Nvidia has reached a nearly $4 trillion valuation by designing and selling artificial-intelligence chips. A handful of lesser-known companies that help make those chips may be better bets for investors.
The largest chip-manufacturing-equipment companies -- including Applied Materials and Lam Research -- are in a market that is growing fast as AI booms. Capital spending on equipment to make advanced chips like those from Nvidia is expected to nearly double between last year and 2028, according to a recent estimate from industry group SEMI.
With global outlays on equipment set to top $100 billion this year, that's a lot of revenue to look forward to. Yet investors haven't given chip-equipment makers the plaudits they have showered on AI chip designers, from Nvidia to Advanced Micro Devices and Broadcom, as well as chip producers like Taiwan Semiconductor Manufacturing Co.
Applied Materials' stock has fallen 23% in the past year, leaving its price at about 18 times forward earnings. Nvidia's stock is up more than 27% over that period, putting it at about 32 times forward earnings.
The big reason for the divergence is the trouble equipment makers like Applied and Lam face in China. Chinese sales, which accounted for a quarter of Applied's revenue and nearly a third of Lam's in their most recent quarters, are taking a beating because of tightening U.S. export controls.
Those controls, which aim to crimp China's chip-making growth, are poised to shift market share there to domestic players. Western equipment companies are on track to lose as much as 7% of their Chinese revenue per year through 2030, New Street Research analysts said in a note last week.
It has also been hard for investors to ignore how the explosion of AI chip demand has been counterbalanced by a so-so market for a lot of other kinds of chips.
Prices for flash-memory chips, for example -- an area where Lam is a big equipment player -- fell more than 15% year-over-year in the first quarter, according to analysis firm TrendForce. The market is starting to recover, but memory-chip makers aren't rushing to add to their production.
Questions about the trajectory of the global economy and the potential impact of U.S. tariffs have also clouded the picture.
Yet those are short-term dynamics in an industry that is notoriously cyclical.
Barring a bursting of the AI bubble, there is little reason to question whether the longer-term trend -- a rise in chip production -- will continue. That global chip sales will rise to more than $1 trillion by 2030 has become an industry mantra. That is probably optimistic but at least directionally correct. Sales were $624 billion last year.
"There are a lot of near-term questions," Bernstein Research analyst Stacy Rasgon said in a recent conference discussion with Lam's chief executive. "But I'm also increasingly getting people looking at the long-term potential of this industry as they start to view it in more secular rather than cyclical terms."
One reason for that optimism is the likelihood that chip-making capital intensity -- the percentage of chip revenue that is spent on equipment -- is going to increase, playing into the industry's hands.
That is happening because more advanced equipment is necessary to produce chips with ever-smaller transistors that can make ever-faster calculations.
The industry is also shifting toward new ways of squeezing out greater performance. This involves stitching together and sometimes even stacking up a bunch of smaller chips.
Such so-called advanced packaging is already used heavily in AI chips. It will require more equipment.
Applied Materials had about $1.7 billion of revenue from advanced packaging in its latest fiscal year. The company expects that amount to double in the coming years.
Equipment companies offer investors something else, beyond those technology and business dynamics: They give them broader exposure to the growth of chip making.
Of course, investors aren't protected from the mistakes of any individual equipment company. But success in the industry tends to track the chip market rather than a narrow set of chips like those Nvidia or AMD sell.
With the equipment makers' relatively attractive valuations, the best value may be found on the chip-factory floor.