What 100 GW of compute really means: 876 TWh a year and a country-scale power system
shyamalanadkat · x · 2026-09-09
- The author sketches what 100 GW of compute in one place implies: at full utilization, 876 TWh a year, requiring a country-scale power system, factories for chips, transformers and cooling gear, and a place to dump 100 GW of heat every second.
- The more interesting question, he argues, is what we'd do with it: throwing that much compute at scientific problems — exploring millions of designs, picking the most promising experiments, running them, and iterating — could target disease, cheaper energy, and other long-standing problems.
- Computation alone doesn't produce cures; the opportunity is accelerating the whole measure-experiment-build cycle, deliberately choosing which problems deserve that industrial-scale attention.
More from AGI Musings
- Dev welcomes labs training on his code: better models benefit me too — intellectronica · 2026-09-09
- Tom Dietterich: AI safety hinges on novel, ill-defined tasks AI can't optimize — tdietterich · 2026-09-09
- Tom Dietterich clarifies: 'new knowledge' in AI debate means environment-interaction knowledge — tdietterich · 2026-09-09
- From GPT-4 failing basic addition to 10,000 agents solving a Millennium Prize problem in 3.5 years — MartinSignoux · 2026-09-09
- If AI proved P=NP, it would imply insights humans lack, argues jessi_cata — jessi_cata · 2026-09-09
- Roko's Bet: If AI Solves P vs NP Soon, It'll Be P=NP Via an Ugly Algorithm — jessi_cata · 2026-09-09