Michael Burry, the investor immortalized in “The Big Short” for calling the 2008 housing crisis before basically anyone else, is placing fresh bets against two companies riding the AI infrastructure wave. He has taken short positions targeting Oracle and Nebius Group, signaling deep skepticism about the financial foundations underpinning the AI boom.
His thesis boils down to something deceptively simple: these companies aren’t depreciating their GPUs fast enough, and that’s making their earnings look better than they actually are. In English, imagine buying a car, pretending it holds its value for ten years, and using that fiction to inflate your net worth. Now do that with billions of dollars worth of AI chips.
The GPU depreciation problem
Burry has been vocal about what he sees as a systemic issue across AI hyperscalers. He estimates that Oracle’s earnings could be overstated by approximately 26-27% by 2028 due to undervalued GPU depreciation. That’s not a rounding error. That’s the kind of gap that turns profitable quarters into mediocre ones.
He has taken put options on Oracle shares, after shorting the stock in recent months. Put options give the holder the right to sell shares at a predetermined price, essentially a leveraged bet that the stock will fall.
Burry has pointed to the broader industry’s projection of $176 billion in AI-related capital expenditures across hyperscalers as context for the scale of the problem. When you’re spending that kind of money on hardware that depreciates faster than companies acknowledge, the gap between reported earnings and economic reality can get very wide, very fast.
The Nebius angle
Nebius Group is a less household name than Oracle, but it’s become a significant player in the AI compute space. The company has secured AI compute contracts totaling up to $27 billion with Meta over five years, and also works with Microsoft. That kind of revenue pipeline looks impressive on paper.
Burry’s skepticism toward Nebius appears rooted in thematic concern about the entire AI infrastructure sector. No reliable regulatory filings confirm new short positions against Nebius beyond Burry’s thematic bearishness on AI investments.
A familiar pattern
Burry has made a career out of spotting situations where market consensus and financial reality have quietly diverged. Before 2008, he noticed that mortgage-backed securities were built on loans that could never be repaid.
His hedge fund, Scion Asset Management, was liquidated by late 2025. But Burry hasn’t gone quiet. He continues sharing his market analysis and investment thinking through a Substack publication, maintaining a bearish outlook on AI-related investments.
What this means for investors
The depreciation question Burry raises is genuinely important, regardless of whether his specific trades pay off. Technology companies have significant discretion in choosing useful life estimates for their hardware assets. A GPU depreciated over five years produces very different financial statements than the same GPU depreciated over three years, even though the underlying economics haven’t changed.
Investors holding positions in AI infrastructure names should at minimum understand how their companies account for GPU depreciation and whether those assumptions look conservative or aggressive relative to actual hardware replacement cycles. The difference between a three-year and a five-year depreciation schedule on $176 billion in capex is not trivial, and Burry is betting real money that the market will eventually care.