Carbon Aware / Grid Aware computing as a way to avoid needing future gas buildout - any other papers exploring this?

[mrchrisadams]

Hi folks,

I'm sharing this piece here because last week an interesting paper was shared, that essentially made the argument that more thoughtful us of the grid in the US to use under utilised capacity more effectively (sound familiar?) would make it possible to accomodate more computation on the grid (particularly computation from AI datacenters) without needing to deploy new gas.

This is because for the most part, the grid is running with a lot of head room, and if you were able to shape demand, you would be able to remove the need to switch on the kinds of peaker gas generation to meet serve grid-wise peaks in demand, but also you would be able to reduce or remove the need to provide longer running the kind of closed cycle gas turbine generation that utility companies are saying they need to meet new demand.

Have you come any other good resources exploring this idea?

Here's one good piece that summarises the report:
https://www.motherjones.com/politics/2025/02/new-duke-study-power-curtailment-ai-data-centers-nuclear-gas-plants/

And this here is the report itself
https://nicholasinstitute.duke.edu/publications/rethinking-load-growth

This is a helpful thread on bluesky from the report author, Tyler Norris:
https://bsky.app/profile/tnorris.bsky.social/post/3lhvx73yvvt2s

This is my query on Bluesky, pretty much asking the same question:
https://bsky.app/profile/mrchrisadams.bsky.social/post/3liecbkvdwc2i