Intel still has a long way to go to reclaim its former glory. The problem is that investors are acting like that’s already happened.
The storied chip maker’s stock price has soared 88% so far this year, and more than tripled over the last 12 months. That has taken the company’s market capitalization close to the $350 billion mark for the first time since 2000, when it was the undisputed leader of both designing and manufacturing the most advanced semiconductors on the planet. Intel also now trades for more than 130 times its projected earnings for this year, far above the peak multiple of 60 times that the stock briefly touched during the dot-com bubble.
That is a risky price for a company still trying to pull off an ambitious turnaround. And even if fully successful, Intel’s stock is pricing in a level of earnings power that looks hard to achieve, given the company’s shifting business model and the competitive realities of the AI chip market.
Intel is scrambling to catch up to TSMC in chip manufacturing technology after ceding its lead several years ago. At the same time, the company is trying to pull off a complex business model shift, whereby it designs its own chips while also manufacturing chips designed by others. And it is trying to win back share for the processors it designs for servers and personal computers.
All these are still works in progress, though some notable progress has been made. Intel struck a deal in September with Nvidia to co-develop a new class of chips for data centers and laptops. That deal kicked off the stock’s recent run, but developments since then have given the shares even more juice. Even social-media posts by President Trump have made an impact; Intel’s stock jumped 11% in one day in early January after Trump praised Intel Chief Executive Lip-Bu Tan and said the U.S. government “is proud to be a shareholder of Intel” on his Truth Social network.
In the past month, Intel landed a major chip supply deal with Google and was named a partner in Elon Musk’s Terrafab project, which aims to build a massive new chip production facility in Texas that would serve the needs of both Tesla and SpaceX.
But the Terrafab announcement was light on details, beyond saying Intel would help with “refactor silicon fab technology” for the project. “Frankly we aren’t even sure what that means,” wrote Stacy Rasgon, an 18-year chip analyst at Bernstein who has a Ph.D. in chemical engineering.
One thing that really is going Intel’s way is a shift in AI computing. That market has so far been dominated by graphics processing chips, or GPUs, that are mostly supplied by Nvidia. But while GPUs are still superior for training AI models, newer AI tools such as agents require a process known as inferencing that is better suited for central processing chips, or CPUs. Those types of chips have typically served as the main “brain” of servers and personal computers that have historically been Intel’s strongest markets.
“This dynamic should provide a renewed growth engine for CPUs that have until recently been viewed by many as ex-growth at best, and at worst left for dead by investors,” TD Cowen analyst Josh Buchalter wrote in a report earlier this month. New Street Research projects that shipments of CPUs for AI servers will average 43% annual growth through 2030.
Not all of this will flow to Intel. Rival AMD has gained a lot of market share in server CPUs over the past few years at Intel’s expense. Still, Intel’s strong position in the server market should help the company capture some of the new momentum, especially as demand also grows for replacements of traditional non-AI servers. Analysts expect Intel’s data center segment to increase revenue by 13% this year compared with an anemic 2% growth projected for the company’s PC business, according to estimates from FactSet.
But even improving server sales are unlikely to return Intel to past levels of profitability. The market for AI-focused CPUs is getting more competitive, with Arm Holdings and even Nvidia now coming in with new chips. Tim Arcuri of UBS sees Intel’s gross margins reaching 50% by 2030 only under a “blue sky scenario.” The company averaged an annual gross margin of over 60% from 2010 to 2020, according to data from S&P Global Market Intelligence.
In the near-term, Intel’s data center business may also remain hampered by supply constraints cited in the last earnings call in January. That could pressure results for the first quarter that Intel will report on Thursday. That doesn’t change the strong prospects for CPU chips over the longer term, but Intel isn’t priced for even a moment of pain these days.
Write to Dan Gallagher at dan.gallagher@wsj.com