Discussion about this post

User's avatar
Stephanie Barret's avatar

Worth putting the fleet-wide number next to the 403-building one: LBNL's 2024 report had all US data centers at 176 TWh in 2023, about 4.4% of US electricity, with 325-580 TWh projected by 2028. The Harvard cohort is the hyperscale top of that stack, so 99 TWh for 403 buildings sits inside a larger total rather than being the ceiling.

The 54% fossil share is the line I would track year over year. It is the one that depends on which grid each building landed on, not on how efficient the servers are.

Stephanie Barret's avatar

Worth flagging the limitation the authors are explicit about: the survey cannot separate AI load from ordinary cloud load. That matters well outside the study. The ratepayer fights now turn on attributing cost to a data-center class - BAAJA's 20 MW threshold, state large-load tariffs, utility cost-of-service filings - and the base for that attribution is estimated and cross-referenced, not metered per customer. The 2% is a useful baseline; the attribution gap is the harder problem.

5 more comments...

Ready for more?