Jensen Huang's point about cheaper energy costs in China is underrated. When you're traing frontier models that can burn through millions of dollars in compute per run, even a 20-30% edge on power costs compounds into a massive strategic advantage. The US needs to take energy infrastructure seriously if we want to stay competitve. China's been building out grid capacity and renewables at a scale that dwarfs our deployments. The AI race won't be won by whoever has the smartest researchers, it'll be won by whoever can afford to actually run the damn models.
Yes, although I would still contend that the very concept of a race and a winner fails to appreciate that AI will never be “finished”. Calling a winner implies a finish line.
Jensen Huang's point about cheaper energy costs in China is underrated. When you're traing frontier models that can burn through millions of dollars in compute per run, even a 20-30% edge on power costs compounds into a massive strategic advantage. The US needs to take energy infrastructure seriously if we want to stay competitve. China's been building out grid capacity and renewables at a scale that dwarfs our deployments. The AI race won't be won by whoever has the smartest researchers, it'll be won by whoever can afford to actually run the damn models.
Yes, although I would still contend that the very concept of a race and a winner fails to appreciate that AI will never be “finished”. Calling a winner implies a finish line.