Are AI Stocks a Bubble?

Are AI stocks a bubble? John Cochrane argues the honest answer is both yes—and no. Valuations are high, and history shows that waves of technological excitement often end with lower returns. But high prices alone don’t mean markets are irrational or that a dramatic crash is inevitable. From railroads to dot-coms, innovation has repeatedly driven heavy investment and rising prices long before the ultimate winners were clear.

What distinguishes the AI boom is how markets process information. Heavy trading and volatility often reflect investors reacting to new data, not blind hype. In this rant, Cochrane explains why stocks can behave like money—valuable not just for returns, but for trading on information. His advice: if you’re not playing that game, don’t pay for liquidity or try to bet against a “bubble.” And never forecast stock prices.

Subscribe now

Transcript

Hi, I’m John Cochrane, senior fellow here at the Hoover Institution and welcome to my weekly rant.

Today, we’ll talk about AI stocks. Are AI stocks a bubble? Well, almost certainly yes and almost certainly no.

What do I mean by that? Valuations are high. And one thing we know historically is stocks that are very high priced relative to earnings book value and other metrics, have let’s put it politely, low returns going forward.

They will go down, though it’s hard to tell when — timing the end of an event like this is very difficult.

There’s also lots of real investment going on. They’re building data centers like nobody’s business, and it’s not clear that all of those would be profitable in the end.

We’ve seen this many times before, a great technical innovation that leads to lots of investment, high stock valuations, and then a period of low returns, to put it politely at the end of it.

Railroads, radio, movies, cars. Those were the hot technical, questions of the 1920s, which led to a stock market boom, lots of investment, and quite a big bust afterwards.

The tech boom of the 1990s is, of course, familiar. We built a lot of internet, built fiber optic cables, stock prices went up, and then they went down, although now they’re back up again. Bubbles bursting aren’t forever.

Houses in 2008 went up, and then went down, and then they’re back up again.

What does a bubble mean?

There’s no clear definition of a bubble and that’s one of the reasons why it’s fun to talk about and arguments never end. I recommend you watch the Nobel Prize debate between Gene Fama and Robert Shiller on whether stock market bubbles exist, and they couldn’t even agree on the definition of a bubble. Well, that’s the problem.

Sometimes it means “I think stock prices are high and I wish I had sold yesterday.” Sometimes it means just “I think people are dumb.” Both of those are kind of silly things to talk about. These are cycles that we’ve seen over and over again. Can we do better in understanding these cycles than to just say “people are dumb?” Except, I’m not smart enough to have made tons and tons of money on it on seeing how dumb they are.

Well, yes.

What’s the pattern that we’re seeing? There’s exciting new technologies. There’s lots of information about these technologies. Lots of betting. Which of the AI companies are going to win out over the other ones? Which technologies are going to work? Is some new chip company going to come in. Will the data centers really prove out worthwhile or not?

There’s lots of trading on that information. The structure of stock markets is that it’s very hard to short something for a long period of time that you think it’s overvalued.

So, if you want to play this game, betting on this week’s news, less next week’s news about what’s up, what’s down in this exciting new technology, you got to buy the stocks. That drives the prices of tech stocks up.

And if you’re playing this game by a little bit now, sell next week, buying the next piece of news, the long-term drag that the level of the stock is too high just isn’t that important compared to getting the next movement right.

In many ways, stocks are like money. Remember, money and bonds are a claim to the same security. They’re both money a year from now. But the bonds pay a lot better return than money.

Is money a bubble in that sense? Well, it’s a bubble, yes, but it’s useful. We all hold a little bit of money at low interest rate. Why? It’s useful to make those transactions.

Similarly, stocks can be have very high prices, but be useful if you’re in that business of making the high volume transaction.

And that pattern holds in each of the “bubbles.” There was new information. It was in the industries that were growing. There was lots of volatility in volume, which is a key point. If people are irrationally sending up the price of something today and buying it, why do they sell it next Thursday and then buy it again Tuesday after that?

That’s the sign of people trading on information, not just irrationally overvaluing things.

So yes, valuations are high, returns are low. And there’s a reason values are high. Same reason money is high, same reason Bitcoin is high. People find these securities useful for trading on the information.

What should you do about it? Well if you’re not playing the game and if you’re not better than other people at playing that game, stay out. That at least is good advice, as well as the advice, “don’t hold too much extra cash.”

Value strategies that buy low price things that are uninteresting, they automatically do that. But overall, don’t pay for liquidity or trading value you don’t need. Betting against the bubble will be hard because of course it’s hard to short sell stocks for a long time.

For more on this view I have an old paper, “Stocks as Money,” great title applied to the last bubble. I think it applies to this one too. We’ll see how it all works out. Never forecast stock prices.

Thank you for listening and don’t forget to hit the subscribe button.

Subscribe now