By Spencer Jakab

This is an online version of my Markets A.M. newsletter. Get investing insights in your inbox each weekday by signing up here-it's free.

You know a company's earnings are a big deal when CNBC displays an on-air countdown clock in the run-up to the results.

In the last two-and-a-half years, Nvidia has gone from a fairly successful chip maker to the most-followed company on the planet-and briefly, the most-valuable one. Another positive earnings surprise this afternoon would be its 10th in a row. That makes Wall Street's use of the word "surprise" questionable.

Even more of a head-scratcher is what has happened to Nvidia's customers during that streak, which dates to the unveiling of ChatGPT in November 2022. Investors sometimes take things that are almost certainly true and spin them into a story that might not be-sort of like large language models themselves.

From December 2022:

Unless you've been living under a rock, you know AI is real. If it hasn't wowed you yet, then nothing will. And we know that the biggest buyers of Nvidia chips, known as hyperscalers, were already the most profitable companies in history. Amazon, Microsoft, Google and Meta are asset-light and fast-growing.

Supercharge those almost perfect business models with AI and you'll see fireworks-right? Well, you have in the stock market. The four have together tacked on $5 trillion in market value since November 2022. But, instead of making them more special, the new technology could weigh on them instead.

Between 2022 and 2026, the four companies will have made a whopping $1.14 trillion in capital expenditures, or three times what they did in the preceding five years, according to FactSet. If you're expecting a financial payoff from all that money then you'll have to keep waiting. Analysts' consensus expectations for their combined net profit this fiscal year has dropped by $22 billion since ChatGPT was unveiled.

Note the difference between capital expenditures and costs. Investments only hit profits gradually, so the bottom-line impact of buying all those Nvidia chips is delayed.

Meanwhile, AI's need for specialized chips, servers and data centers has made companies that needed little in the way of physical assets a lot more dependent on them. Business models built on bits, bytes and code bear at least a slight resemblance to smokestack industries of yore.

Like so many technologies that came before it-electricity, railroads and the internet-the economic benefits of AI could be huge and generally a good thing for the stock market, too. But they are best for a company like Nvidia selling the proverbial picks and shovels. Its projected revenue and net income this year are five and nine times higher than in November 2022, respectively.

The companies paying top dollar for those chips? They might just deliver the wrong type of surprise one day.

This item is part of a Wall Street Journal live coverage event. The full stream can be found by searching P/WSJL (WSJ Live Coverage).

(END) Dow Jones Newswires

May 28, 2025 06:57 ET (10:57 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.