Michael Burry and Steve Eisman both gained notoriety for betting against the US housing market, but they now disagree on the future of AI trading.
Burry’s bearish posts often target the unbridled enthusiasm for AI. Earlier this year, he announced bets against some of the sector’s leading names and predicted that the AI boom will end in disaster.
Burry has zeroed in on a specific part of the AI story: the booming demand for high-tech chips to run and train models. The data center explosion has been a boon for hardware makers like Nvidia, but Burry argues that depreciating chip values are a major problem for companies.
Earlier this year, he predicted that the top hyperscalers would “understate depreciation” by more than $175 billion between 2026 and 2028. Famous short-seller Jim Chanos has voiced similar concerns about chip costs and returns on capex spending.
But on an episode of the New Money podcast, fellow “Big Short” alum Steve Eisman hit back at Burry’s views.
“With all due respect to Michael, I think his argument is too academic,” Eisman stated. “If AI succeeds because Anthropic and OpenAI grow like crazy, and the hyperscalers do well, it won’t matter whether the depreciation schedule changed from three or four years to five or six years.”
Burry has shared this take on AI and the broader market over the past year on his Substack, Cassandra Unchained, outlining what he sees as the risks for investors due to hyperscalers overstating profits and under-reporting depreciating hardware.
Eisman noted that he understands Burry’s argument when it comes to AI chip depreciation schedules, an accounting tactic that made them appear significantly more profitable on paper. While Eisman implied that he agrees with Burry on the potential impact of chip depreciation, he said it’s not a major problem right now.
“Where I think he is wrong, for the moment, is that there is such demand for chips right now that there is still huge demand for the older chips,” Eisman said.
Recent news from Silicon Valley has challenged Burry’s thesis that AI chips will rapidly lose value. In August, CoreWeave announced a contract to rent A100 GPUs made by Nvidia, a type of chip that debuted in 2020, through 2029.
Eisman may have hit back at Burry’s thesis, but he’s not universally bullish on the AI trade. He’s issued his own warnings recently, primarily focused on the market’s reliance on OpenAI and Anthropic, which he described as the “Achilles’ heel” of the AI trade.
“If OpenAI fails and the whole chain goes in reverse, we’ll have a massive correction that has nothing to do with the depreciation schedule,” Eisman said.