AI's real bottleneck is power, not chips: missing warm-shell data centers strand GPU capital
rohanpaul_ai · x · 2026-10-26
Nadella admitted last year Microsoft lacks electricity for all its AI GPUs — the constraint isn't chip supply but 'warm shells' (finished, utility-ready buildings near large power feeds).
- OpenAI is lobbying the U.S. government to add 100GW/year of generation as a strategic AI asset
- The power gap creates stranded capital: GPUs depreciate while waiting for buildings, substations, and transmission
- Even centralized training can't escape it — inference demand compounds the strain
More from Infra
- Matthew Berman: Image, video and voice gen are now good enough to run locally — MatthewBerman · 2026-10-26
- Reddit asks: Why can't AI coding agents run locally like Stable Diffusion? — Kaporalhart · 2026-10-26
- Vjeux joins Muse Open Kernel Challenge; colinlikescode leads 4 of 5 problems — Vjeux · 2026-10-26
- Gandalf paper: KV cache offloading wears out SSDs in months; attention-aware admission extends life 7.3x — ThePrimeClock · 2026-10-26
- Gary Marcus publicly wagers on whether CoreWeave stock drops below $40 this week — GaryMarcus · 2026-10-26
- Resin: a runtime that lets agents write forward passes to build standalone LLM inference engines — andersonbcdefg · 2026-10-26