Intel bets on Terafab to help it reassert itself in the AI chip race

As it struggles to keep up in the AI chip race, Intel is forging a partnership that could, literally, have galactic proportions.

The semiconductor giant is joining Elon Musk’s ambitious Terafab project, billed as the largest-ever chip manufacturing facility, one that will power space travel and advanced robotics. Tesla, SpaceX, and xAI announced the joint $25 billion venture in March, with plans to eventually scale out to one terawatt (1,000 gigawatts) of AI compute capacity a year.

Intel announced that it had joined the partnership on Musk’s social media site, X, on Tuesday, saying the goal is to “help refactor silicon fab technology.”

Scott Bickley, advisory fellow at Info-Tech Research Group, said that while details are still scant, that phrasing may indicate the true scale of the initiative, which “implies a potential redesign or improvement of existing methods.”

“This is no small feat, and in light of Intel’s recent struggles with certain chipsets (Raptor Lake and Arrow Lake), it is not short on ambition either,” he noted.

What Terafab is aiming at

During its March unveiling, Musk called the $25 billion Terafab “the next step towards becoming a galactic civilization.” The initiative will launch at a sprawling facility in Austin, Texas, with next-gen chip-building focused on space travel, Optimus humanoid robots, and electric vehicles.

Tesla, SpaceX, and xAI say that Terafab will be the largest chip manufacturing facility ever, outputting 1TW a year of compute power and “combining logic, memory and advanced packaging under one roof.”

Intel’s ability to “design, fabricate, and package ultra-high-performance chips at scale” will help accelerate those 1TW/year ambitions, the company said in its X post. The Santa Clara-headquartered tech giant hosted the world’s richest man over the Easter weekend, and CEO Lip-Bu Tan lauded Musk’s “proven track record of reimagining entire industries.”

“This is exactly what is needed in semiconductor manufacturing today,” Tan wrote in a separate X post. “Terafab represents a step change in how silicon logic, memory and packaging will get built in the future.”

The advanced chip center is targeting 2-nanometer process (2nm) technology, the most advanced chip generation technology in commercial production. It is designed for an initial output of 100,000 raw silicon wafers a month, with a goal to scale to one million wafer starts per month at full capacity.

Musk’s ambitions are to ultimately produce between 100 billion  and 200 billion custom AI and memory chips a year. During the unveiling event, the near-trillionaire said that Terafab will create an “incredibly fast recursive loop” to improve chip development.

The partnership comes as Intel pushes further into advanced chip packaging, where multiple chiplets, or smaller components, are combined onto a single custom chip. Tan has called this a “very big differentiator” for Intel.

Considerations for enterprises without galactic ambitions

That said, it’s Musk we’re talking about here, so enterprise buyers shouldn’t panic (yet), analysts note.

“This should not be a near-term concern for IT buyers, as this seemingly is more of a long-term design, development, and, if it makes it that far, manufacturing and supply chain story,” said Info-Tech’s Bickley.

Should this partnership become more formalized and survive Musk’s “pressures and expectations” (as we have seen, many have not, Bickley noted), the initiative would actually add another potential foundry vector to Intel’s existing roadmap, versus displacing its current product roadmap.

The reality is that foundries are working at capacity through 2026 and likely 2027 at this point, Bickley noted. “It would take some measure of months to years for a net new fab and chip design to approach the pilot stage, making the near-term impact probability for this year close to 0%,” he said.

Tan, who took over as Intel CEO just over a year ago, has been focusing on rebuilding process technology and advancing foundry strategy. This partnership might indicate that he is “striving to land a marquee foundry customer” for Intel, not over-investing resources and capital into Musk’s empire.

“Tan’s messaging since assuming the CEO role has been one of simplification, faster execution, and discipline that flows through the balance sheet, aligning capital structure with strategy,” said Bickley. “In short, this deal would need to be accretive to Intel.”

Longer term, though, should Terafab become a “materially large and technically demanding customer” for leading-edge processes and advanced packaging, it would place stress on Intel, he noted. The company would have to weigh internal product priorities against other high-value, high-visibility opportunities with Terafab.

However, “right now, we just don’t know; there is no term sheet, capacity allocation framework, or other details to go on,” Bickley said.