Thanks for the comment.

Thanks for the comment. Your question about where the energy would come from was a good one, good enough that I wrote a full follow-up to answer it: https://medium.com/@brian-curry-research/there-is-plenty-of-power-for-ai-the-shortage-story-is-misinformed-or-self-serving-4debd2e075b6

The short answer is that there is plenty of power. Solar is a big part of it. It is already the cheapest new power in most of the world, with installed costs down 87% since 2010, and it is spreading because it is profitable, not in spite of profit. Add wind, batteries, nuclear restarts and uprates, geothermal and data centers that flex their demand, and the supply is there. The real problem is who pays for it, not whether it exists.

Where I'd push back is on the rest.

AI does not have unlimited free energy, and it can be switched off. Every model runs on physical chips in data centers that buy metered electricity. A typical chatbot prompt uses about 0.24 watt-hours, and operators already throttle AI workloads when the grid is stressed. In one test in Phoenix, a GPU cluster cut its power 25% for three hours on request. That is the opposite of uncontrollable.

AI also does not learn whatever it wants. Today's models do not keep learning on their own after deployment. Their weights are fixed when training ends, and that training is designed, funded and run by people. They talk to us through text we type, not telepathy.

On deity-level intelligence and a superorganism, I'd suggest reading Yann LeCun. He has argued for years that current language models lack a real understanding of the physical world, persistent memory and the ability to plan, and that the doomsday framing is not supported. A lot of the fear comes from frontier labs, tech commentators, media and officials who each have a stake in a dramatic story.

Where I agree with you: power over AI is concentrated. But it sits with the companies that own the chips, the data centers and the power contracts, not with the AI itself. That is exactly why the rules matter, and why I think AI can be the most amazing thing to happen to people, if we design the economics so the gains are shared. That was the point of The Price of an Hour, and the energy piece shows the power side of it can work too.

That's where I land after 10 years working in AI and energy. Cheers.

Related research