
0:00
41:14
Is AI a giant bubble? Jonathan Koomey has spent more than a quarter century measuring how much electricity computers and data centers really use. We talk about the $700-800 billion going into AI infrastructure this year and whether it will ever be paid back, what the fiber boom of the 1990s tells us, why GPUs lose their value so fast, and why 2027 could be an interesting year for Nvidia, the Nasdaq and the midterms.
We also go through the electricity and water claims that did not hold up, the xAI data center in Memphis and its jet engines, who pays when the promised demand never comes, what happens to workers, and why lawyers keep citing cases that do not exist.
Reports mentioned, from Jonathan Koomey:
Understanding AI energy use (Koomey and Masanet, Veolia and Microsoft, February 2026): https://www.institut.veolia.org/en/publications/veolia-institute-review-facts-reports/ai-energy-water-and-waste-management
Separating fact from fiction in data center electricity forecasts: A guide for regulators (Koomey, Schmidt, Sreedharan, Kumar and McNair, Koomey Analytics and Gridlab, March 2026): https://gridlab.org/portfolio-item/data-center-load-forecast-report/
Electricity Demand Growth and Data Centers: A Guide for the Perplexed (Koomey, Schmidt and Das, Bipartisan Policy Center, February 2025): https://bipartisanpolicy.org/report/electricity-demand-growth-and-data-centers/
Jonathan Koomey: https://www.koomey.com
Watch the full conversation on YouTube: https://youtu.be/rhB6mVbZ_KY
More at loadbearingpodcast.com
Music: "Hall of the Mountain King" by Kevin MacLeod, licensed under Creative Commons Attribution 4.0 (https://creativecommons.org/licenses/by/4.0/)
Mais episódios de "Load Bearing Podcast"



Não percas um episódio de “Load Bearing Podcast” e subscrevê-lo na aplicação GetPodcast.








