A recent reflection, based on conversations with policy leaders, is that there are two superficially similar but importantly different perspectives one can hold with respect to US manufacturing:

Affinity for manufacturing and physical production is an anachronistic fetish, embodied by populists with outdated attraction to hard hats and clanging forges. A great deal of manufacturing has departed the US, which is certainly fine and probably even quite good. It’s unpleasant labor, and countries ought to each specialize in their respective comparative advantages.

Manufacturing is the ultimate network effects and economies-of-scale business. As services are substituted by AI, and as datacenter deployment accelerates, the relative importance of manufacturing is likely to grow. To think that one can pick and choose sectors in which one will excel (“let’s win in drones but not in dishwashers”) is a fallacy. Manufacturing is hence of paramount strategic importance. However, we don’t know how to make the US the world’s preeminent manufacturing power (given its cost base and given the current center of gravity in China) -- indeed, we don’t know whether it’s even possible -- and this is a significant strategic problem for the country.

I have zero direct expertise here, but my outside view is closer to #2 than #1: it seems that the ecosystems and supply chains create strong gravity across the board. I also asked @elonmusk, who has clearly done more over the past decade to advance sophisticated US manufacturing than anyone else, and this appears to be his view. Most economists, on the other hand, are much closer to #1, and I don’t think that the economics profession considers the absence of good ideas for reviving US manufacturing to be a problem of particular significance. (There are lots of snide epithets about the efficacy of industrial policy.) It seems to me that there’s even some amount of backwards reasoning happening, where, because we don’t know how to do #2, #1 is subconsciously a much more comfortable position to hold.

Talking about winning particular manufacturing sectors feels to me a bit like talking about winning individual biological research sectors or winning particular software sectors. That is: it seems that the strong default assumption should be that “the place that is best at biology research sector X will also be best at sector Y”, and similarly in software, because the skills and inputs needed are so transferable. As such, my guess is that if the US seeks meaningful sovereignty or preeminence in any of drones, robotics, solar, batteries, pharma, etc., we need to bite the bullet, and win at manufacturing across the board.

Overall, I’d love to read more arguments for and against these perspectives, particularly from those with direct expertise.