Tech stocks await AI’s real reckoning

Once more with feeling. The artificial intelligence boom was big on the way up, and it has already been big on the way down. A selloff in technology stocks wiped around $3 trillion of paper wealth from the industry in under a month. The rout looks more driven by feelings and flows than facts. If that changes, a harder reset is possible.

Tech firms added $5 trillion of market value in the first six months of 2024, based on the 30% rise in the 416 stocks comprising the FR US Technology Index. Those handy gains made tech an easy place to sell if nerves got tested. And they roundly did, due to political shocks, a Japanese interest-rate shift, and rising U.S. unemployment. Between July 10 and Aug. 5, the CBOE Market Volatility Index tripled.

What hasn’t deteriorated is the market’s view of companies’ performance. Nvidia, maker of the chips powering AI giants’ efforts, lost one-fifth of its market value since July 10, even as estimates of next year’s earnings rose. That’s the case elsewhere, too: analysts expected S&P 500 tech firms’ earnings to increase 21% in 2025, as of Aug. 6 – more than the 19% predicted in April. Instead, it’s the multiple on which investors value those earnings that has come down. Citigroup strategists estimated in June that only one-third of AI-related stocks’ gains in the year to date came from rising profit estimates. Air is now being let out of overinflated tires.

Still, expectations change. AI is dominated by companies making huge investments for uncertain payoff. While Meta Platforms' 22% increase in ad sales last quarter suggests the recipe is working, that’s a tiny sliver of the eventual promised bounty. Getting the rest of the way requires a lot of cash: A Morgan Stanley analysis shows that 8 of the top companies investing in AI will plow $380 billion into capital expenditures in the next two years – 50% more than in the previous three. For Nvidia and its kin providing them the tools, that’s great: Meta’s Llama 4 AI model may use 10 times the computing power of its predecessor, says CEO Mark Zuckerberg.

The size of the prize itself is also up for grabs. AI could add nearly $18 trillion to the global economy, Accenture reckons. McKinsey has identified $4.4 trillion of productivity gains. But these are just assumptions; facts are yet to come in. Executives at Meta and Alphabet have conceded that it’s better to spend too much too soon than too little too late. That leaves room for volatility - and further upsets.