Automotive market realities and uncertainties, and what it all means to semiconductor design and manufacturing.
An estimated 74.39 million automobiles are forecast to be sold this year, according to Statista. That’s up about 2.8% over 2015, which on the surface doesn’t look like fabulous growth.
What isn’t apparent in the numbers, though, is the amount and type of semiconductor content. Electronic control units, which are primarily driven by MCUs, increasingly are being replaced by SoCs. Automotive companies are even making inquiries about finFETs these days, which means that ultimately there could by an average of 8 to 10 SoCs per car rather than ECUs, at the most advanced process nodes. Moreover, because these chips have to last longer, they will include more rigorous design and verfication/debug methodologies, with added testing post-manufacturing.
Given that the advanced smartphone market is flattening, automotive chips could make up some of the difference in volume, and all of the difference in price. While price is always an issue for big systems companies—in this case the automakers—it’s not nearly as price-sensitive as a smartphone. Adding an additional $1,000 worth of electronic content to a car is a much smaller percentage increase than in a smartphone, or even a server, and the number could go considerably higher as vehicles move from assisted to autonomous driving sometime in the next decade or so.
Not all of this will be at the most advanced process nodes, of course. Automotive companies always are weighing what the market will bear and where to add costs. In the ECU world, they generally choose specialties that define their brand, such as engine control, infotainment, comfort level such as variable suspension, and fuel efficiency. With ECUs, no carmaker could afford to be all things to all consumers. Whether that changes with the shift to SoCs remains to be seen, because putting more things on a chip, or at least in the same package, can add economies of scale that never existed in the automotive market.
This helps explain why there is such a scramble among foundries and OSATs to beef up their capabilities across a wide spectrum of processes, materials and and packaging. It also positions them for other market slices that fall under the connected-everything/IoT umbrella, which will be narrowly defined, purpose-built devices that in aggregate could greatly accelerate growth across the semiconductor industry.
Automotive is the biggest cohesive market opportunity in the short-term, but the capabilities the chip manufacturing sector is making ultimately will apply to all of those sectors. How fast automotive grows remains to be seen. There is no clear understanding about when autonomous vehicles will begin reaching the market en masse, or how quickly advanced semiconductor content will begin infiltrating cars and trucks. Some of that depends on government regulations, proof of reliability in real-world harsh conditions, and how quickly automakers are willing to shift to new technology.
There used to be a saying on Wall Street in the heyday of the U.S. auto industry that when Detroit sneezes, the rest of the country catches cold. At that time, more than 50% of the economy was tied to the automobile industry in one way or another. Those numbers are far lower today. Car companies are scattered around the world and fueled by a global supply chain, and economies are much more diversified than in the past. General Motors is no longer the largest company on the planet. It now ranks 23 on the list of largest public companies. (Wal-Mart is No. 1, and six of the top 10 are oil and gas companies.)
But as electronic content increases in cars, and in other markets, this list could change significantly—and so could the list of acquisitions across markets rather than within them. Change is coming, and the chip manufacturing and design sectors are repositioning themselves to prepare for it. The big race has begun.