GeekPark models how a freeze on unstarted US data centers would slow AI compute growth
Overview
GeekPark models an extreme scenario in which every planned US data center project that has not yet started is frozen, while projects already under construction finish on schedule.
Under that assumption, the article estimates US data center capacity would plateau at roughly 120 to 150GW around 2028, and annual effective compute growth would fall from about 12 to 15 times to about 6 times.
GeekPark concludes that AI development would slow sharply but not stop, with labs shifting effort toward inference, post-training and efficiency work. The figures come from the article's own model and assumptions, not from reported data on actual construction freezes.
Written by AI from the articles below · updated Nov 11, 1:13 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- GeekParkNewsHow a US data center freeze could slow AI compute growth
AIGeekPark models an extreme scenario in which all planned, unstarted US data center projects are frozen, while already-under-construction projects finish on schedule. Under that assumption, US data center capacity would plateau at roughly 120 to 150GW around 2028, and annual effective compute growth would fall from about 12 to 15 times to about 6 times. The article argues AI would slow sharply but not stop, with labs shifting toward inference, post-training, and efficiency work.
Heat Trend
- Current comparable heat
- 9
- Peak comparable heat
- 10Nov 11, 2:00 AM ET
- Change in past 24 hours
- –
The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.