Paul van Gerven

7 April

Yes. No. Maybe. Let’s not rush into things.

If European Commissioner Thierry Breton wasn’t a religious man already, he might have become one after he got a call from Intel the other day. As the driving force behind the European Union’s ambition to establish cutting-edge semiconductor manufacturing on European soil, Breton has desperately been trying to get Europe’s chipmakers and research institutes on board, only to be told that his goal is unrealistic. The research institutes are at least sympathetic, no doubt partially because they wouldn’t mind more funding, but even they know that one simply can’t skip multiple nodes. Europe first has to maneuver itself into a position in which building a cutting-edge fab actually makes sense – technologically, but also from a demand perspective. That process will easily take a decade.

How about luring one of the world’s foundries to Europe, then? After all, both Samsung and TSMC were recently persuaded to build fabs in the United States. “In the US, we committed to building a fab after the authorities made clear that they would subsidize the cost gap. But in Europe, the case isn’t that strong, and the Europeans really should figure out what exactly it is they want, and whether they can maybe achieve it with their own chipmakers,” a senior TSMC executive told the Financial Times. Ouch.

And then, all of a sudden, the world’s biggest semiconductor company pledges support for your cause, offering to build a foundry in Europe on top of the already ongoing expansion of the fab in Ireland. “We have a shared ambition with the EU to deliver state-of-the-art semiconductor technology to Europe and create a more geographically balanced manufacturing capacity. Intel is uniquely positioned to support the EU’s vision for a digital transformation by 2030, thanks to our technology leadership, continuous investment in our manufacturing and development of future foundry capabilities in Europe,” Intel Ireland’s general manager, Eamonn Sinnott, wrote in a press statement.

It would make for an interesting alliance: both Europe and Intel have been backed into a corner recently. For Europe, the blanket of the liberal international trade order doesn’t seem to be as comfortable as it used to be. Feeling the cold creep in, the European Commission, as well as France and Germany, have started pushing for ways to reduce Europe’s technological dependence.

Intel, meanwhile, has dropped the ball in process technology. After leading the semiconductor industry for a decade or two, it has forfeited technological leadership to TSMC. Adding insult to injury, the once X86-dominated leading-edge semiconductor landscape has become a much more varied ecosystem. Arm, GPU and other architectures are slowly encroaching on Intel’s bread and butter.

Faced with these existential threats, Intel was about to wave the white flag. Pressured by investors, it even hinted at moving out of manufacturing entirely. The new CEO, Pat Gelsinger, however, went on the offensive recently, unfolding a bold (some say foolish) new strategy to take Intel back to the top. This includes bolstering manufacturing operations and starting an independent foundry business offering not only X86 but also Arm and Risc-V cores and other IP.

Intel Ocotillo Fab 42
Credit: Intel

Massive

Given its struggles of late, the big question is whether Intel really is the godsend Breton has been hoping for. The company has to get its manufacturing woes sorted out, make up ground technologically, get customers to sign up and build up scale while at the same time keeping those fabs filled. And that’s while battling a formidable opponent that has executed flawlessly over the past years and that isn’t exactly going to sit on its hands.

The biggest challenge by far, however, is successfully setting up a foundry. This will require a massive culture reset. Historically, Intel has never been good at anything else than doing one thing and doing it very well. That’s a long way from the mother of all customer-centric business models. On top of that, Intel’s recent track record likely doesn’t inspire confidence with potential customers. And how eager will they be to share their sensitive designs, especially since many compete with Intel in one way or another?

There are also many things working in Intel’s favor, of course. Many believe that the upcoming Silicon Everywhere era is more likely to create a semiconductor supply squeeze than a surplus. Governments are doling out a lot of money to anyone willing to build fabs in the West; for the US government, in fact, Intel may be too strategic to fail. And, surely, the Western fabless community wouldn’t mind if TSMC got some serious competition – Samsung is hanging in there, but it doesn’t challenge the Taiwanese (hey, here’s a crazy idea: what if Intel teams up with Samsung?).

Technologically, even if Intel can’t catch up with TSMC in density any time soon, there are plenty of opportunities to compete on performance by introducing disruptive transistor and advanced packaging technologies. Intel now also has IBM’s innovation potential to draw upon: the two former rivals have forged an intimate R&D partnership. IBM may not be in manufacturing anymore, but it still has some excellent semiconductor R&D to offer – just ask Samsung.

Complementary

All in all, Breton would do well to not immediately give in to Intel’s charms. Better be patient and see how things progress – getting 7nm out without any further delays might be a good start. Europe may be desperate, but there’s no reason to be. Getting a leading-edge fab somehow has become the focal point of the technological sovereignty ambitions, but the fact of the matter is that Europe’s industries don’t even need that many leading-edge chips right now. It still makes much more sense to invest in semiconductor technologies that European (and other) industries will actually need.

That said, European demand for leading-edge chips is bound to increase, considering future applications such as automotive electronics, communications infrastructure, AI and Industry 4.0. Being able to manufacture them ‘in-house’ would be highly desirable, for which a leading-edge partner is a virtual sine qua non. As long as it’s complementary to evolving Europe’s own semiconductor base and semiconductor-using industries, and it’s looking like Intel might actually pull off the stunt of the century, it would be worth risking some taxpayers’ money to fund a European Intel fab.

Main photo credit: Intel