Intel’s foundry shake-up doesn’t go far enough

Intel has been under pressure from some investors to get out of the money-losing business of making chips for others. It announced earlier this month that it was turning the foundry division into an independent subsidiary, a move designed to unlock growth by reassuring foundry customers that their manufacturing concerns wouldn’t be subordinated to those of Intel itself.

That may fix some problems that were inherent in the old structure. But it has the feel of a hedge, and probably doesn’t go far enough.

One complicating factor is the foundry’s role in public policy, which can be difficult to reconcile with the demands of investors.

The U.S. is investing billions of dollars into Intel and many other companies as it races to reduce its overreliance on other countries for chip production. 

The Commerce Department has proposed as much as $8.5 billion in direct funding of Intel, for example, to support factory construction projects in Arizona, New Mexico, Ohio and Oregon. And Intel said on Sept. 16 that it had been awarded up to $3 billion in separate funding under the Chips and Science Act for the government’s Secure Enclave program, “designed to expand the trusted manufacturing of leading-edge semiconductors for the U.S. government."

Rebuilding domestic chip manufacturing capacity is important, for reasons of national security and competitiveness, as well as job creation. But that’s a long-term goal, with a mission that transcends profitability. Intel right now is under pressure from investors to immediately fix huge challenges in its products business—the non-foundry operation—such as taking leadership in the design of advanced chips for AI.

“Their biggest problem is foundry. There’s a very small chance it will succeed," Christopher Danely, semiconductor analyst at Citigroup, said in an interview.

The foundry business has never made money, losing $5.3 billion in the first half of this year, $7 billion in 2023 as a whole and $5.2 billion the year before.

Intel said in April that foundry’s operating losses are expected to peak in 2024, and that it is looking to achieve break-even operating margins midway between now and the end of 2030. Its goal is to become the world’s second-largest foundry by 2030.

Intel’s margin for the second quarter was 35.4%, meanwhile, down from 35.8% in the second quarter of last year. Rival Advanced Micro Devices reported a gross margin of 49% in the second quarter, up from 46% in the second quarter of last year.

A recent multibillion-dollar agreement for Amazon.com’s cloud-computing arm to manufacture AI “fabric" chips at Intel factories provides some validation of the current foundry strategy, Danely acknowledged. “It’s a mild positive, I would admit," he said.

Should Intel exit the foundry business, however, its gross margin would exceed 50%, according to Danely. Barring that, he argues that the foundry business will continue to lose money, diverting engineering talent from more promising lines of business and depressing morale. He also argues that the company’s internal manufacturing culture doesn’t lend itself to the production of a more diverse array of products for outside customers.

Intel has said it is developing “the service-oriented mindset" required to succeed as a foundry.

“Intel’s capabilities across design and manufacturing are a source of competitive differentiation and strength," the company said in a statement Wednesday to The Wall Street Journal. “We are fully focused on executing the plan we have laid out to drive the next phase of our transformation and build a stronger Intel for the future."

Intel Chief Executive Pat Gelsinger argues that the products and foundry businesses are distinct operations but “clearly better together," as he said during a Sept. 16 interview with CNBC. The company has said the products and foundry businesses complement one another for several reasons, including the ability to co-develop technology, get products and services to market faster, and achieve greater scale.

The foundry business has different operating requirements, Gelsinger said on CNBC, “and we need to have the operational skills to make that great."

Rebuilding domestic semiconductor manufacturing capacity in the U.S. is critical. In an ideal world, Intel would play a major role in that project. But the reality is that the greater good may be best served by allowing Intel to get its own house in order before it takes on the broader interests of U.S. security.