Worries about an AI bubble have gone mainstream. From the pages of the New York Times to the trends on Google, people are looking at the stock market’s rapid rise, the ascension of companies such as CoreWeave and Palantir Technologies, and the circular dealmaking among Nvidia, OpenAI, and tech companies of all stripes to draw the conclusion that yes, yes, there is a speculative bubble. And when bubbles are bemoaned, you can bet that mentions of the dot-com boom and bust won’t be far behind.
For the bears, “since the dot-com bubble” has become a mantra of sorts, especially when talking about valuations. And for the bulls, the fact that there is no obvious Pets.com is a sign that speculation hasn’t reached the silly state just yet. But these pronouncements miss what made the internet bust so devastating—and it’s about more than just speculative fervor in the stock market.
The word “bubble” may be one of the most overused in the financial markets. It gets tossed around whenever a group of stocks’ gains look ridiculous compared with their fundamentals. The gains in nonprofitable tech stocks favored by Cathie Wood and her ARK Innovation exchange-traded fund from 2020 to 2022 were dubbed a bubble, as was the craze for meme stocks such as GameStop and AMC Entertainment that began in 2021. There was also a bubble in special purpose acquisition companies in 2021, and in Bitcoin in 2017 and 2021. All of them popped, and aside from a market correction in 2022, none of them caused lasting damage to the stock market or the U.S. economy. People made money, people lost money, and life went on.
The dot-com bubble was different. Sure, there was the speculative fervor that caused shares of companies like Pets.com, Kozmo.com, and even BN.com, the online business of Barnes & Noble, to pop. But the internet boom was about far more than sock puppets. The real damage came when the fervor bled out into the real world—when utilities like Montana Power decided to pivot to building a fiberoptic cable network, when good businesses like Time Warner merged with bad ones like AOL, and when everyday Americans spent the paper wealth they had accumulated before watching it evaporate. (I still remember the drummer in a band I played with asking me what to do with his Excite@Home stock.) By the end of the bubble, the gains weren’t even in websites anymore. In 1999, they were in chip makers like Qualcomm, which gained 2,600%, while Amazon.com rose a paltry 28%.
AI isn’t quite there yet, but it’s well on its way. Despite those who doubt its utility, it is a big deal. It can already do menial tasks as well as many human beings. OpenAI’s ChatGPT, Google’s Gemini, and others can do tasks like simple coding, drafting emails, and basic research as well as an entry-level worker. Perhaps one day it will even be able to write this column in the style of Alan Abelson—and put me out of a job.
AI has now gone from a cool toy to a prime mover of both the economy and the stock market. Though measuring its economic impact is tricky—the contribution from AI capital spending to gross domestic product during the first six months of the year has been estimated to be anything from 0.1 percentage point to all of it—it’s nearly everything for the 10 biggest stocks in the S&P 500 index: 22V data show that nearly three-quarters of their returns are due to AI. The same is true among sectors. Industrial companies that have mentioned AI favorably have gained 14% this year, while those that haven’t mentioned it at all have risen just 7.6%.
“Broadly speaking, the U.S. economy and markets are essentially all in on this GenAI ecosystem construct,” writes Peter Boockvar, author of the Boock Report newsletter.
Still, there are concerns that the spending is unsustainable. The level of circular financing—companies lending money to others to buy their products—has been rising. OpenAI has gotten a piece of Advanced Micro Devices in exchange for agreeing to buy its chips, while also signing a $300 billion cloud deal with Oracle. All told, OpenAI plans to spend $1 trillion dollars of money it doesn’t yet have to build out its AI infrastructure. (For more, see the Tech Trader column.)
If it were just OpenAI, it wouldn’t be so worrisome. But lately we’ve seen an initial public offering for Fermi, a company that plans to develop power for data centers in Texas; for Via Transportation, which uses AI to create transportation solutions; and for WhiteFiber, a designer of AI data centers and other infrastructure. It isn’t quite silly season yet, but the giggles are starting.
Still, there’s no need to back away from it yet. The S&P 500, after all, gained 20% or more a year from 1995 through 1999, which suggests that there could be more upside ahead.
Wealth Consulting Group’s Talley Leger is watching four potential disruptors to the rally. First, the U.S. can’t produce enough power to satisfy AI’s demands. Second, investors start doubting the potential profitability of all the spending on AI. Third, investors sour on negative returns on invested capital at Big Tech companies. Fourth, tech companies start returning cash to shareholders through dividends or buybacks instead of spending on AI.
Until they come to fruition, it’s time to stop worrying and love the bubble.
email: ben.levisohn@barrons.com
