We have more measures of peak AI intelligence than of AI's proliferation through the economy but the latter matters quite a bit now (new growth is roughly the area under the curve of intelligence x proliferation, and the latter seems like the marginal opportunity) Re: USA vs China, I'd bet we proliferate AI faster in the end but it's not obvious––there's vast inefficiency in our efficient markets (firm productivity dispersal is very high) and an accelerationist Politburo could mandate every firm use DeepSeek Code tomorrow, a classic in the ELO contest of freedom vs command