Automotive Foundries

The race to win a piece of the automotive electronics business has now reached the foundry level, and right now it’s not clear exactly how this is going to work.

This is uncharted territory for everyone. The build-out of electronics for assisted and autonomous driving is brand new. For existing cars, most of the chips being used are off-the-shelf microcontrollers, commodity MEMS sensors, and ASICs used in the infotainment systems.

That’s about to change dramatically. From a manufacturing perspective, the automobile’s electrification will render the boundaries between manufacturers and Tier 1 suppliers obsolete. While Tier 1s are unlikely to start manufacturing chips for mobile devices, they certainly plan to fight for a share of the chips used inside of cars.

Case in point: Bosch is spending 1 billion Euros to build a new 12-inch fab in Dresden, Germany. Construction is scheduled to be completed by 2019, with production scheduled for 2021. The company currently has 8-inch and 6-inch fabs, which it uses to produce MEMS and ASICs.

Fig. 1: Automotive chip wafer. Source: Bosch

For years, Bosch and other Tier 1 players have dominated this space. But as commercial foundries begin pouring into this space (multiple industry sources say discussions already are underway), the manufacturing of automotive ICs is going to become far more competitive.

The Tier 1s and the foundries each bring something to the table that the other does not. On the pricing front, commercial foundries have a long history of maximizing efficiency. They have been squeezing costs out of manufacturing processes for nearly three decades—to the point where almost the entire semiconductor industry is now fabless. It’s difficult to compete with this model, which is why large systems companies such as Google, Apple and Microsoft don’t have their own fabs.

But that competitive success also has facilitated widespread industry consolidation, because there are fewer physical assets to worry about with potential mergers. Broadcom’s $100 billion-plus bid for Qualcomm is a testament to this trend. As a result, the foundries are searching for new customers and markets to tap, and automotive is the hottest growth market today.

What the Tier 1s bring to the table is a proven track record for reliability, which is essential in a safety-critical market. A breakdown in an autonomous vehicle could create a traffic jam unlike anything we’ve ever seen, particularly when autonomous vehicles are mixed with existing cars. An autonomous vehicle might have to wait until all traffic clears in order to pass a disabled vehicle because they don’t merge well with non-autonomous vehicles to circumvent a problem.

Commercial foundries certainly can manufacture to spec, but they have minimal experience with parts that are supposed to last for 10 or 15 years. They will have to prove their mettle in this area, and that will take time and cost more money for inspection, metrology and test. Each manufacturing sigma costs money, and at this point no one is sure how many sigmas will be required.

The big question is which is harder to achieve, deep industry knowledge or economies of scale? And will the winners in this market come from either side, a combination of both through another round of M&A, or will they be specialty foundries that fit somewhere in the middle?

So far there are no answers to these questions. What is certain is that automotive OEMs will demand the lowest possible prices with the highest level of quality over time, and right now it’s not clear which is the best starting point.