It's obvious that there are lines of flight (to take a Deleuzian tack, a la Land) away from the current political-economic assemblage. For example, a strategic nuclear exchange starting tomorrow (which can always happen -- technical errors, a rogue submarine, etc.) would almost certainly set back technological development enough that we'd have no shot at AI for the next few decades. I don't know whether you agree with him, but I think the fact that he ignores this fact is quite unserious, especially given the likely destabilizing effects sub-AGI AI will have on international politics.