Michael Burry built his reputation on a single, spectacular bet: shorting the U.S. housing market before the 2008 collapse, a trade that earned his small fund over $700 million and became the centerpiece of Michael Lewis’s book The Big Short. Now the hedge fund manager is staking out another contrarian position, this time against the artificial intelligence rally that has dominated Wall Street for the past two years. In a post on X that he later deleted, Burry wrote that the AI market “feels like the last months of the 1999-2000 bubble.” His firm’s federal filings show he is backing that view with real money.
Burry’s public warnings and how he’s positioned
Burry, who runs Scion Asset Management from Saratoga, California, has a habit of posting pointed market commentary on social media and then scrubbing it within hours. His AI bubble comparison was captured and widely reported by outlets including Reuters and Bloomberg before he took it down. The pattern is familiar to anyone who has followed him: a blunt warning, a burst of attention, then silence.
But the warning is not just talk. Scion’s most recent Form 13F-HR, filed with the Securities and Exchange Commission and available through the agency’s EDGAR database, lists put options on several publicly traded companies alongside the fund’s long equity holdings. The filing’s information table details each position by issuer name, CUSIP number, share or contract count, and option type. Put options gain value when stock prices fall, so their presence is a documented signal that Scion is positioned for declines in at least some of its holdings.
The 13F does not organize positions by investment thesis. It does not say “short AI.” But several of the companies where Scion holds puts rank among the largest beneficiaries of the AI boom, a connection that analysts have noted in reviewing the disclosure. Scion is a relatively small fund, so the positions are modest by institutional standards. Still, the directional bet is clear.
The dot-com parallel: what happened in 1999-2000
Burry’s comparison points to a specific and well-documented period. Between roughly March 1999 and March 2000, the Nasdaq Composite more than doubled. The rally was overwhelmingly concentrated in the index’s largest names, most of them technology and internet companies riding a wave of speculative enthusiasm. According to multiple financial retrospectives on the bubble period, the top 10 Nasdaq stocks posted cumulative gains of approximately 784% over that 12-month stretch, dwarfing the performance of the broader index.
What followed was severe. The Nasdaq peaked at 5,048.62 on March 10, 2000, according to historical index data, then lost nearly 80% of its value over the next two and a half years. Cisco Systems, which traded at more than 150 times earnings near its peak, saw its stock price collapse by roughly 86%. Dozens of smaller dot-com firms went bankrupt entirely. The crash wiped out an estimated $5 trillion in market value.
The Federal Reserve Bank of St. Louis maintains a data series called the Nasdaq-100 Top 30 Total Return Index, sourced from Nasdaq Global Indexes and available through the FRED portal. While it tracks the top 30 components rather than the top 10, the series provides a government-hosted benchmark for measuring how much of the Nasdaq-100’s performance has been driven by its biggest names. That concentration pattern is central to Burry’s argument: when a small number of stocks account for a disproportionate share of an index’s gains, the index becomes vulnerable to sharp reversals if sentiment shifts.
Market concentration in 2026 echoes the late 1990s
The structural similarities are hard to dismiss. Through the first half of 2026, a small group of mega-cap technology companies, often referred to as the “Magnificent Seven,” continues to drive the bulk of the S&P 500’s and Nasdaq-100’s returns. Nvidia, whose chips power most AI training and inference workloads, saw its market capitalization surge past $3 trillion during 2024 and has remained among the world’s most valuable companies. Microsoft, Amazon, Alphabet, Apple, Meta Platforms, and Tesla have similarly commanded outsized influence on index-level performance.
According to S&P Dow Jones Indices data, the top 10 stocks in the S&P 500 accounted for more than 35% of the index’s total market capitalization by late 2024, a level of concentration not seen since the dot-com era. That figure has remained elevated into 2026. The Nasdaq-100 is even more top-heavy. When a handful of names drive the headline numbers, broad index gains can mask weakness across the rest of the market, a dynamic that was also present in 1999.
Valuations in the AI sector have drawn sustained scrutiny. Nvidia traded at roughly 60 to 70 times trailing earnings through much of 2024 and 2025, while some smaller AI-adjacent companies carried even steeper multiples. Bulls argue that AI revenue growth, which has been substantial for companies like Nvidia and Microsoft, justifies premium pricing. Skeptics counter that the market is pricing in adoption curves and profit margins that may take years to materialize, if they materialize at all. Burry falls squarely in the skeptic camp.
Burry’s track record: right on housing, uneven since
The 2008 call remains one of the most celebrated trades in modern finance. Burry identified the fragility of subprime mortgage-backed securities years before the broader market caught on, and his credit default swaps paid off spectacularly when the housing market collapsed. The trade earned him a permanent place in financial history and a level of public credibility that few hedge fund managers enjoy.
His subsequent public calls have been more uneven. In January 2023, Burry posted a single word on Twitter: “Sell.” The S&P 500 went on to rally more than 24% that year, according to index data. He has also flagged concerns about passive investing, index fund concentration, and water scarcity, positions that reflect long-term structural thinking but have not always translated into near-term market moves. In mid-2023, his 13F showed he had reversed course and taken long positions in several tech names, only to later rotate back toward bearish bets.
That mixed record matters. Being early and being wrong can look identical for extended periods, something Burry himself experienced during the housing trade when his fund’s investors pressured him to abandon the position before it paid off. His current AI skepticism could prove similarly premature. Or it could be another instance of spotting a structural vulnerability before the crowd does.
How AI revenue complicates the dot-com comparison
Several things are documented and verifiable. Burry’s fund holds put options on prominent stocks, confirmed by federal filings that anyone can review on EDGAR. Market concentration in a small number of technology names has reached levels that echo the late 1990s, supported by index data from both government and private sources. And Burry has publicly compared the current environment to the final phase of the dot-com bubble, a comparison reported by major financial news organizations.
What the evidence does not support is a guaranteed timeline or outcome. The dot-com bubble took years to fully inflate, and the Nasdaq continued rising for months after early skeptics raised alarms. Burry’s 13F is a quarterly snapshot, not a real-time trading diary; his positions may have changed since the filing date.
And this is where the AI boom diverges most sharply from the dot-com era. The dot-com period was defined in large part by companies with business plans but no revenue. The current AI cycle, by contrast, is generating substantial actual revenue for companies like Nvidia, Microsoft, and Alphabet. Corporate capital expenditure on AI infrastructure has been significant, producing real products, real efficiency gains, and real earnings growth at the companies building the underlying technology. The question is not whether AI is real. It is whether the stocks pricing in AI’s future have gotten ahead of what that future will actually deliver, and whether the concentration of gains in a handful of names has created the same kind of fragility that made the Nasdaq’s March 2000 peak the starting line for a two-and-a-half-year collapse.
