The risk model is different. The cost of setting up an gorgeous tech company office in pre-Covid San Francisco is trivial compared to setting up even poor-quality fab. Not that we (the US) should shy away from doing it - the Silicon Valley is called that because semiconductors are made out of silicon, but SV is also full of Superfund sites (for many reasons, but IC manufacturing isn't a light industrial process, and is ridiculously capital intensive. A few million dollars, and a few months, and that social media company will have an MVP and started proving traction in their chosen niche. (It'll take closer to a decade to actually get somewhere with it eg tiktok, but MVP can be demoed to investors far sooner.) A few years, and closer to a billion dollars, and you might have broken ground on the fab.