Can anybody save Intel?

INTEL has spent two decades missing the next big thing. The chipmaker’s dominant PC business blinded it to the opportunity from mobile phones in the 2000s. More recently, the firm was slow to adopt extreme-ultraviolet lithography, an expensive chipmaking process that was originally funded by Intel itself.

Now, Nvidia dominates the white-hot market for designing artificial intelligence (AI) chips, becoming the world’s most valuable semiconductor company. Investors in Intel have voted with their feet.

As when any corporate icon falls on hard times, dealmaking rumours are swirling. Qualcomm, an American chip designer, is reported to be interested in buying Intel. Apollo, a financial firm, is also mulling an investment.

Any buyer must confront a vexing problem. Intel’s manufacturing business, or “foundry”, is viewed as strategically important by American policymakers, who want more chips to be made at home. But it is also deeply unprofitable. Enormous and relentless investment is required for it to compete with TSMC, a Taiwanese chipmaking giant.

The story of Intel is a marvel of American engineering. The firm’s survival now requires a financial-engineering miracle, too.

Pat Gelsinger, Intel’s boss, acknowledged as much on Sep 16 when he said that Intel Foundry would become a distinct subsidiary with its own board. The firm’s separation of church and state should convince potential customers that Intel’s manufacturing arm is not entirely captive to its chip-design division.

At least that is the theory. Only 1 per cent of Intel Foundry’s revenue came from external customers during the first half of this year. A splashy announcement that Intel will make custom AI chips for Amazon’s cloud-computing arm has failed to convince many people that it can leap from making its own chips to ones for outside customers, as TSMC does.

“I’m like five foot six (1.7 m) and 50 years old, and even if all the politicians in the world would love for me to play in the NBA (National Basketball Association), it’s probably never going to happen,” said Christopher Danely of Citigroup.

Without profits to reinvest – and with US$53 billion of debt already – Intel relies on a growing pile of subsidies and private financing. The firm has been promised more than any other under America’s Chips Act, legislation passed in 2022 to boost domestic production.

On Sep 16, it was awarded up to US$3 billion to make chips for the armed forces, in addition to as much as US$8.5 billion of grants and US$11 billion of loans announced earlier this year. In June, Intel said it would finance a plant in Ireland through a joint venture with Apollo, which has a big life-insurance arm.

“Intel has bank debt. Intel has public bonds. And now, Intel has US$11 billion of investment-grade private credit,” said Apollo’s boss of the deal. What the chipmaker does not have, to the torment of its increasingly subordinated shareholders, is a credible plan to turn a profit.

Neither America’s government nor its financiers can fund Intel forever. But beyond firing workers and delaying projects, it has few options to raise cash. One may be to sell Altera, the programmable-chip business it bought for US$16.7 billion in 2015. It could offload its majority stake in Mobileye, though the automotive technology firm’s valuation would surely reflect the current troubles in the carmaking industry.

A radical deal involving the full separation of Intel Foundry is hard to imagine, given its precarious financial position, even in the unlikely scenario that potential customers decided to invest in the business.

What about a full takeover? An acquisition by Qualcomm, which designs chips for phones, would be the largest in the industry’s history. It would produce a chipmaking giant – call it Qualtel, Incomm or Americhip – with US$90 billion of annual sales, and create a huge new customer for Intel Foundry.

For American regulators, the perceived security benefits of a stronger combined firm could allay concerns about antitrust. “I think the US government would be a huge proponent of the deal – it would create a massive US-centric company that they can then throw a lot of support to,” said Angelo Zino of CFRA Research.

Yet, any deal would be hard to pull off. Qualcomm has no manufacturing experience, designs its chips using the architecture of Arm, a British rival to Intel, and would struggle to afford such a deal – it has US$13 billion of cash and securities, and its market value is less than twice that of Intel. Regulators outside America would also baulk at a tie-up.

Intel recently shelved a project in Germany, in effect scuppering Europe’s chipmaking ambitions; the continent’s regulators will be in no mood to help. Nor will those in China, as Qualcomm’s board already knows – in 2018, Qualcomm abandoned a US$44 billion deal to buy NXP, a Dutch manufacturer, after failing to win Chinese approval.

That would leave Intel returning to the drawing board, by which time things may have gotten worse. Inaction, however, is not an option.

@2024 The Economist Newspaper Limited. All rights reserved