The US government continues to toy with ways to “encourage” the US semiconductor industry. We think this is a bad idea at any time, and a particularly bad idea right now.
This was brought to our attention via this article in Politico published yesterday. The framing of this article seemed written to trigger us as it asserts that the US needs to subsidize chip makers to ‘win the race to 5G’. Following on TSMC’s heavily subsidized plan to open a small fab in Arizona and reports of legislation for “tens of billions of dollars” to subsidize US semiconductors R&D and manufacturing.
This effort is misguided on two fronts.
First, the devil is in the details and those are short on the ground right now. Having read and watched hours of content on this topic it is clear to us the US policy makers do not have a good grasp on how the semiconductor industry works. The distinction between foundries and fabless, academic R&D vs. corporate R&D, all seem to get regularly conflated. But the details matter. We have written on this in the past and will not delve into here, but put simply, the US semiconductor industry is in really good shape right now, world leading in almost every category. Why do companies enjoying record stock prices and margins need government support? It would probably help those companies more if the government just spent the money to protect workers more.
Absent an understanding of the industry, industrial policy risks sliding into a form of crony capitalism, with the most “persuasive” rather than the most deserving winning subsidies. We can say this with a high degree of confidence because this is how the US semiconductor industry has been built over 50 years, with a general absence of any subsidies.
Of course, there is the very real competitive threat posed by China’s massive subsidies of its semiconductor industry, largely in the form of unlimited capital for venture/private equity deals. We are not anarchists or libertarians, and acknowledge the need for the US government and US industry to offer a coordinated response to this. Blindly throwing money at companies with the best lobbyists is not that.
The US is not lacking in semiconductor companies, and contrary to that first Politico story, it is not missing a “5G chip”, whatever that is . Buying Nokia and/or Ericsson is not going to solve those companies’ disadvantages against Huawei. We imagine Intel would appreciate a subsidy for its next fab, but that is not going to make them competitive in 5G. They already spent $20 billion on that effort and failed. Qualcomm might benefit from a few billion dollars, but how much of that would end up boosting their 5G prospects (which already look pretty good) and how much into further inflating real estate prices in Del Mar? Would a US government semiconductor venture fund be any better at finding commercial opportunities than the 1,000+ US VCs who have not done a semis deal in over a decade?
Instead, we think the US government’s time is better spent looking at structural reasons for the overall decline in US manufacturing over the past decade+. And to the government’s credit, China’s stance towards US semis companies is a central complaint in the ongoing trade war. But there are other issues like taxation of foreign subsidiaries, the strength of the US dollar and the fragility of our patent system, which could all withstand a thorough evaluation.
And this leads to our second problem with the current government thinking. The US has done a pretty good job of building a leading economy and an incredibly robust technology industry. Those obviously both face their own problems today, but it strikes us as folly to think that the solution to these problems is to entirely abandon the principles and fundamentals that the industry is built on. As we noted in our piece on China’s semiconductor industry dynamics, it is ironic that just as China is seeking to emulate the way the US built its semis industry the US is abandoning those principles. China’s efforts are formidable but not guaranteed to succeed. The US is in a strong position in semiconductors and should not sacrifice that with ill conceived policies.