Aswath Damodaran, a finance professor at New York University, is warning that a future AI crash could be more painful than the bursting of the dot-com bubble around 2000. His concern is not simply that investors may be too optimistic. It is that the AI boom is being built on a cost structure that could make any correction harder to contain.
Why this AI boom looks different
In the podcast Intangible Economy, Damodaran argues that the current AI cycle is not a clean repeat of the dot-com period. The key difference is infrastructure. AI companies need massive investments in physical infrastructure, and much of that spending is financed with debt.
That matters because debt changes who gets hurt when expectations fall. In a typical stock-market correction, shareholders take the direct hit. Damodaran’s warning is that an AI correction could have broader effects because the sector’s buildout reaches beyond software valuations and into factories, infrastructure, borrowing and depreciation.
His point is not that AI cannot be important. It is that the financial machinery around AI is heavier than many investors may be treating it. The larger the physical buildout, the more the business depends on future demand being large enough, durable enough and profitable enough to justify what has already been spent.
The scaling question
Damodaran also challenges a common assumption about AI business models: that they will scale like traditional software. In many software businesses, adding users can be highly profitable because the cost of serving one more customer can become very low. He says AI does not fit that pattern as neatly.
Every additional use of AI consumes compute. Damodaran compares this more to Spotify, which pays for each stream, than to a business where costs are spread across a fixed base. He contrasts that with Netflix, where high content costs can be spread over a growing subscriber base.
This difference affects margins. If growth brings rising usage costs with it, more users do not automatically mean better economics. Growth paired with thin margins could destroy value rather than create it.
Damodaran also points to the risk of price erosion from Chinese competitors like Deepseek. With margins already low, weaker pricing power would make the economics even more difficult. The result is a business model that may need enormous demand while still struggling to turn that demand into strong profits.
The uncomfortable bull case
Damodaran’s warning is not limited to the downside case. He also sees danger in the most optimistic version of the AI story. If AI is valued on the idea that it can replace entire jobs rather than simply help people work, then the social consequences become central to the investment case.
In that scenario, he says, if AI delivers on the promise being used to justify its value, "half of white-collar workers" would lose their jobs. That is why Damodaran frames the strongest AI narrative as a risk to society as well as to portfolios.
"The scary thing is the big stories you tell that can justify AI, if they come true, are going to create some insane costs for society that we better start thinking about right now,"
He calls this scenario the "AI fever dream." The phrase captures the tension in the current market: the most exciting version of the technology may also be the version with the most disruptive consequences.
Big tech’s changing financial profile
Damodaran says he owns five of the seven so-called Magnificent Seven stocks, including Amazon, which he has held on and off since 1997. But he says these companies are changing at a fundamental level because of their heavy AI investments.
For companies that once grew with minimal capital spending, this is unfamiliar territory. Damodaran says the analysis can no longer focus only on margins and new business lines. He now also has to study capital expenditures and depreciation.
The issue is that large infrastructure investments carry accounting and business risks. Damodaran says companies are building massive factories and infrastructure that will be depreciated over ten years but could be obsolete after five.
"I'm not sure they really know what they're getting themselves into,"
That comment cuts to the heart of his concern. Big tech companies have long been treated as capital-light, high-margin businesses. AI may push them toward a model where they must make vast upfront commitments before knowing whether the economics will hold.
Why restraint may matter
Damodaran views Apple’s more cautious approach as smarter than many critics suggest. Many analysts have faulted Apple for not jumping in headfirst, but he sees restraint as a strength.
His reasoning is direct: Apple can observe what others do, learn from their mistakes and avoid pouring billions into areas where it has no experience. In his words, "we undervalue restraint in business."
That does not mean every company should avoid AI investment. It means the current AI boom may reward discipline as much as ambition. If the technology proves powerful but expensive, the winners may not simply be the companies that spend the most. They may be the ones that understand the costs, the pricing pressure and the social risks before the market forces them to.