Steve Eisman warns about the potential collapse of AI spending

The same investor who accurately predicted the 2008 subprime mortgage collapse and made significant profits from it, Steve Eisman, is now turning his attention to a new risk territory: uncontrolled spending on artificial intelligence by tech giants. His recent analysis, published on his YouTube channel, draws a concerning parallel between current AI investments and the speculative bubble that devastated the tech sector over two decades ago.

The investor who was right on Wall Street now sees worrying parallels

Eisman’s credibility comes from his proven ability to identify speculative bubbles. Before the world experienced the 2008 global financial crisis, he had already spotted cracks in the U.S. mortgage system. Now, with the same analytical lens, he observes how Meta, Google, Amazon, and other tech companies are channeling unprecedented resources into artificial intelligence.

Over $300 billion annually in AI CapEx

Major tech companies are combining their efforts and budgets into AI-related capital investments, exceeding $300 billion. This spending represents an unstoppable acceleration in the development of language models, computing infrastructure, and AI systems. Everyone is racing to stay ahead in this new technological frontier.

The 1999 lesson: when internet gold rush turned into recession

Steve Eisman references a historical precedent worth noting. In 1999, internet analysts proclaimed that the web would conquer the world. Their prediction was correct, but what happened along the way was massive, disorderly investing. Capital flowed too quickly, in too large amounts, and the resulting overinvestment was a key cause of the 2001 recession. Even after that economic contraction ended, tech stocks remained stagnant for years.

The parallel Eisman suggests is potential but not negligible: if the pattern repeats, current excessive AI spending could precede a significant correction period.

Early signs that AI innovation is slowing down

Although Eisman admits that AI is not his main area of expertise, he points to observations from sector critics suggesting that the current model of AI development—progressively scaling large language models—is beginning to lose effectiveness. As an indicator, the recently launched ChatGPT 5.0 has not shown dramatic improvements over its predecessor ChatGPT 4.0.

These early signals suggest that tangible results from innovation may not justify the current level of investment. If confirmed, this scenario could be the prelude to a slowdown.

The risk of disappointing returns

The central question Steve Eisman raises is crucial: what will be the actual return on investment for all this AI spending? If initial yields fall short of expectations, the pace of investment will slow dramatically from its current dizzying rate. What would follow is a “painful digestion” period similar to what the tech industry experienced after 2001.

This scenario is not a definitive prediction but a cautious warning from someone whose analytical ability has proven extraordinarily accurate in critical market junctures.