Everyone is admitting the AI boom might be a bubble

At least it’s out in the open.

Over the past couple of weeks, a number of investors, bankers, and tech oligarchs have frankly acknowledged that the surge of AI investment powering the stock market and the economy may meet the criteria of a bubble.

On Friday, Amazon founder Jeff Bezos said the investor rush — from retail traders jumping on highly valued stocks like Palantir, CoreWeave, and SoundHound AI to the massive investment funds and corporations bankrolling the build-out of data centers across the country — might be what he described as a “good” kind of bubble.

Right on cue, Monday morning brought an announcement that OpenAI and AMD had signed a gargantuan computing deal that was worth “tens of billions of dollars in revenue,” sending AMD, a company that was already worth roughly $270 billion, up more than 25% premarket. Other AI-adjacent stocks climbed, too.

Just a couple of weeks ago, Meta CEO Mark Zuckerberg said in a podcast interview that an AI bubble was “quite possible.”

They aren’t the only ones: Goldman Sachs CEO David Solomon, tech investors James Anderson and Roger McNamee — heck, even Sam Altman himself, the CEO of a company that burns cash at an alarming rate and will need to test the limits of private markets to raise enough cash to fulfill its mission — has made a similar concession.

Points for candor. But not too many points.

After all, it would look kind of foolish to argue there’s zero chance of a bubble, after a quick glance at the record spending key companies are doing...

...or the remarkable upsurge in construction spending on data centers.

But so what? Speculative investment booms have been features of market economies since shares in the Dutch East India Company were the hottest trade in Amsterdam.

And for centibillionaires Bezos and Zuckerberg, whose companies (and shareholders) are making some of the biggest bets in financial history on AI, the risks are justified by the potential benefits of the technology — even after factoring in the risk that a sudden crash in prices could occur. Bezos said as much on Friday, the Financial Times reported:

“‘This is kind of an industrial bubble as opposed to financial bubbles,’ Bezos said at a tech conference in Turin on Friday, drawing parallels with the dotcom-era investment in fibre-optic cable that outlasted many of the companies who deployed it and the ‘life-saving drugs’ that emerged from the 1990s biotech boom and bust.

‘The banking bubble, the crisis in the banking system, that’s just bad, that’s like 2008. Those bubbles society wants to avoid,’ he said.”

Zuckerberg, likewise, minimized the potential downside of any eventual AI crash.

“If we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate, obviously,” he said. “But what I’d say is I actually think the risk is higher on the other side.”

They may have a point.

Economists and historians who have studied the aftermath of market crashes over the centuries say busts hurt the economy the most when they are financed by debt or loans from the banking system. That’s not what’s happening at the moment.

Much of the spending on AI is fed by the giant profits that companies like Amazon, Microsoft, and Nvidia produce, rather than the bond market — which, for instance, fueled the US housing market boom that turned into a bust and financial crisis in 2008.

“It makes a big difference in terms of what we would expect the economic consequences of the bubble to be,” said William Quinn, an associate professor of history at Queen’s University Belfast and the coauthor of “Boom and Bust: A Global History of Financial Bubbles.”

For bubbles financed by stock market equity or corporate cash flows, “you need to worry less from the perspective of an economic agent or a consumer in the economy. But obviously as an investor, it’s still a problem for you if we’re in a bubble and if the bubble bursts.”

That’s not exactly a small point. In the aggregate, Americans are more exposed to the stock market than ever before, with stock portfolios accounting for roughly 35% of the net worth of American households.

And while most of that stock market wealth is concentrated among the wealthiest families, there are still a lot of people out there that own at least some stock. Gallup data shows that some 62% of Americans said they have money in the markets, up from the low 50s in the aftermath of the market bust in 2009.

So clearly, there’s a broad swath of the public with at least some stake in the AI boom. And maddeningly, even if we were to all agree that a bust is coming, there’s no real way to know when it will hit, Quinn said.

“A lot of the time the bursting of a bubble seems to be completely random,” he said. “So with the Wall Street crash [of 1929], there was no obvious trigger. People sort of come up with various proposals for what might have caused it, but none of them are plausible.

“It’s just one day, some people started to sell. And then other people saw that they were selling and they started to sell, and then you started to get lots of margin calls and the whole thing spiraled.”

Still, there are valuable lessons to be learned from past bubbles, Quinn said.

“What we are able to see from history is kind of who gets out whenever the bubble bursts. And rather than retail investors, it tends to be people with good inside information.”