Ex-Google CEO: China is 6 months behind US on AI and has ‘infinite energy’
Six months is a remarkably thin margin when you consider that just a few years ago most analysts put China years behind the U.S. on frontier AI. Schmidt has been around long enough to know what he's looking at, even if his predictions have been wrong before. The energy angle is the part that deserves more attention than it usually gets. Training the largest models requires enormous amounts of electricity, and if China has cheaper, more abundant power available, that's a structural advantage no amount of clever algorithm work can fully offset.
I'd want to know what exactly he's measuring. Six months behind in what? Research output, model capability, commercialization, deployment across industries? Those are very different gaps. My sense is the U.S. still leads on the cutting edge of research, but China's edge in manufacturing and scale means it can push AI into real-world applications faster. That's the part that should worry people, not the headline number.
Neither country is going to slow down because of a report like this. The real question is whether Washington and Beijing are spending more energy competing with each other than building the infrastructure their own AI sectors need.