AI's next bottleneck is cheap electricity

"We've spent several years now obsessing over models and assistants, but here's a new interesting truth: the next competitive edge in AI won't be another benchmark, but electrons. And not just any electrons, but cheap ones. As the " AI wars" heat up, the winners won't simply be those with the best UX or the most compute. They'll be the firms that can secure abundant low-cost power at scale, hour after hour, year after year."

"That's where AI is colliding with the physical world, and where the story stops being about software and starts being about grids, turbines, and price curves. Most recent analyses show that AI-driven data centers are now a visible driver of U.S. electricity demand and are starting to send retail prices higher, a clear signal that the constraint is shifting from graphics processing units (GPUs) to kilowatt-hours."

"In data centers, much of the water involved in most cooling systems is withdrawn, then used to absorb heat, and later returned. Warmer, yes, but reentering the water cycle after discharge is brought back within permitted temperature ranges. Only some designs (notably evaporative cooling) consume water through vapor losses; others trade water for electricity by leaning on air-cooled chillers or direct-to-chip liquid loops that dramatically cut onsite withdrawals."

The next competitive advantage in AI will be abundant, low-cost electricity rather than benchmarks, UX, or raw compute. AI-driven data centers are becoming a visible driver of U.S. electricity demand and are beginning to raise retail prices, shifting the binding constraint from GPUs to kilowatt-hours. Cooling choices affect water impacts: much water in cooling systems is withdrawn and returned at permissible temperatures, while only evaporative designs consume water through vapor loss. Air-cooled chillers and direct-to-chip liquid loops reduce onsite water withdrawals by trading electricity for water. Water stress is location-specific and depends on siting and cooling technology.

Read at Fast Company