AI Regulations Risk Extreme Concentration Among Large Firms

At some level of regulation things may look extremely concentrated where frontier capability only applies to large firms with enough capital/political favor/existing market advantage/compliance capacity and everyone else is out of luck. Stocks would go up but relative economic mobility of people and firms would go to zero for almost all

Also seems to increase the likelihood that labs will just choose to internalize all the economic gains of ai, run ahead of legacy companies wherever possible, etc (ie a more extreme version of the above)

Tbh I’d be surprised if a few regulations hold back the development of the next 1-2 model generations (though the broader public may never see them), but beyond that capital might slow if tam is truly bounded by stringent frontier regulations and the market balance between commodity and frontier stays roughly how it is now (or gap narrows)

Fair, but that said, how many model generation do you think we have left until we get where we’re going? When the USG has line-of-sight on the machine-making-machine I think capital considerations ~stop coming into play

Money is good, and the state has plenty of it, but it’s fundamentally just a proxy for power, of which the state has even more. If the powers that be come to the (correct, post-RSI) conclusion that the only thing they need to win is more GPUs, then they’ll get more GPUs, TAM be damned