~$40, the stock at $105 is ~2.5x tangible book (lowest since March 2009) and under 10x earnings, uniquely so among US clothing retailers, with $1.5 billion net cash, 108 million shares and $1 billion of buyback left. His IV15 is $155.81, a 0.72x P/IV15 that puts LULU alongside Adobe at about IV18 and an expected ~18% CAGR. He added the prior day at an average cost of $124.65. The acknowledged weak spot: O'Neill ran Nike's stumbling DTC push, the stock is down 31% since her appointment, and the thesis is entirely execution-dependent.
Key metrics
- IV15 of $155.81 per share; P/IV15 = 0.72x, an IV18, implying roughly 18% CAGR over a 15-year hold
- Average cost after the latest add: $124.65; shares recently touched $104 versus $423.32 on January 30, 2025 and the $516.39 peak of December 29, 2023
- Tangible book value per share doubled from ~$20 to ~$40 in 3+ years; at $105 the stock is ~2.5x tangible book, lowest since 1Q 2009 (2.14x on March 9, 2009)
- Only US clothing retailer of any market cap trading below 3x tangible book and below 10x earnings
- Q1 net revenue $2.5 billion, +4% (+2% constant currency), comps -2%; North America revenue -3% (-4% cc) with comps -6%; China Mainland +30% (+23% cc), comps +13% with 8 points from the Chinese New Year shift; Rest of World +13% (+9% cc)
- Q1 product margin down 330bp: tariffs -280bp, efficiency initiatives +100bp, markdowns -40bp; ex-tariffs gross margin would have been up ~60bp
- Guidance: FY2026 revenue $11.0-11.15 billion (flat to -1%), North America down high single digits, China Mainland up ~20%, Rest of World up mid-teens; Q2 gross margin -410bp and SG&A deleverage of 500bp
- Balance sheet: ~$1.5 billion cash, insignificant financial debt, ~$600 million undrawn revolver, Q1 inventory $1.7 billion (+2% in dollars, -4% in units), 108 million shares out versus 145 million a decade ago, ~$12 billion market cap and ~$10 billion enterprise value
- Repurchased ~2.2 million shares in Q1 at an average price of $165, with ~$1 billion left on the authorization
- Ross Stores went from $1.82 (post-split) in October 2000 to $227.42 today, a 20.76% CAGR over 25 years, versus ~10% for the S&P 500 and 12% for Microsoft ($22 to $411)
- Only 2 of 32 covering analysts rate LULU a buy; the stock is down 31% since Heidi O'Neill was named CEO
- Michigan consumer sentiment: Democrats near historic lows around 32 versus Republicans at 87; he attributes possibly 150-250bp of comp weakness to politically depressed customers
- 800+ stores, international revenue over $3 billion (~30% of total), men's now 24% of revenue; LULU has traded 16.6x its shares outstanding since the December 2023 peak
Key insights
- The setup is explicitly a repeat of 2013-2018: same transparency scandal, same inventory and CEO chaos, same Trump tax or tariff hit to gross margin - and that cycle ended with margins back above 58% and the stock quadrupling off the 2014 low to $160 by September 2018.
- Valuation is the catalyst: at ~$105 LULU is ~2.5x tangible book, the cheapest since 2.14x on March 9, 2009, and the only US clothing retailer of any size below 3x tangible book and 10x earnings.
- Tangible book value per share doubling from ~$20 to ~$40 in three-plus years, achieved alongside buybacks, is his proof that returns on equity and invested capital remain genuinely high despite the operational mess.
- The de minimis exemption repeal is a bigger near-term margin problem than tariffs because it forces all US e-commerce to ship from Canada, on top of the 280bp tariff hit in Q1.
- Competitive fear is overdone: at least 95% of LULU customers buy neither ALO nor Vuori, Vuori is male-skewed and lifestyle-led, and ALO's growth leans on expensive influencer marketing off a small base.
- A political read on the comps - Michigan sentiment at ~32 for Democrats versus 87 for Republicans against LULU's uniquely left-leaning 25-45 female, $100k+ household customer - suggests 150-250bp of the comp weakness is macro-political rather than competitive.
- Elliott owns ~5% and Chip Wilson 8%, and the board picked Heidi O'Neill over both of their preferences; her failed Nike DTC record is why the stock is down 31% since her appointment, which is also why expectations are already priced in.
- Extreme turnover - 16.6x shares outstanding traded since the December 2023 peak at $516.39 - means trapped high-basis holders are largely gone, removing overhead supply and allowing a fast move if the business turns.
- Only 2 of 32 analysts rate the stock a buy, and Bernstein's two objections (newness that was not new enough, margin pressure for another year) both require the new CEO to resolve - which Burry is willing to wait a year for.
No reports match your filters.