Michael Burry, the investor who made his name predicting the 2008 housing collapse, has found his next short: Nvidia Corp. (NVDA). And he's not being subtle about it.
The Nvidia Math Problem
In a Substack post over the weekend, Burry laid out why he's betting against the AI chip darling. His argument is simple but striking: Nvidia is "simply the purest play" on artificial intelligence, which makes it dangerously exposed when the numbers don't add up.
"I do not see how that math works," Burry wrote, pointing out that Nvidia could sell roughly $400 billion worth of chips this year while actual revenue-generating AI applications are generating "less than $100 billion in application layer use cases."
That's a pretty significant gap. Companies are spending wildly on AI infrastructure, but the actual money-making applications haven't caught up. Burry sees Nvidia as "the most loved, and least doubted" stock in the AI trade, which in his view creates an attractive short opportunity because sky-high expectations leave almost no room for disappointment.
Why Other Tech Giants Get a Pass
Burry also explained why he's not shorting other AI-heavy companies like Meta Platforms Inc. (META), Alphabet Inc. (GOOG) (GOOGL), or Microsoft Corp. (MSFT). The answer comes down to business fundamentals.
Betting against Meta would mean shorting "its social media/advertising dominance," he wrote. Going after Microsoft would be "shorting a global office productivity SaaS goliath." These companies have massive, profitable core businesses that could cushion the blow if AI spending doesn't pan out.
They're not "pure shorts on AI," Burry said. They can afford to lose money on artificial intelligence experiments because their cash-generating businesses will keep humming along. That's not the case with Nvidia, which is far more dependent on continued hyperscaler spending.
Burry also flagged another risk at Nvidia: technological obsolescence. The company releases new chip solutions every year or less, which raises the possibility of future writedowns as older inventory loses value.




