Chipmaker TSMC's Europe factory plan gives it a new global dimension

The two regions have a few things in common when it comes to semiconductor strategy. Both wish to rekindle their perceived loss of chip prowess. And governments on either side of the Atlantic are willing to spend tens of billions of dollars (and euros) to lure the biggest and most-powerful players to their shores. In each case, local authorities are taking a lead role in doling out incentives such as tax breaks and cheap land.

A key purpose of the Arizona project is to help the US catch up on leading-edge semiconductor technology after having once been the global leader. Yet in Europe, the goal is to ensure continued supply of crucial components to the auto industry, an economically and politically important sector for many countries in the region. Crippling shortages through the Covid-19 pandemic highlighted how important chips are for carmakers.

In 2020, the state won the grand prize when TSMC announced it had chosen a site near Phoenix as home for a $12 billion project that would make almost-leading-edge chips (Oregon was believed to be short-listed). That program was later increased threefold and there’s a good chance further expansion will be announced.

TSMC doesn’t make $40 billion commitments on a whim. Its decision to invest an unprecedented sum into what amounts to a greenfield project comes because key clients such as Apple Inc. and Nvidia Corp. told the chipmaker, either implicitly or explicitly, that they will buy from these new factories. Their chips are largely logic processors: components that act as the brain of devices ranging from iPhones to artificial intelligence servers. Washington politicians and industry leaders are hoping that one day the world’s best chips will be made on US soil. They won’t. Not by TSMC, anyway.

Europe is different, and TSMC’s strategy there makes more sense. Even though the Taiwanese company introduced its 28-nanometer process 12 years ago, it is expected to be at the heart of the chipmaker’s European strategy. There’s two reasons for this: most automotive chips still use that node, and TSMC has pivoted this technology to more niche applications instead of processors. As industry observer Paul McLellan once wrote: “Old TSMC fabs don't die, they become specialty technology fabs.”

Whereas 28nm was used to make the A7 chip for Apple’s iPhone 5s in 2013, today it’s commonly deployed to manufacture sensors (eg LiDAR, used to detect a car’s surroundings), microcontrollers and radio frequency transmitters. A modern car includes semiconductors to detect traffic, control the side mirrors, and monitor battery levels. That’s not to say newer processes aren’t also found in cars: TSMC’s 5, 7 and 16nm offerings are also deployed to make components for advanced driver-assistance systems, infotainment and control circuits.

Apple, Nvidia and Advanced Micro Devices Inc. are eager to shift to new manufacturing techniques annually because that allows their own products to be more powerful, a key selling point for logic processors. Automotive chip designers such as NXP, Infineon and STMicroelectronics NV, on the other hand, work on longer product cycles which require detailed safety checks and integration across more complex systems. There’s far more electronic components in a car than a smartphone, and the automobile industry upgrades their vehicles at a slower pace. In their case, newer is not always better.

Even though this older technology only accounts for around 10% of TSMC’s revenue, compared with more than 50% for processes introduced within the last seven years, the chip giant is still a global leader by market share. A belief that electric vehicles and continued deployment of sensors and electronics in cars is a long-term trend justifies a bet on Europe. Having local partners and governments cover much of that wager makes it the kind of project TSMC will find hard to resist.