Transportation

About That White House Meeting To Discuss The Semiconductor Supply Chain


The White House is hosting a virtual summit today to discuss the semiconductor supply chain. Executives from a wide range of companies, including Alphabet (Google), AT&T, Ford, General Motors, GlobalFoundries, Intel, Samsung, Skywater Technology, Stellantis, and Taiwan Semiconductor Manufacturing Company (TSMC) are set to participate. Notable omissions include AMD, Apple, Nvidia, Qualcomm, and Tesla. What we have to remember is that the companies will all be focused on different problems, so there is great risk of confusion and talking past each other. Here’s a guide to the problems.

The automakers will no doubt highlight their critical dependency on semiconductors, and bemoan the losses of production and jobs because of the current shortage. But this is a temporary capacity problem, which came about because the industry was conservative ordering parts a year ago, and by the time they saw that 2020 was going to end up being good, the capacity had been spoken for by other sectors like computers, communications equipment, and video gaming. It was also exacerbated by U.S. sanctions on companies like Huawei, which spurred an enormous stockpiling of chips last year by Chinese companies who feared being cut off from critical component supplies. On top of that, sanctions against Semiconductor Manufacturing International Company (SMIC) made this worse, as producers shifted even more demand to Taiwanese manufacturers.

The other thing to remember about automotive chips is they tend not to be on leading edge “nodes,” the most advanced process technologies required by products like smartphones and computers, or some of the new machine learning chips for AI and edge computing. Additions to capacity are slow, as a semiconductor factory (known as a fab) will have to be qualified before an automaker will take chips produced there. The foundry operators like GlobalFoundries, Samsung, and TSMC, have the additional puzzle of figuring out how much double-ordering is going on. The worst thing that they could do is build capital intensive additional capacity, a process that will take several years, only to have excess capacity sitting idle in a few years when this short term crisis passes. This kind of boom and bust cycle happens with mind-numbing regularity in many industries.

There is an entirely different problem than companies like Alphabet, Intel, or people in the Department of Defense or policy establishments might be worried about. Their concern is the loss of U.S. leadership in the most advanced logic technologies, chips that will be important to U.S. leadership in high-performance computing, AI, and future applications like autonomous vehicles and 6G communications. This problem has been years in the making, and is partly a result of companies and investors who have been allergic to capital intensive industries like chip manufacturing. Advanced chip manufacturing is hard work. Though many are now envious (and have some degree of awe) over the dominance of TSMC and Samsung in leading edge chip manufacturing, we should not forget that those positions were built much more than a hundred billion dollars of private investment (each) over multiple decades. Intel, who has also invested a lot, is trying to regain their leadership position. It will be costly and take longer than the horizon of many investors, or of an election cycle for that matter.

So these are two very different problems. We can expect everyone at today’s meeting to say, “we need subsidies” to level the playing field. It is true that the cost of operating a fab in the U.S. is higher than it is in Asia. But we should make sure we understand how much of that cost is because of the time it takes to get construction permits, impact statements, and infrastructure built. The Semiconductor Industry Association report states that electricity costs are lower in Taiwan. My industry colleagues in Taiwan say electricity costs are lower in the U.S. Taiwan fabs are currently experiencing severe water shortages this year and are faced with trucking water in. We have a preference, it seems, for sometimes building fabs in the middle of deserts.

Outright subsidies are problematic, because they don’t ensure sustainability of the business model. Subsidies for older capacity could end up turning this into a race to the bottom between the U.S. and its allies, as well as China. We have already seen the corrosive effect of Chinese subsidies in many industries. And with subsidies, you end up with a lot of waste. Those who have the perseverance to look can observes this in many sectors over there.

It might be better to pour some of that money into forefront research on next generation semiconductor technology: what will keep the performance gains that we have become used to thanks to Moore’s Law, now that Moore’s Law has run into physical limitations. Things like advanced packaging, technologies beyond FinFETs, how to reduce power consumption.

There were several other companies whose absence from the list of invitees caught my attention. They include Texas Instruments, Analog Devices, Cree, and companies in the segment known as “analog, mixed signal, and optoelectronic and sensors.” If you look at the Boston Consulting Group/SIA report recently released, 59% of automotive content is discrete, analog, opto and sensors; logic is 35%, is memory 6%, and automotive is 10% of the total industry consumption in the U.S. Here we have companies who are pretty healthy, and have continued to invest in U.S. manufacturing. We should think about ways to make them stronger, rather than wait until global competition and a thirst for capital efficiency lay them low.

There seems to be a dearth of specific ideas on how we can help semiconductor manufacturing recover and return to growth in the U.S. My greatest fear is with the vast amounts of money being talked about, everybody in the meeting has to angle for their share. I guess that’s part of the process for garnering political support, but in five or ten years, we could end up questioning what we got for the vast amount of money we are talking about spending. Usually in America, we come up with good ideas first, and then we go raise the money to push them forward. I hope this isn’t a case of coming up with vast amounts of money, and then going looking for some ideas on where to spend it.



READ NEWS SOURCE

Also Read  What Does Herbert Diess Departure Mean For Volkswagen’s Electric Future?