So what is the right model?

Yesterday we ranted a bit about proposed US government industrial policy for semiconductors. A well known strategy commentator messaged us to ask essentially “If not this, then what?” Which is a reasonable question, unfortunately one without an easy answer.

The US is unarguably the world’s leader in fabless semiconductor design. Factor in Intel (who admittedly need a bit of work), and companies like Applied Materials and ASML (not a US company, but close enough for our purposes here), and the US is the clear leader in “Semiconductors”. By contrast, China is clearly looking to reduce its reliance on US semiconductors, and arguably seeking to challenge the US directly. What would be a sensible strategy for the US to respond?

We argued yesterday that direct subsidies of US chip companies was problematic on a number of fronts and seems unlikely to affect the change needed.

That being said, there a number of other policies which might be more productive:

  • A national semis venture fund. This would be a direct response to China’s approach. The US government does not have much expertise in running such a fund, and it would run in to the same problems of direct subsidies. So a better approach might be:
  • The US government would pledge to be the leading LP (Limited Partner) investor in a Semis Venture fund . The distinction is one of governance. Instead of having the civil service make investment decisions, let people with technical and venture experience make them. Pay them well, but based entirely on fund returns. Even better, the government should invest in a number of funds with competing fund managers. True, determining the recipients of these LP investments could be subject to political pressure, but if done in a transparent manner, with compensation determined by the funds’ returns, we could reduce that source of friction. To be clear, this is actually the model that the Chinese government has implemented.
  • The US government could position itself as a major customer for new semis companies. This one is complicated but reflects the actual origins of much of the US electronics industry going back to World War II and the Vietnam War. By consolidating semis purchasing and putting out some ambitious goals, the government could use commercial mechanisms to guide investment in new materials and new chip designs (e.g. AI). The major chip companies already sell a lot to the US government, what we are proposing here is more coordination on the government’s side. Recognizing that strategic interests may trump traditional bidding parameters and consolidating orders in a directed fashion. Not easy and lots could go wrong, but achievable.
  • A massive increase in research grants to universities to encourage semis R&D. As above, the government already does a lot of this, but could benefit from a clear, unified strategy. This is unlikely to result in near-term results, but could be a big boost over the long-term.
  • Immigration – we try to avoid politics on this site, so we will try to approach this from a strictly commercial point of view. If the world wants to send its best and brightest to contribute to research here, that seems like a major gain for the US overall. The trick with this is ensuring that all that talent stays here, and we do not end up training some other country’s future workforce. There is no reasonable guarantee that this won’t happen, but tying visas to longer-term commitments to remain in the US, or some other mechanism, this could go a long way to keeping that talent. Our sense is that most people who come to the US to get advanced scientific degrees would prefer to stay here. The government just has to make that easy for them to do.
  • Deeper structural reform. This one is a bit abstract. There are numerous policies that have led to a decline in US manufacturing and R&D over the past 20 years. These include among many others the position of the US dollar relative to other currencies (notably Taiwan and South Korea), and tax policies which still offer big incentives for companies to offshore much technical activity. We recognize that these are not easy nor quick to resolve, but acknowledging that past policies have had consequences which have led us here would be a good first step in addressing them.

The unifying philosophy behind all of these is to remove the government from the role of picking winners, a task better left to the market and to academia. Instead, the government should harness and organize what it is already doing to act in a more coordinated manner, and leave other parts of society to do what they do best.

2 responses to “So what is the right model?

  1. Pingback: Is America losing the R&D race? | DIGITS to DOLLARS·

  2. Pingback: And then there were ten | DIGITS to DOLLARS·

Leave a Reply