Nvidia Soars, History Advises Caution. These 12 Stocks Are Now a Safer Way to Play AI.

Looking at it in a different way, we’re already seeing growth slow. Over Nvidia’s past four earnings reports, quarter-over-quarter growth has slowed from 88% to 34% to 22% to 18%.

I’m not dumb enough to jump in front of a roaring freight train—I wouldn’t suggest anyone short Nvidia stock. But the whole thing seems a little, if not unhinged, let’s just say unsustainable.

I’m no AI skeptic. I’m not even accounting for increased competition in AI chips, or the fact that enterprise software companies are having some alarming problems turning AI promise into real-world profit. But Nvidia is already worth more than any company on the planet, now or ever.

Another 20% rally would take the stock’s market cap to $4 trillion, a milestone never before reached. It might be time to temper your enthusiasm, at least a little.

For one thing, Nvidia’s forward price/earnings multiple has gone from roughly 25 at the end of last year to a current 45. More eye-opening is that the stock trades for 20 times expected revenue for the January 2026 fiscal year, based on Wall Street estimates.

A few years back, I wrote a column citing data from Bernstein analyst Toni Sacconaghi showing that stocks trading at price-to-sales multiples above 15 times tend to be terrible investments. From 1970 to 2020, Bernstein’s data showed, those stocks underperformed the market by 18 percentage points over the subsequent three years—and 28 points over five years. For stocks trading above 20 times sales, the returns were even worse.

I know what you’re thinking. It’s different this time. This is AI! And sure, maybe AI really is the most important thing to happen in technology since cloud computing, or the internet, or mobile phones, or even the personal computer. But the numbers worry me.

Nvidia’s market value is now nearly five times the industry estimate for next year’s global chip sales—yes, the total from every company worldwide.

Microsoft

has seven times the number of employees Nvidia does, and twice the sales.

Apple

has five times the staff, and triple the sales volume. Nonetheless, this past week, Nvidia’s market cap vaulted past them both.

And here’s the other thing. There are picks-and-shovels AI bets that don’t require the same heroic assumptions that Nvidia stock now entails. Here are a dozen ideas, many of which have been highlighted previously in this column or elsewhere in the magazine. None of these is growing triple digits, but they’ll all benefit from the continued growth of AI.

Micron Technology

: You know what AI data centers need beside GPUs? Lots of high-bandwidth memory. Micron is seeing more demand for this memory than it can fulfill. The push for AI capable phones and PCs is going to increase memory demand too.

Arista Networks

: You know what else is needed as data centers grow? Networking hardware. Microsoft and

Meta Platforms
,

which account for about half of Arista’s sales, are both ratcheting up spending to respond to growing AI needs.

Corning

: I laid this one out in detail last week, but the bottom line is that the more GPUs you combine into AI server racks, the more fiberoptic cable you need to connect them all.

Arm

/

Qualcomm

/

SoftBank

: Qualcomm’s Snapdragon X processors are poised to steal market share from

Intel

and

Advanced Micro Devices

as AI-capable PCs become a real market. Those processors are based on Arm’s chip designs. SoftBank, meanwhile, owns 90% of Arm, while trading at about a 50% discount to net asset value. Their fates are intertwined.

Oracle

: I was early on this one—Oracle is now competing head-to-head with

Amazon
.

com,

Alphabet
,

and Microsoft in the cloud, and gaining share. While growth had been stagnant at the enterprise software giant for years, it is now poised to return to double-digit expansion. Oracle recently signed OpenAI as a customer, and it’s partnering with Microsoft and Google to make their AI clouds more easily interoperable.

Taiwan Semiconductor

: There is geopolitical risk here, but when it comes to building AI chips, there aren’t really good alternatives to TSMC. Nvidia, AMD, and even Intel are relying on Taiwan Semi’s cutting edge fabs to make AI chips.

HP Enterprise

/

Dell

: Both hardware companies are seeing surging demand for their Nvidia-based AI servers; Dell is also poised to benefit from the coming AI PC wave. And they’re bargains. Both trade for about one times forward revenue.

Microsoft: Microsoft is a multipronged bet, which benefits from the rise of OpenAI, improving demand for the company’s Azure cloud, an assortment of AI Copilots, and the gradual return of Bing as at least a marginal competitor to Google in search. No software company is better positioned.

Adobe

: The dominant player in creative and marketing software, Adobe has gone all in on AI. Wall Street has been worried about competition from AI upstarts—the stock is off 12% this year—but Adobe has already started generating real revenue from AI-enhanced versions of its content creation software. Buy the dip.

Write to Eric J. Savitz at eric.savitz@barrons.com