‘Big Short’ investor Steve Eisman warned that falling AI prices, moderating usage and data-center obstacles could expose weaknesses across the stock market, where AI has become a dominant investment theme.
Eisman discussed the risks during a podcast, Behind the Business with Michelle Toh, last week. Eisman said OpenAI recently lowered the price of its latest model after computing capacity increased substantially. That move could put pressure on Anthropic to reduce prices as well, intensifying competition among leading AI developers.
He also pointed to the retreat from "token maxing," a practice in the recent past wherein companies encouraged employees to use AI extensively regardless of expense. Eisman said businesses are increasingly focused on controlling costs and evaluating whether usage produces measurable value.
Eisman said financial disclosures from OpenAI and Anthropic could become more transparent if either company goes public. He suggested Anthropic might pursue a listing before weaker growth figures potentially affect investor interest.
The investor said AI influences more than half the stock market when technology companies outside the information technology sector are included. Consumer staples, health care and real estate are the least connected sectors, while utilities, banks and industrial companies remain linked through power, financing and construction.
Eisman warned that an AI downturn could extend beyond a recession and trigger a "massive correction" in stocks. He cited pricing pressure, slower token usage and permitting challenges affecting data-center construction.
AI Spending Risks Could Trigger Recession
Speaking about the economy, Eisman argued that the AI boom is built on a highly concentrated chain of spending, making it vulnerable to financial trouble at OpenAI or Anthropic.
Citing Nvidia Corp. (NASDAQ:NVDA) as an example, the company stated that its revenue growth of more than 100% masks customer concentration, with 70% of its accounts receivable coming from just five customers, according to its 10-Q filing. Meanwhile, OpenAI and Anthropic reportedly account for a significant share of hyperscalers’ AI revenue, driving their massive chip purchases.
He warned that financial trouble at either AI startup could disrupt the entire chain, hurting hyperscalers, Nvidia and companies such as Oracle Corp. (NYSE:ORCL), which reportedly has roughly half of its $600 billion backlog tied to OpenAI, thus cautioning that a reversal in AI spending could have broader economic consequences, potentially pushing the U.S. into a “recession immediately.”
AI CapEx Faces a Reality Check
Last week, Apollo Global Management Chief Economist Torsten Slok said that AI’s financial benefits remain concentrated among large technology companies, with little evidence of broader productivity gains across corporate America. Slok found that technology companies’ revenue per employee rose from roughly $550,000 in 2022 to nearly $800,000, while the figure remained largely unchanged for the rest of the S&P 500. Tech profit margins also expanded, but margins across other sectors stayed near 10%.
Meanwhile, Yorkville America CEO Steve Neamtz told Benzinga that investors are increasingly scrutinizing whether surging AI capital expenditures are translating into measurable revenue growth and returns. He said demand, financing, power availability, customer commitments and capacity utilization will determine the sustainability of infrastructure spending.
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
Photo: Hryshchyshen Serhii / Shutterstock
© 2026 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.
