AI revolution: So big, Texas grid needs the equivalent of 30 Nuclear Reactors By 2030 to meet Power Demand
Data centres and Electricity demand,
By Jo Nova
The insatiable hunger for electricity
The world is about to flip from an energy diet to an electrical boom. Look at Texas.
Here in Australia our top Blob Scientists tell us it will take 15 years to build one nuclear plant. But in Texas, which has two nuclear plants already, the AI revolution is beating down the door, and it’s saying “Feed me 30 plants for breakfast ” — or at least by 2030. It’s like a different planet.
There are already 340 datacentres in Texas which use 8GW of power, but new projects are so large, they are starting to ask for a whole gigawatt up front. And the sum total of requests for new electrical supply add up to 99 gigawatts — most of which have materialized in the last year. The new level of demand is so big, the grid managers are starting to worry that single new industrial loads are large enough to threaten the grid.
We’re talking of a seismic shift:
The ERCOT grid peak load last summer was 86 gigawatts. The new peak demand by 2030 is expected to be 75% bigger. It may not all be nuclear, ERCOT did ask for “the equivalent” of 30 new nuclear plants. But it will be big and fast.
Texas has 31 million people. Australia has 26 million (plus a million foreign uni students). We’re performing a national pantomime over whether it’s realistic to build a few nuclear plants and a couple of SMR’s, as if we’re plotting to build the first colony on Venus. Meanwhile Texas is leading the revolution.
https://joannenova.com.au/2025/03/ai-revolution-so-big-texas-grid-needs-the-equivalent-of-30-nuclear-reactors-by-2030-to-meet-power-demand/?utm_source=rss&utm_medium=rss&utm_campaign=ai-revolution-so-big-texas-grid-needs-the-equivalent-of-30-nuclear-reactors-by-2030-to-meet-power-demand
