Steve Eisman: Massive investment in AI could create a tech bubble like the one in 2001

Steve Eisman, the investor who correctly predicted the 2008 mortgage crisis and made extraordinary profits from the subprime market collapse, has spoken again about the risks threatening the financial sector. This time, his warning is not about mortgages but about the uncontrolled rise of artificial intelligence and the corporate spending that accompanies it. Eisman argues that tech giants are channeling unprecedented resources into AI but warns that if these expenditures do not generate tangible returns, the sector could experience a correction as severe as the one technology faced more than two decades ago.

The investor who foresaw the 2008 disaster now warns of tech excess

The current race to dominate artificial intelligence, in Eisman’s view, resembles the years leading up to the 2001 tech recession. During that period, large corporations engaged in frantic overinvestment, funding internet projects with exaggerated expectations. Although the internet eventually conquered the world as some predicted, the path to that success was riddled with sharp declines and disillusionment.

“The analogy we can draw, even if only potentially, is interesting,” Eisman states. “In 1999, tech analysts weren’t wrong in projecting that the internet would transform society. Reality eventually proved them right. However, the gold rush to invest in the internet was disproportionate: too much capital, too much speed, too many failed projects. That overinvestment was largely responsible for the recession that hit the sector in 2001. Tech stocks took years to recover after that crash.”

$300 billion annually: a winning bet or reckless investment?

Meta, Google, Amazon, and other giants are collectively spending over $300 billion annually on CapEx—capital expenditures—dedicated solely to AI projects. These corporations fiercely compete to lead the AI revolution, but no one is certain what the true return on this colossal spending will be.

Steve Eisman admits that AI is not his area of expertise but notes that some experts have begun questioning the current AI development model. According to critics, the prevailing strategy of simply scaling larger language models may be losing effectiveness. Early evidence includes the release of ChatGPT 5.0, which analysts say does not represent a significant advance over its predecessor, ChatGPT 4.0.

Is the rapid AI innovation reaching a plateau?

Initial signs of a possible slowdown in tangible results from AI innovation are emerging. If these symptoms are confirmed and the returns from the enormous expenditures turn out to be disappointing—at least in the short term—tech companies would drastically slow their investment pace. This would lead to a prolonged period of economic contraction in the sector, comparable to the industry downturn after the dot-com bubble burst over 20 years ago.

Steve Eisman emphasizes that these predictions are not certainties but potential analogies based on historical patterns. However, the warning is clear: overinvestment cycles followed by severe corrections are recurrent in speculative markets, and the tech sector’s history repeatedly demonstrates this. The real risk is not only that AI fails as a technology—after all, the internet succeeded—but that the financial returns on current spending are deeply unsatisfactory for years, leading to a “painful digestion” period similar to that of the early 2000s.