Michael Burry is back at the center of market debate after a fresh set of disclosed trades put Nvidia at the top of his latest short list.
Burry, the fund manager whose subprime short was immortalized in The Big Short, disclosed on June 30 through his Substack newsletter Cassandra Unchained that he had opened bearish positions against Nvidia, Tesla, Applied Materials, Caterpillar, and the Philadelphia semiconductor ETF SOXX. Nvidia was listed with an entry price of $198.09. On July 1, he opened a new short in Micron Technology at $1051.87. On July 25, he published another note saying he had increased his bearish exposure to Nvidia, Micron, and SOXX.
Burry's case centers on AI infrastructure accounting
The core of Burry's argument is familiar. He says AI infrastructure buildout may be supported by two weak points: what he calls off-balance-sheet circular financing and overstated depreciation lives. In his view, the way Nvidia's customers account for spending could make profit across the supply chain look healthier on paper than it really is.
The reaction on social media was immediate and predictable. For many traders, it looked like another round of Burry's standard bearish script. What mattered more was how Nvidia traded after the disclosures.
On June 30, the day Burry first disclosed the position, Nvidia closed at about $200, roughly in line with his stated $198.09 entry. In early July, the Philadelphia semiconductor index dropped more than 6% in a single session after Meta said it would lease out idle AI computing capacity, among other factors. Nvidia fell toward $194, then rebounded quickly and moved back above $210 by mid-July.
After Burry added to the trade on July 25, Nvidia fell nearly 5% on July 27 to $196.51. That drop was not driven directly by his disclosure. The immediate catalyst was a report saying Nvidia was discussing about $250 billion in financing guarantees for OpenAI. Burry then wrote on social media that the arrangement would amount to “Nvidia guaranteeing customers to buy its own chips,” echoing his circular-financing thesis and adding to market concern.
Nvidia then traded in a $190 to $200 range, touched a low of $190 on July 29, recovered quickly, and closed at $200.75 on July 31. At that level, Burry's June 30 short entered at $198.09 was roughly flat with a slight unrealized loss, while the portion added on July 25 at $210.28 was down about 4.5% on paper.
After 2008, Burry's record has been mixed
That pattern is familiar. Burry has spent years issuing bearish calls, and the market has spent years questioning how often he gets the timing right. That leads to a harder question than whether his latest Nvidia short works: when a professional pessimist speaks again, how much weight should investors give the view, and how many people on Wall Street are actually willing to put real size behind the same trade.
Burry's defining victory came in 2008. Before the subprime crisis broke open, he reviewed large numbers of mortgage contracts one by one and concluded that many securities rated AAA were backed by borrowers who had little chance of repaying. The trade made him famous and gave him the label he still carries.
What gets less attention is what happened after that.
At the end of 2020, Burry built a short against Tesla. The stock then rose rather than fell. By the end of the first quarter of 2021, the disclosed position was interpreted by the market as a wager worth hundreds of millions of dollars, though much of that reflected notional option value rather than premium actually paid. The position had disappeared from holdings by the third quarter of 2021, and multiple reports concluded that he most likely exited with losses.
In GameStop, he sold weeks before the stock's explosive rise and missed the most violent part of the squeeze. Years later he bought again, then exited on his own after disagreeing with the debt risk and valuation logic behind the company's acquisition of eBay.
His history with Nvidia and Palantir shows two different versions of the same instinct. In early 2023, he posted a tweet that simply said “sell.” It was a broad market warning, not a company-specific attack. The market then rallied hard, and by late March Burry acknowledged that he had called it wrong. He did not target Nvidia and Palantir directly until November 2025, when he disclosed that about 80% of his portfolio was tied to put options on the two names. Palantir then delivered better-than-expected earnings, the stock dipped only briefly, and quickly recovered to new highs.
By the first half of 2026, those two shorts had gone in different directions. Nvidia kept climbing, with its market capitalization at one point topping $5 trillion, leaving his position under water. Palantir, by contrast, fell more than 10% in one week after Burry publicly questioned whether rivals were eating into its AI business.
That said, his record is not made up only of misses. He warned in 2019 that passive money flowing indiscriminately into large-cap stocks could distort pricing. He later told the media that he shorted into the 2020 pandemic crash and realized gains. He has also said that he expected the meme-stock frenzy to fade by mid-2021.
He is often better at spotting risk than timing it
Put the wins and losses side by side and one line stands out. The times Burry was right tended to combine a structural problem with a clear catalyst. Subprime defaults hitting maturity and the pandemic shock both fit that pattern. Calls built mainly on “valuation is too high,” without a visible trigger, have usually been weaker on timing.
That fits the framework he has followed for years. He puts little faith in metrics that can be dressed up, such as price-to-earnings ratios and return on equity. He focuses on free cash flow and prefers to go through raw financial materials himself rather than rely on agencies or analyst reports.
That method is strong at answering whether an asset is overvalued or whether hidden stress exists. It is much weaker at telling investors when the market will finally care.
Some of the most useful data in the current debate sits in the background. Nvidia's short interest is about 1.3% to 1.4% of free float, a fairly low level. In dollar terms, though, Nvidia is still one of the biggest short positions in the U.S. equity market. A tally through February this year showed that institutions shorting Nvidia had lost more than $5 billion in aggregate, making it the most painful short at that time. Short positioning had contracted earlier this year, but has since risen again.
Even so, the number of investors willing to short the stock publicly and at scale remains small. Burry is the loudest voice, but he is not alone, and he is far from the consensus.
The three pillars of his argument do not carry equal weight
Burry's latest case rests on three separate claims, and they are not equally persuasive.
Depreciation lives are the strongest part of the case
The first and strongest point concerns depreciation. The accounting logic is basic: when a company buys equipment intended for long-term use, it does not expense the entire cost immediately. Instead, it spreads the cost over the useful life of the asset. The longer that period, the lower the annual expense, and the better reported profit looks.
Burry's complaint is that AI chips evolve so quickly that their real useful life may be only two to three years, while cloud buyers such as Microsoft, Google, and Meta are using depreciation periods of six years. Stretch the life and annual expense falls. Reported profit rises.
Nvidia and some analysts have responded directly, but their response is aimed at customer depreciation policy rather than Nvidia's own books. The dispute will have to be tested over the next few years, when future financial statements show whether meaningful impairments appear.
Off-balance-sheet circular financing is harder to verify
The second point is the off-balance-sheet circular financing claim. The idea is that not all of Nvidia's strong chip demand comes from genuine end-user need. Some of it, under this theory, may reflect a funding loop that creates demand for Nvidia products from within the ecosystem itself.
One version circulating in the market is that Nvidia provides financing support to an AI company expanding computing capacity, that company uses the support to rent cloud services and buy Nvidia chips, and Nvidia then records the resulting revenue and orders as real sales. If the money eventually cycles back to Nvidia and parts of those flows do not need to be fully disclosed in financial statements, the arrangement looks stronger on paper than in economic substance.
Burry's concern is that if those customer-supported purchases were stripped out, Nvidia's true end demand might not be as strong as reported. Nvidia has pushed back by saying its strategic investments in outside companies are very small and account for only a tiny share of revenue. Compared with the depreciation argument, this claim depends more heavily on information that has not been fully disclosed and is therefore harder to test.
The buyback dilution critique is the weakest of the three
The third argument concerns share repurchases and earnings per share. Burry has argued that Nvidia's buybacks in recent years exaggerated EPS performance. Nvidia answered that his calculation included taxes tied to employee stock compensation and used an incorrect basis. Burry did not offer more detailed numbers to rebut that response, which leaves this as the least convincing of the three arguments at this stage.
Eisman and Chanos share some concerns, but not Burry's trade
One of the most notable counterpoints comes from Steve Eisman, another real-life figure featured in The Big Short. Eisman has said he is not planning to short Nvidia for now. His reasons are straightforward: revenue is still growing sharply, buybacks and dividends keep increasing, and major cloud companies are still expanding capital spending commitments. For the moment, he said, those numbers hold up.
Eisman is not fully comfortable, though. He has also said he is nervous about how long the rally can last, has already cut exposure, and warned at the end of July that if any major technology company reduces AI capital spending, U.S. stocks could face a straight-line decline.
That leaves two veterans shaped by 2008, both known for contrarian instincts, on opposite sides of the same AI question.
Jim Chanos takes a more nuanced position. He agrees with Burry in broad terms on the issue of accounting mismatch. Chip suppliers can recognize revenue immediately, while cloud buyers spread massive AI spending over four to seven years. To Chanos, that looks strikingly similar to the setup before the 2001 internet bubble burst. Cisco's numbers remained strong while telecom operators masked overcapacity by extending amortization schedules, and WorldCom went further with outright accounting fraud before the scandal finally surfaced and produced one of the largest bankruptcies in U.S. history.
But Chanos has chosen a very different trade. In his public comments, he has said he is cautious about cyclical bets in memory chips and largely avoids Micron. He is also not shorting Nvidia directly and has even said Nvidia looks more attractive on valuation than some peers. The targets he is actually shorting are private equity firms that are exposed both to AI infrastructure and commercial real estate while operating in a low-cap-rate environment. His bet is on the leveraged financial edge of the AI boom, not on chip stocks themselves.
Put the three veterans side by side and a useful distinction appears. There is meaningful agreement on the broad question of whether there is excess in AI. There is much less agreement on what to short and how to express that view.
What matters more than whether Burry is right this time
Direction and execution are not the same thing.
There is a recurring paradox in cyclical sectors such as memory chips. On paper, they should be among the easiest places to say that prices will eventually come back down. In practice, professional short sellers repeatedly struggle there. Samsung, SK Hynix, and Micron can cut production and support pricing. Irrational enthusiasm in the market can outlast a short seller's staying power. Accounting rules can delay the visible damage for years.
None of those variables can be mapped with precision simply by doing more research.
That is the shared lesson running through Burry, Eisman, and Chanos. Calling direction depends on disciplined work, sensitivity to cash flow, skepticism toward accounting optics, and an ability to identify structural risk. Those are skills that can be learned. Calling the exact break point and the depth of the decline depends on crowd psychology, industry game theory, and random catalysts arriving at the same time. At that stage, the problem starts to look less like analysis and more like probability.
For ordinary investors, the most durable takeaway may be the way these short sellers frame questions. Which reported numbers are most open to cosmetic improvement? Which structural risks are easiest to ignore? When everyone says “this time is different,” what should be checked first?
Whether one specific short position ultimately pays off is not the only thing worth watching in this episode.
The article was first published on Titanium Media App. It was written by Silicon Valley Tech_news and edited by Lin Shen.