Burry warns semiconductor rally echoes dot-com bubble final phase
Michael Burry, the investor who predicted the 2008 housing crash, has warned the AI-driven semiconductor rally mirrors the final months of the 1999-2000 dot-com bubble, disclosing put options on Nvidia, Palantir and chip ETFs representing 80 per cent of his hedge fund portfolio.

Michael Burry, the investor who predicted the 2008 housing market collapse, has warned that the artificial-intelligence-driven semiconductor rally carries the hallmarks of the final months of the 1999-2000 dot-com bubble, disclosing put options on Nvidia, Palantir and a broad chip-sector exchange-traded fund that now account for roughly 80 per cent of his hedge fund’s portfolio.
The founder of Scion Asset Management published the warning in a 10 May Substack column, comparing the Philadelphia Semiconductor Index’s trajectory to the late-stage dot-com pattern in which stocks climb “on momentum simply because they have already risen.” He disclosed positions across SOXX puts, QQQ puts, and single-stock puts on Nvidia with a $110 strike, Palantir with $100 and $50 strikes, and Oracle, all expiring between December 2026 and mid-2027.
“It feels similar to the final phase of the 1999-2000 dot-com bubble,” Burry wrote, adding that chip companies are “gaining decades of future earnings in their valuations within days.” He described financial media as consumed by “no end to the discussion of AI” and said “the market is being whipped around by a two-letter theme.”
The SOX index has surged 65 per cent year-to-date in 2026, with the iShares Semiconductor ETF trading roughly 60 per cent above its 200-day moving average, an extreme historical deviation. Nine individual components within the SOXX trade at even more extended levels than the index itself. Intel, which Burry called the “poster child for the top,” has risen 190 per cent this year despite persistent competitive disadvantages in the AI accelerator market where Nvidia commands an estimated 70 to 90 per cent share.
Burry’s bearish thesis follows growing caution among institutional investors about overheated AI-linked equities. On 9 May, NewSquare Capital disclosed it had trimmed a $14.8m position in the First Trust NASDAQ-100 Technology Sector Index Fund after a 60 per cent run, while the S&P 500 has pushed to a fresh record even as earnings growth lags the valuation expansion.
The notional value of the underlying shares behind Burry’s put positions exceeds $1.1bn, according to a TradingKey analysis of Scion’s regulatory filings. The Nvidia puts alone cover one million shares at a $110 strike price, well below the stock’s roughly $215 level at Friday’s close and its all-time high of $217.80. Nvidia’s market capitalisation stands at approximately $5.3tn. Burry’s Palantir short is structured in two tranches: $100 strike puts expiring December 2026 and $50 strike puts expiring June 2027, reflecting his view that the data-analytics company’s fair value sits in the “single digits to low double digits,” far below its recent price around $137.
On Palantir specifically, Burry stated he was “shorting the inherent unsustainability of its business model itself, rather than merely its valuation.” The stock has already declined roughly 34 per cent from its 52-week high of $207.
The long side: software stocks “scared away” by AI
Burry has paired the semiconductor shorts with long positions in traditional enterprise software companies he argues have been unfairly discounted by AI enthusiasm. His disclosed holdings include Adobe, Autodesk, Salesforce, and Veeva Systems, firms he described as “AI victims” whose share-price declines “stemmed from panic selling rather than deteriorating fundamentals.”
The strategy amounts to a paired trade: short the AI beneficiaries riding what Burry sees as a speculative wave, long the incumbents whose businesses he believes will prove durable once the capital-allocation cycle turns. It is a concentrated bet. The short side alone represents roughly 80 per cent of Scion’s reported portfolio, with the remaining 9.5 per cent split across QQQ, SOXX, and Oracle puts.
A mixed record on timing
Burry’s 2008 housing bet, immortalised in the film “The Big Short,” produced one of the most profitable trades in hedge-fund history. His subsequent record has been less consistent. He called Tesla a bubble in 2021, drawing public mockery from Elon Musk, and exited the position before the electric-vehicle maker’s stock continued its four-year rally. He shut down Scion Asset Management entirely in late 2025 before re-emerging with the current AI-short portfolio.
A Hankyung report noted that Burry’s post-2008 bearish calls have “missed the mark” more often than not, though the scale and concentration of the current semiconductor bet distinguish it from his earlier, more diversified wagers. The options structure also gives him a defined timeline: if the chip sector does not correct materially by January 2027, the puts expire worthless and the trade is a total loss on the premium paid.
What the bulls say
The bull case rests on the sheer scale of committed capital expenditure. Microsoft, Google, Amazon and Meta have together pledged hundreds of billions of dollars toward AI data centre infrastructure through 2027, creating a demand pipeline for Nvidia’s H200 and Blackwell chips that is neither speculative nor deferred. The Philadelphia Semiconductor Index’s 65 per cent gain this year, on this view, reflects the acceleration of a real capital cycle rather than speculative multiple expansion of the sort seen during the dot-com era.
Market pricing also embeds an assumption that the Federal Reserve will deliver minimal rate cuts in 2026, a posture the central bank has signalled it is comfortable maintaining while inflation remains above target and the labour market stays tight. Higher-for-longer rates have historically acted as a headwind for growth-stock valuations, yet semiconductor shares have rallied through the rate environment regardless.
The precedent
The dot-com comparison is not casual. Between March 2000 and October 2002, the Nasdaq Composite fell 78 per cent, wiping trillions in market value. The Philadelphia Semiconductor Index, which peaked alongside the broader tech complex, took more than 17 years to reclaim its 2000 high. Burry’s argument is not that today’s chip companies are fraudulent or unprofitable (unlike many dot-com era firms, Nvidia, Broadcom and TSMC generate enormous free cash flow) but that valuations have compressed decades of future earnings into a single year’s price action.
The SOX index recorded its strongest 25-day rolling performance since 9 March 2000 on 6 May, and inflows into semiconductor ETFs continued through the first week of May. Burry’s options expire in January 2027, giving the trade roughly eight months to play out. The consumer sentiment backdrop is deteriorating: the University of Michigan index fell to a record low in early May, even as the S&P 500 traded at an all-time high, a divergence that echoed the late-1999 pattern Burry is betting will resolve in his favour.
Marcus Holloway
Markets editor covering UK gilts, sterling and the Bank of England. Previously a fixed-income strategist in the City.


