Author: Claude, DeepChain TechFlow
DeepChaotrend Summary: As the Nasdaq continues to hit new all-time highs and NVIDIA's market capitalization approaches $5.3 trillion, Michael Burry—the investor who rose to fame during the 2008 financial crisis by shorting subprime mortgages and the real-life inspiration for the movie "The Big Short"—is now betting the opposite.
He not only maintained his short positions on NVIDIA and Palantir but also expanded his shorting to semiconductor ETFs and the Nasdaq ETF, while buying traditional software stocks that have been pressured by the AI narrative, constructing a comprehensive portfolio for the "AI bubble repricing."
The Nasdaq Index hit a new all-time high for the week, closing at approximately 26,247 on May 8, while the S&P 500 also reached a record level on the same day. Since the second quarter, the Philadelphia Semiconductor Index has risen by about 55%. NVIDIA’s stock price is nearing its historical high of $217.80, with its market capitalization surpassing $5.2 trillion. The AI-driven tech stock rally is currently at its most intense phase.
But at the height of market euphoria, an investor known for betting against the trend is significantly increasing his position in another direction.
According to Foreign Policy Journal on May 7, Michael Burry, the hedge fund manager whose prediction of the 2008 subprime mortgage crisis was adapted into the film "The Big Short," disclosed his latest portfolio adjustments this week on his Substack column, "Cassandra Unchained":
He not only maintained his put options on NVIDIA and Palantir but also added a direct short position on Palantir and increased his put bets on the semiconductor ETF (SOXX), the Nasdaq-100 ETF (QQQ), and Oracle.
Meanwhile, he began buying a group of traditional software companies that had been sidelined by the AI boom, such as Adobe, Autodesk, Salesforce, and Veeva Systems, arguing that their stock price declines were due to panic selling rather than deteriorating fundamentals.
At this point, a complete short hedge position has emerged, with the core logic being to short AI-benefiting stocks and go long on AI-harming stocks.
Starting from the $1.1 billion bet in November last year
Burry's short position on the AI sector began in the third quarter of 2025.
At the time, its hedge fund Scion Asset Management’s 13F filing revealed that he purchased put options with a notional value of approximately $912 million in Palantir and $187 million in NVIDIA. The announcement in November last year triggered market turbulence, causing temporary downward pressure on the stock prices of Palantir and NVIDIA.
However, Burry later clarified on X that his actual investment was approximately $9.2 million, not the $9.12 billion widely reported by the media—the latter being the notional value of the options contracts, a difference of nearly a hundredfold. This detail is crucial: the notional value listed in 13F filings is often misinterpreted as actual capital invested, thereby inflating the perceived scale of the trades.
Shortly after the disclosure, Burry announced the closure of Scion Asset Management and the cancellation of its SEC registration, ending his career of managing external funds.
He later transitioned into a private investor and launched a Substack column under the name "Cassandra Unchained" (Cassandra being the Greek mythological prophetess who foretold truths no one believed), consistently publishing market analysis.
Shorting Palantir has paid off; Burry says "it hasn't fallen enough yet."
Based on the trading results, Burry’s bet on Palantir is currently in profit. Palantir’s stock price has declined from around $161 when he entered to approximately $137 today, a cumulative drop of about 34% from its 52-week high of $207. Despite the company recently reporting strong first-quarter fiscal 2026 earnings—with revenue up 85% year-over-year—the stock price fell after the earnings announcement.
Burry did not close his position. According to his Substack post, he currently holds put options expiring in December 2026 with a strike price of $100 and put options expiring in June 2027 with a strike price of $50, indicating he expects Palantir to decline more than 60% from its current level within the next year. He explicitly stated in his post that Palantir’s fair valuation is only in the “low single digits to low double digits.”
In April this year, Burry posted on Substack that Anthropic was “eating Palantir’s lunch,” noting that the AI safety company’s revenue growth had surpassed a $30 billion annualized rate, and its more user-friendly, lower-cost AI integration tools were replacing Palantir’s complex enterprise deployments. After the post was published, Palantir’s stock price dropped 13.7% within a week, after which Burry deleted the post. Wedbush analyst Dan Ives dismissed the claim as a “fictional narrative,” and Palantir CEO Alex Karp had previously publicly stated he could not understand Burry’s short position.
NVIDIA short positions are still losing money, but Burry stands by his view that "AI is a bubble."
In contrast to Palantir’s success, Burry’s position in NVIDIA is entirely different.
NVIDIA's stock closed at approximately $215 on May 8, nearing its all-time high of $217.80, with a market capitalization of about $5.3 trillion. According to reports, Burry holds NVIDIA put options with a strike price of $110, expiring in December 2027, which are currently deeply out of the money. However, he has not reduced his position; instead, he increased his stake in recent portfolio adjustments.
Burry’s core rationale for shorting NVIDIA is “overbuilding of AI infrastructure.” In his first Substack article last November, he drew a parallel between the current AI investment boom and the late 1990s internet bubble, comparing NVIDIA to Cisco of that era. Cisco’s stock rose 3,800% between 1995 and 2000, briefly becoming the world’s most valuable company, before plunging more than 80% following the dot-com bubble burst.
Burry’s core arguments include: hyperscale customers such as Microsoft, Google, Meta, Amazon, and Oracle are extending the depreciation periods for GPUs to improve their financial statements; he estimates that these accounting practices will collectively understate approximately $176 billion in depreciation expenses between 2026 and 2028, artificially inflating industry profits. Additionally, he believes that the current massive capital expenditures on AI infrastructure are based on overly optimistic demand forecasts, mirroring the frenzy with which telecom companies laid fiber optic cables around 2000.
This perspective prompted a direct rebuttal from NVIDIA. According to CNBC, NVIDIA privately distributed a seven-page memo to Wall Street sell-side analysts, addressing Burry’s claims point by point and specifically citing his X platform posts as sources requiring correction. In the memo, NVIDIA stated that its customers set GPU depreciation periods at four to six years based on actual useful life, and that earlier products, such as the A100 released in 2020, continue to maintain high utilization rates. Burry responded by saying, “I’m not saying NVIDIA is Enron,” but stood by his analysis.
Go long on software stocks suppressed by AI: a complete bubble hedging portfolio
Perhaps the most noteworthy aspect of Burry’s position adjustments is not the shorting itself, but his long positions.
He recently purchased stocks such as Adobe, Autodesk, Salesforce, Veeva Systems, and MSCI. These companies share a common trait: their business fundamentals remain solid, but their stock prices have declined sharply due to market narratives about being disrupted by AI and forced selling by private credit funds.
Adobe is currently down approximately 30% from its 52-week high, and Autodesk is down about 22% year-to-date; both companies' forward P/E ratios have retreated to levels seen between 2018 and 2019.
Burry explained on Substack that he "does not believe the technical selling pressure from private credit and software debt is sufficient to negatively impact these stocks in the long term." In other words, he believes the market has overly punished companies labeled as "AI losers" while overhyping those labeled as "AI winners"—and he is betting on a correction of this mispricing.
Looking at both the short and long positions together, Burry has constructed a classic long-short hedge: if the AI bubble narrative bursts, high-valuation beneficiaries like NVIDIA and Palantir will be hit first, while undervalued traditional software stocks may see a valuation rebound. Even in a broadly declining market, this structure could generate positive returns.
In his letter to investors when closing Scion, Burry admitted: “My assessment of security values has been out of sync with the market for a long time.” This statement is both an introspection and a hallmark of his一贯 stance.
At the height of the AI craze, he chose to stand opposite the crowd.