A burning question that’s on a lot of people’s minds right now is: Why is the U.S. economy still holding up? The manufacturing industry is hurting badly from Trump’s tariffs, the payroll numbers are looking weak, and consumer sentiment is at Great Recession levels:

And yet despite those warning signs, there has been nothing even remotely resembling an economic crash yet. Unemployment is rising a little bit but still extremely low, while the prime-age employment rate — my favorite single indicator of the health of the labor market — is still near all-time highs. The New York Fed’s GDP nowcast thinks that GDP growth is currently running at a little over 2%, while the Atlanta Fed’s nowcast puts it even higher.

One possibility is that everything is just fine with the economy — that Trump’s tariffs aren’t actually that high because of all the exemptions, and/or that economists are exaggerating the negative effects of tariffs in the first place. Weak consumer confidence could be a partisan “vibecession”, payroll slowdown could be from illegal immigrants being deported or leaving en masse, and manufacturing’s woes could be from some other sector-specific factor.

Another possibility is that tariffs are bad, but are being canceled out by an even more powerful force — the AI boom. The FT reports:

Pantheon Macroeconomics estimates that US GDP would have grown at a mere 0.6 per cent annualised rate in the first half were it not for AI-related spending, or half the actual rate.

Paul Kedrosky came up with similar numbers. Jason Furman does a slightly different calculation, and arrives at an even starker number:

And here’s an impressive chart:

The Economist writes:

[L]ook beyond AI and much of the economy appears sluggish. Real consumption has flatlined since December. Jobs growth is weak. Housebuilding has slumped, as has business investment in non-AI parts of the economy[.]

And in a post entitled “America is now one big bet on AI”, Ruchir Sharma writes that “AI companies have accounted for 80 per cent of the gains in US stocks so far in 2025.” In fact, more than a fifth of the entire S&P 500 market cap is now just three companies — Nvidia, Microsoft, and Apple — two of which are basically big bets on AI.

Now as Furman points out, this doesn’t necessarily mean that without AI, the U.S. economy would be stalling out. If the economy wasn’t pouring resources into AI, it might be pouring them into something else, spurring growth that was almost as fast as what we actually saw. But it’s also possible that without AI, America would be crashing from tariffs.

Trump certainly seems to think AI is a golden goose worth protecting. Joey Politano points out that even as Trump has slapped tariffs on a plethora of industries, he has left AI and its supply chain mostly untouched:

But despite Trump’s tariff exemptions, the AI sector could very well crash in the next year or two. And if it does, it could do a lot more than just hurt Americans’ employment prospects and stock portfolios.

If AI is really the only thing protecting America from the scourge of Trump’s tariffs, then a bust in the sector could change the country’s entire political economy. A crash and recession would immediately flip the narrative on Trump’s whole presidency, much as the housing crash of 2008 cemented George W. Bush’s legacy as a failure. And because Trump’s second term is looking so transformative, the fate of the AI sector could potentially determine the entire fate of the country.

So a whole lot is riding on the question of whether an AI bust will crash the economy. The stakes could hardly be higher.

A lot of bubbles are purely financial beasts, driven by irrationality or coordination problems in the markets for stocks, bonds, and derivatives. For example, you can have a speculative bubble, in which a bunch of people know an asset is overpriced, but think they can sell out before the crash, and so they keep buying and buying and pushing the price up and up. You can also have an extrapolative bubble, when people see the price of something going up and up, and mistakenly decide that it must be due to some underlying positive trend.

But a much simpler possibility is that investors could make a big mistake about how valuable some technology is. They could honestly believe that AI is going to create immense amounts of value, and they could just end up being wrong. Then when they realize that the technology isn’t all it’s cracked up to be, they could temper their expectations, which would cause a price crash in AI stocks. But the stock crash wouldn’t be the real problem; far more painful would be the wave of loan defaults and financial distress that would result from AI’s actual shortcomings.

If there’s an AI crash, it’ll probably be this latter type. Jeff Bezos calls it an “industrial bubble”, and I think that’s as good a name as any. This kind of bubble is still a financial phenomenon, since the banking system gets hurt. But the cause is a mistake about real technology, rather than asset markets going haywire.

Everyone who’s talking about an AI bubble is basically warning that the technology itself might disappoint. For example, here are some excerpts from a big Bloomberg feature about the possibility of an AI bubble:

Even some of AI’s biggest cheerleaders acknowledge the market is frothy, while still professing their belief in the technology’s long-term potential. AI, they say, is poised to reshape multiple industries, cure diseases and generally accelerate human progress…Yet never before has so much money been spent so rapidly on a technology that, for all its potential, remains somewhat unproven as a profit-making business model…

The data center spending spree is overshadowed by persistent skepticism about the payoff from AI technology. In August, investors were rattled after researchers at the Massachusetts Institute of Technology found that 95% of organizations saw zero return on their investment in AI initiatives.

More recently, researchers at Harvard and Stanford offered a possible explanation for why. Employees are using AI to create “workslop,” which the researchers define as “AI generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task.”…

AI developers have also been confronting a different challenge. OpenAI…Anthropic and others have for years bet on the so-called scaling laws…Over the past year, however, these developers have experienced diminishing returns…Some have also struggled to match their own hype. After months of touting GPT-5 as a significant leap, OpenAI’s release of its latest AI model in August was met with mixed reviews…

There’s also the risk that the AI industry’s vast data center buildout, entailing a huge increase in electricity consumption, will be held back by the realities of strained national power networks.

When you bring up concerns like this to an AI engineer, executive, or founder, they tend to just smile at you indulgently, secure in the knowledge that their invention is everything it’s cracked up to be, and that much better things are already in the pipeline.

But this doesn’t reassure me. Because when we look at the history of industrial bubbles, and of new technologies in general, it becomes clear that in order to cause a crash, AI doesn’t have to fail. It just has to mildly disappoint the most ardent optimists.

This is why I think an AI crash is more likely than a lot of people in the tech world — or the Trump administration — realize.