Steve Eisman has issued a stark warning that markets will go 'straight down' if any hyperscaler cuts its capital expenditure, as reported by CNBC. During his latest interview, Eisman emphasized that "it all boils down to, in a sense, Nvidia" when it comes to market response to potential capex reductions. He stated that "the market will go straight down" if any frontier model developers face funding constraints, questioning whether these companies have moats around their businesses. This warning comes as the July meeting of the Federal Open Market Committee looms later this week, adding to market uncertainty about AI investment sustainability.
Steve Eisman has sold his long-held Google position to reduce artificial intelligence exposure, according to reports from CNBC's Squawk Box. The former Neuberger Berman portfolio manager, who gained fame for shorting subprime mortgages at FrontPoint Partners, made the deliberate exit from his Google position after owning it for an extended period. Eisman explained that he wanted to reduce his exposure to AI-related investments, stating he has not purchased a replacement for his Google position. Speaking to CNBC on July 27, 2026, Eisman confirmed he sold the position "a couple of months ago" specifically to dial back his exposure to the artificial intelligence trade, framing the sale as risk reduction rather than a bearish call on Alphabet's business.
The timing of Eisman's exit appears strategic, as Alphabet peaked at $408.61 on May 18 before closing at $319.74 on July 24, representing a roughly 20% decline in approximately two months. According to reports, Alphabet fell 7.1% on July 23 following Q2 earnings, when the company raised 2026 capital spending guidance to a range of $195 billion to $205 billion. Eisman noted that he is currently holding cash and does not expect the AI debate to settle within the next two weeks. The investor highlighted how the AI industry has become a lot more capital intensive now, with the increasing presence of cheaper AI models from China adding complexity and the possibility of a price war looming. He also pointed to Alphabet's capex increase from $190 billion to $205 billion as a trigger for recent stock declines despite better-than-expected Q2 results.
Eisman warns that the entire market has become concentrated in AI-related investments, as reported by CNBC. He argues that even investors who believe they are diversified with 60% stocks and 40% bonds are actually not diversified, as more than 50% of their portfolio is tech and AI-related. The investor highlighted the concentration of the tech and AI trade across equities as well as bonds, noting that of the 40% in bonds, most of the new issuance is AI related. Even investors who bought stocks like Goldman Sachs wouldn't be diversifying beyond the ongoing AI trade, given that investment banks are undertaking AI financing. The Bank of England data supports this concern, showing that five AI hyperscalers held just 3% of outstanding US investment-grade debt at the end of 2025 but accounted for over 15% of this year's issuance by early May. High Technology made up 14.2% of US corporate bond issuance in Q2, with Amazon pricing $37 billion of notes on March 10 as the largest deal.
When asked about potential market corrections if AI fails commercially, Eisman stated he believes there would be a big correction, as reported by CNBC. He emphasized the concentration risk, stating "it's all one trade. So it better succeed." The investor noted that the overall sentiment a year ago was great, with companies spending money more freely on AI than they are now, but the current environment is much more complicated. Eisman warned that the equity markets will see a "big correction" if the AI technology does not succeed, highlighting how the catch-22 situation that the AI trade is in right now creates vulnerability. He also flagged specific pressure points at the model layer, stating "I think one of the bottlenecks is that Anthropic and OpenAI are responsible for a lot of the spending that people are doing. And Anthropic and OpenAI have models that are now much more expensive than the competition." Both companies are privately held, so retail investors cannot directly own or short them, but their pricing shapes the unit economics of every hyperscaler CapEx dollar.
