AI as a Grid Bully: Who Gets Throttled First? скачать в хорошем качестве

AI as a Grid Bully: Who Gets Throttled First?

Article link: https://open.substack.com/pub/johanos... The “software is eating the world” story is giving way to a harsher reality: AI is colliding with physics. Modern AI isn’t just a clever layer of code; it is an industrial system that converts enormous amounts of electricity into heat, at extreme density. That is why data centres are shifting from simple air-cooling to complex liquid-cooling, why water access is becoming strategic, and why energy availability increasingly shapes where AI capacity can be built. As AI becomes embedded in “always-on” services, reliability stops being a nice-to-have and becomes a core economic requirement. The deeper risk is political, not just technical. When power is scarce and demand is volatile, grids start to be redesigned around the largest loads, and communities can find themselves competing with hyperscale buyers for electricity, land, and long-term infrastructure investment. That pressure is already prompting new approaches: more direct procurement of generation, more private “behind-the-meter” power deals, and renewed debates about nuclear and other firm baseload options. If we get this wrong, we normalise a future where critical infrastructure decisions tilt towards a few buyers, rather than broader public interest, resilience, and fairness.