A recent trading disclosure has once again thrust Michael Burry into the spotlight.
This fund manager, immortalized in the movie *The Big Short* for his prescient bet against subprime mortgages, disclosed a list of short positions via his Substack newsletter, "Cassandra Unchained," on June 30. The list included Nvidia (entry price $198.09), Tesla, Applied Materials, construction machinery giant Caterpillar, and the Philadelphia Semiconductor ETF (SOXX). On July 1, he initiated a new short position on Micron Technology (entry price $1051.87). On July 25, Burry announced he had further increased his short positions on Nvidia, Micron, and SOXX.
His rationale is not new. He posits two major issues in AI infrastructure build-out: "off-balance sheet circular financing" and "inflated depreciation schedules." The way Nvidia's clients handle their finances may make the entire supply chain's paper profits appear healthier than they actually are.
The reaction on social media was almost a conditioned reflex: just another round of bearish talk.
Nvidia's stock price experienced significant volatility following the disclosures. On June 30, when Burry first disclosed his position, Nvidia closed around $200, roughly in line with his $198.09 entry price. In early July, influenced by factors such as Meta's announcement of leasing idle AI compute capacity, the Philadelphia Semiconductor Index fell over 6% in a single day, and Nvidia briefly dipped near $194 before quickly rebounding, climbing back above $210 by mid-July.
After Burry's second increase on July 25, Nvidia dropped nearly 5% to $196.51 on July 27. However, the direct driver of this decline was a report published that day claiming Nvidia was discussing providing OpenAI with approximately $250 billion in financing guarantees. Burry promptly commented on social media, equating it to "Nvidia guaranteeing its own customers' chip purchases," echoing his "off-balance sheet circular financing" argument and amplifying market concerns.
Thereafter, Nvidia traded within a range of $190 to $200. It touched a low of $190 on July 29 before quickly recovering, closing at $200.75 on July 31. Based on these levels, Burry's initial June 30 entry at $198.09 is essentially flat, with a slight paper loss; the July 25 addition at $210.28 is sitting on a paper loss of about 4.5%.
This is almost Burry's fixed script. Over the past few years, he has frequently sounded alarms on the market, yet his hit rate has repeatedly been questioned. This leads to a more pertinent question: When a "professional pessimist" speaks again, how credible is his judgment? Furthermore, within Wall Street as a whole, are those shorting these companies a few individuals making a desperate bet, or part of a tacit, collective retreat?
The Other Side of the Legend: After Deification
Burry's legendary triumph came in 2008. Before the subprime crisis erupted, he meticulously combed through thousands of mortgage contracts and discovered that vast amounts of securities labeled AAA by rating agencies were actually built on borrowers who couldn't repay. This short bet made him famous overnight and cemented the moniker "The Big Short."
What is rarely mentioned, however, is that his track record after that victory has been less than stellar.
In late 2020, he began building a short position against Tesla. In the following months, the stock price rose instead of falling. By the end of Q1 2021, the disclosed size of his position was interpreted by the market as a multi-billion dollar bet (though this often reflected the notional value of options, with the actual premium paid being far less). This short position disappeared from his holdings by Q3 2021, with multiple reports concluding he likely closed it at a loss.
With GameStop, he cleared his position weeks before the explosive squeeze, missing the most violent part of the rally; years later, he bought back in but then actively sold out again due to concerns over the debt risk and valuation logic of its acquisition of eBay.
On the Nvidia and Palantir front, he has submitted two different report cards. In early 2023, he posted a one-word tweet: "Sell." This was a generic market warning. Subsequently, the market staged a strong rebound, and by late March, Burry himself acknowledged the misjudgment. His direct targeting of these two specific stocks didn't come until November 2025, when he disclosed that about 80% of his portfolio was in put options on Nvidia and Palantir. This move was immediately countered by Palantir's better-than-expected earnings, which caused only a brief dip before the stock quickly recovered and hit new highs.
By the first half of 2026, the fates of these two short bets diverged again. Nvidia's stock price continued to surge, its market cap once surpassing $5 trillion, leaving his short position in the red. Palantir, however, dropped over 10% in a single week after Burry publicly questioned whether its AI business was being eroded by competitors.
Of course, his report card isn't all misses. He warned as early as 2019 that indiscriminate passive investment flows into large-cap stocks were distorting prices; he disclosed to media that he had shorted and profited from the market before the March 2020 pandemic crash; he also mentioned correctly calling the end of the meme stock mania in mid-2021.
Many Are Bearish, But Few Dare Bet Big
Examining his hits and misses reveals a clear dividing line. The times he was right invariably involved a "structural problem" coupled with a "specific catalyst"—the wave of subprime mortgage defaults, the pandemic black swan. Judgments based solely on "overvaluation" without a concrete trigger have almost always been wrong on timing.
Underpinning this is his consistent analytical framework: distrusting easily manipulated metrics like P/E ratios and ROE, trusting only free cash flow; disbelieving rating agencies and analyst reports, preferring to dig into the original financial documents like reviewing medical records.
This methodology excels at answering "Is the asset overvalued? Are there hidden risks?" but fails at answering "When will the risk materialize?" This is precisely the key premise for understanding his current short play.
First, consider some easily overlooked data: Nvidia's current short interest sits at around 1.3% to 1.4% of its float, a relatively low level (though, when converted to notional value, Nvidia is also one of the most-shorted stocks by absolute size in the US market, with losses for short-sellers not being trivial). A February report indicated that short sellers of Nvidia had collectively lost over $5 billion, making it the stock causing the most significant losses for short-sellers at that time. It's worth noting, however, that short interest, after shrinking earlier this year, has recently climbed again.
Those bold enough to publicly, significantly short these stocks are an extreme minority. Burry may have the loudest voice, but he is far from the only one and is decidedly not mainstream.
Let's examine the three arguments he presented this time, which vary in persuasiveness.
First, and arguably the most substantive: Depreciation Schedules.
This involves basic accounting logic. When a company buys equipment for long-term use, it cannot expense the entire cost immediately. Instead, it must amortize the cost over the asset's "useful life" across multiple future accounting periods, deducting only a portion each year, thereby making profits on paper appear healthier.
Burry's contention is that for hardware like AI chips, which iterate extremely fast, the actual useful life may be only 2-3 years. However, cloud providers like Microsoft, Google, and Meta, which purchase these chips, are stretching the depreciation schedule to 6 years. The longer the schedule, the lower the annual amortized cost, artificially inflating paper profits.
Nvidia and some analysts have responded, though it's important to note their response addresses the depreciation policies of *downstream customers*, not Nvidia itself. This debate will ultimately be settled by future financial reports (whether large impairments occur).
Second: Off-Balance Sheet Circular Financing.
This argument suggests that the robust chip orders Nvidia is currently seeing may not all stem from genuine end-user demand. There might exist a funding cycle that "self-creates demand."
One specific version circulating in the market posits: Nvidia provides financing guarantees for an AI company's compute expansion; that company uses the guaranteed funds to lease cloud services and purchase Nvidia chips; Nvidia then books this as "real" revenue and orders. The money, after circulating, ultimately flows back to Nvidia itself, with many of the intermediate funding flows not fully disclosed on financial statements—hence "off-balance sheet."
Burry's concern is that if such cycles of "funding customers to buy your own products" are stripped away, Nvidia's true end-user demand might not be as dazzling as the reported numbers suggest. Nvidia counters this by stating its external strategic investments are minuscule, representing a tiny fraction of revenue, and that the allegation lacks basis. This argument is even more reliant on information not fully disclosed and is harder to verify than the first.
Third: Share Buyback Dilution—the weakest argument.
Burry alleged that Nvidia's share buybacks in recent years have inflated its per-share earnings performance. However, Nvidia subsequently pointed out that Burry's calculation incorrectly included tax expenses related to employee stock-based compensation, a clear error in methodology. Burry has not provided more detailed data to refute this, making this currently the least convincing of the three points.
It is noteworthy that holding the opposite stance is another *Big Short* protagonist, Steve Eisman. Eisman stated he has no immediate plans to short Nvidia, citing reasons including: revenue growth remains strong, buybacks and dividends are increasing, and major cloud providers' capex commitments continue to expand—numbers that currently hold up to scrutiny. However, he also admitted feeling uneasy about the sustainability of this rally, has actively reduced his holdings, and warned in late July that if any major player cuts AI capital expenditure, the US stock market risks a sharp decline.
Two veterans, both shaped by 2008 and renowned for their contrarian thinking, have arrived at diametrically opposed answers to the AI question.
The stance of legendary short-seller Jim Chanos is even more nuanced. He agrees with Burry's general direction regarding the "accounting mismatch"—chip equipment makers booking revenue immediately, while cloud providers buying the chips amortize massive expenditures over 4-7 years. This, he notes, bears a striking resemblance to the period leading up to the 2001 dot-com bubble. Back then, Cisco's reports looked strong, but telecom carriers were using extended amortization schedules to mask overcapacity, and WorldCom engaged in outright fraud. The problems were only revealed en masse when scandals broke, leading to one of the largest bankruptcies in US history.
But in terms of actual positioning, Chanos has chosen a completely different path. According to his public statements, he is cautious about the cyclical play in memory chips and largely avoids Micron. He also does not directly short Nvidia, believing it holds more valuation appeal than peers. His actual short bets are placed on private equity firms simultaneously exposed to both AI infrastructure and commercial real estate in a low-cap rate environment—targeting the financial leverage at the periphery of the AI frenzy, not the chip stocks themselves.
Placing these three viewpoints side-by-side reveals a considerable consensus among seasoned short-sellers on the macro judgment "Is there an AI bubble?" But on "*Whom* to short specifically and *how* to position," the divergence is significant—almost everyone is fighting their own battle.
More Important Than "Is He Right This Time?"
Direction and execution have always been two different things.
A notable paradox is that for highly cyclical industries like memory chips—theoretically the easiest to judge as "destined for a correction"—professional short-sellers have repeatedly suffered setbacks in reality.
The reasons aren't hard to understand. Oligopolies like Samsung, SK Hynix, and Micron can collude to cut production at any time, artificially controlling price cycles; market irrational exuberance can outlast the financial endurance of short-sellers; and accounting rules themselves can delay the full exposure of real problems for years. None of these variables can be accurately predicted merely by "thorough research."
This pattern underscores the common thread among Burry, Eisman, and Chanos. Identifying the *direction* relies on solid research methodology, sensitivity to cash flow, vigilance on accounting practices, and a nose for structural risk—these are skills that can be learned and replicated. But pinpointing the exact *timing* of a decline, or its *depth*, essentially depends on whether irrational sentiment, industry博弈, and random catalysts align—a realm that transcends analytical capability and more closely resembles a game of probability.
For the average investor, the truly valuable takeaway is *how* these individuals pose questions: which financial figures are easily manipulated, which structural risks are easily overlooked, what to investigate when everyone chants "this time is different." Replicating a specific short position or betting on whether a particular bearish call will materialize is not the point.
This, perhaps, is what's truly worth retaining from this latest round of short-selling controversy.
(This article was first published on Titanium Media APP. Author: Silicon Valley Tech_news. Editor: Lin Shen)
