Intel splits itself in two to aid CEO Pat Gelsinger’s turnaround plans—and links up with archrival Arm

Once the undisputed leader in computer processors that weren’t for mobile handsets or embedded devices, Intel’s x86 architecture is increasingly being challenged by Arm’s more efficient architecture everywhere from laptops to data centers (even Nvidia, whose stock continues to soar on the back of AI mania, is using Arm’s architecture in its latest AI data center processors). So, over the past few years, Intel CEO Pat Gelsinger has tried to pivot his company toward a foundry model, i.e., making chips for others, like industry leader TSMC does. And yesterday, Intel took a huge step in that direction by effectively splitting itself in two.

Under the dramatic reorganization and rebranding, we now have Intel Foundry and Intel Product—separate legal entities that remain parts of the same overall company but even have their own sales forces and back-end business systems. Intel Foundry will obviously make the processors that Intel Product develops but via arm’s-length transactions. Intel Foundry’s other customers get strict confidentiality, and everyone’s welcome. “If we’re going to be the Western foundry at scale, we can’t be discriminating in who’s participating in that,” Gelsinger said yesterday. Of course, this wouldn’t be a 2024 chip story without copious lashings of AI, so he described Intel Foundry as “the world’s first systems foundry for the AI era.”

Despite early customer wins a couple years back in the form of Qualcomm and AWS, Intel’s self-prioritization has so far largely held back the company’s foundry hopes. With this clear separation of church and state, Gelsinger hopes everyone will now flock to Intel. The ambition here is stark: He wants Intel to be No. 2 to TSMC by 2030, which would mean leapfrogging the likes of China’s SMIC and South Korea’s Samsung. And that requires linking up with Arm, whose architecture is being used by many potential Foundry customers.

The details on Intel and Arm’s vaguely titled “Emerging Business Initiative” remain scarce, but the gist is that Intel will be working closely with startups developing their own Arm-based system-on-chip processors, “offering essential IP, manufacturing support, and financial assistance,” according to a statement given to Fierce Electronics.

Describing Intel’s incoming chipmaking technologies as “industry-changing,” Haas said Arm had to be “a part of it … We need total transparency, seamless communication. We need to act like we’re working with Intel Foundry and not the side of the house that we might consider a competitor, and you guys have been terrific,” Haas told Intel Foundry general manager Stuart Pann, with whom he shared the stage.

Gelsinger used the Intel Foundry Services Direct Connect event to show off wafers of the company’s upcoming Clearwater Forest processors, which will be made using Intel Foundry’s online-next-year 18A manufacturing process—with which Intel hopes to overtake TSMC as maker of the world’s fastest chips. But he also announced a potentially major new customer: Microsoft, whose CEO, Satya Nadella, briefly popped up on video to lend his support to Intel in its efforts to build up its U.S. supply chain (Intel’s reportedly in line for $10 billion in CHIPS Act onshoring subsidies) and to mention that Intel will build a Microsoft chip on the 18A process—and yes, that’s about as much detail as we have on that front, though Microsoft has recently said it plans to make an AI chip and a computer processor.

Gelsinger also chatted with OpenAI CEO Sam Altman, who reiterated for the millionth time how the world needs loads more AI chips. Who knows? If Altman succeeds in getting funding for his new chip project, maybe Intel will make the things. More news below.

David Meyer

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

NEWSWORTHY

India forces X censorship. The Indian government has secretively ordered X to withhold certain accounts and posts in the country, X’s global government affairs department disclosed yesterday. As TechCrunch reports, X was not allowed to publish the executive orders themselves, but the company says it is appealing them and has notified affected users.

Google AI news. At the end of last week, it emerged that Reddit had signed a deal allowing a big AI company to use Reddit’s content as training fodder, for around $60 million a year. Now Reuters reports that the company in question was Google. Meanwhile, Google has had to apologize and suspend Gemini picture generations after the AI model started injecting conspicuously ahistorical diversity into certain historical requests—suggesting for example that Nazi-era German soldiers might have been non-white, or that U.S. senators from the 1800s may have included Black and Native American women.

U.S. counters China threat. The Biden administration has announced plans to spend $20 billion on upgrading U.S. port infrastructure over the next five years, specifically to protect them from Chinese hackers. As the Financial Times reports, the president’s executive order on the matter will see more port cranes produced locally.

ON OUR FEED

“The Board’s expansion to Threads builds on our foundation of helping Meta solve the most difficult questions around content moderation. Having independent accountability early on for a new app such as Threads is vitally important.”

—Meta’s external Oversight Board, which adjudicates on tough content-moderation decisions, extends its reach to the Threads app, the company’s new-ish X rival.

IN CASE YOU MISSED IT

In a ‘great signal for creators,’ YouTube star Marques Brownlee adds a corporate gig to his résumé at accessories company Ridge, by Alexandra Sternlicht

Nvidia just crushed earnings again. Top analyst says it’s another ‘drop the mic’ moment that confirms the AI revolution, by Will Daniel

$1.7 trillion chip giant Nvidia just gained over $100 billion in value after a blowout quarter, but this mega-bearish analyst says the tech industry is in an AI bubble, by Paolo Confino

Customer demand for Nvidia chips is so far above supply that CEO Jensen Huang had to discuss how ‘fairly’ the company decides who can buy them, by Kylie Robison

Nvidia’s China sales are down to a ‘mid-single digit percentage,’ as U.S. controls restrict exports of the $1.7 trillion chipmaker’s leading AI chips, by Lionel Lim

Apple unveils Sports, which will put real-time scores on your iPhone’s lock screen, by Chris Morris

Marc Lore’s startup Wonder is opening food halls inside Walmart stores, in a sequel to his e-commerce startup’s $3 billion acquisition by Walmart, by Jason Del Rey

BEFORE YOU GO

Stable Diffusion 3. The race for ever-better AI-generated images isn't slowing down. Today, Stability AI opened its waitlist for an early preview of Stable Diffusion 3, promising "numerous safeguards" to "prevent the misuse of Stable Diffusion 3 by bad actors."

This is the web version of Data Sheet, a daily newsletter on the business of tech. Sign up to get it delivered free to your inbox.