Michael Burry's Portfolio: What the Big Short Investor Owns Now
$1.1 Billion Against AI
Michael Burry's Q3 2025 13F filingA quarterly SEC report where large investors must disclose their stock holdings. showed a portfolio worth $1.38 billion, up 139% from the prior quarter. Nearly 80% of it was concentrated in put optionsContracts giving the right to sell a stock at a set price. They gain value when the stock drops. on two stocks: Palantir and NVIDIA. In plain English, he's betting that the two highest-profile AI companies are overvalued and heading down.
The Palantir put alone covered 5 million shares worth $912 million. The NVIDIA put covered 1 million shares worth $187 million. On the other side of the portfolio, he held call options on Pfizer and Halliburton, plus small equity positions in Molina Healthcare, Lululemon, and SLM Corp.
Then, on October 27, 2025, Burry sent a letter to investors announcing he was shutting down Scion Asset Management. He deregistered the fund on November 10. His stated reason: he didn't want to manage other people's money through the prolonged downturn he expected.
The Record Since the Big Short
Burry's fame comes from one trade. Between 2005 and 2008, he bet against subprime mortgages and turned Scion Capital's $600 million into a 489% return while the S&P 500 returned less than 3%. That trade made him a legend. Everything since has been more complicated.
| Period | Call | Result | Verdict |
|---|---|---|---|
| 2005-2008 | Short subprime mortgages | +489% return | Right |
| 2017 | Global financial meltdown | S&P +14.5% that year | Wrong |
| 2019 | Index funds are the next CDOs | Record inflows continued | Wrong |
| 2019-2020 | Long GameStop at $0.83 | Sold at $3.38, 4x return | Right (but early) |
| 2023 | Long China tech at 11.6x P/E | 39.9% gain in early 2025 | Right (then reversed) |
| 2025 | Short Palantir and NVIDIA | Open / Fund Closed | TBD |
Since relaunching Scion Asset Management in 2013, Burry has returned 460% total, with a three-year annualized return of 25.48%. The S&P 500 returned 23.4% annualized over the same three years. He's beaten the market, but not by the margin you'd expect from someone whose name is synonymous with prescience.
The GameStop Lesson
Burry's GameStop trade is instructive because it shows both his strength and his weakness. He first bought in the summer of 2018 when nobody was paying attention, built a 3-million-share position by Q3 2019, and laid out a clear thesis: share buybacks, a console refresh cycle, and potential buyout made the stock cheap at under a dollar (split-adjusted).
He was right. The stock eventually went from under $1 to over $120 in the WSB-fueled short squeeze of January 2021. But Burry sold his entire position in November 2020 at an average price of $3.38. He turned 4x on his money, which is excellent. He also left a potential 14,400% return on the table.
Burry later admitted he had no idea the short squeeze was coming. This is the Burry pattern in miniature: identify a genuinely mispriced asset, take a position, get the direction right, but exit before the biggest move because the catalyst isn't what he predicted.
Why Burry Is Shorting AI
Burry's AI thesis has three parts, and he's laid them out in detail through his Substack newsletter "Cassandra Unchained," which he launched in November 2025 at $379 per year.
Why He Might Be Wrong
Burry has called a bubble roughly once every two years since 2008. He predicted a financial meltdown in 2017 that didn't happen. He called index funds the next CDOs in 2019 and they've only grown. He posted "Sell" on Twitter in February 2023, then deleted his account; the S&P 500 gained 24% that year.
The pattern suggests that Burry's framework is permanently calibrated for disaster. He looks at every market through the lens of 2008, searching for the structural fraud or hidden risk that will cause a collapse. Sometimes that lens reveals real problems. More often, it produces false alarms.
| Prediction | Date | What Happened |
|---|---|---|
| Global financial meltdown + WW3 | May 2017 | S&P 500 +14.5% by year-end |
| Index funds are the next CDOs | Sep 2019 | Record index fund inflows |
| "Sell" (deleted tweet) | Feb 2023 | S&P 500 +24% in 2023 |
| AI is a bubble | 2025 | NVIDIA +200%+ since thesis formed |
There's a structural difference between AI and the dot-com bubble that Burry may be underweighting. Dot-com companies had revenue plans. AI companies have revenue. NVIDIA generated $130 billion in revenue in fiscal 2025. Microsoft's AI-powered cloud revenue is growing 35%+ year-over-year. The AI trade is expensive, but it's not imaginary.
How to Use Burry (Without Following Him)
Burry is most useful as a signal, not a strategy. When he identifies a structural problem, pay attention. His analysis of depreciation accounting in AI is worth investigating regardless of whether his short thesis works. When he takes a position, don't copy it. His timing is historically inconsistent, his positions are sized for a hedge fund's risk tolerance, and he's exited winning trades early multiple times.
What Burry Gets Right
He identifies structural risks before anyone else is looking. His depreciation thesis on AI chips raises a genuine accounting question. He finds value in forgotten corners of the market (GameStop, China tech, Pfizer). His willingness to bet against consensus is rare and occasionally very profitable.
What Burry Gets Wrong
Timing. Timing. Timing. He sold GameStop at $3.38 before it hit $120. He called the 2017 crash that never came. He views every bull market through the lens of 2008. He closed his fund rather than manage money through his own predicted crash. His post-2008 outperformance over the S&P is only 2.1 percentage points annualized.
Methodology
