US Watchdog Warns That AI Will Drain American Power Grid

NERC (North American Electric Reliability Corporation), an energy watchdog, has warned that the rising demand for artificial intelligence could outpace power grids.

This article was written by Sam Shedden and originally published by ReadWrite.

AI is currently experiencing major issues with power supply. So much so, that companies like Microsoft have begun to invest in reopening nuclear power plants. NERC now warns that there’s a possibility that this could lead to blackouts in both the US and Canada.

On top of this, cryptocurrency’s revived boom since the election of Donald Trump has continued to put stress on power grids across the country. Bitcoin currently sits at $105,000, and its breaking of the $100K mark has sparked a renewed interest in mining.

This involves running heavy-load PCs – in some cases en masse – running constantly to crack codes to mine a Bitcoin. In 2024, it is estimated that 0.6% to 2.3% is used by cryptocurrency mining operations.

On the AI front, it is estimated by a Goldman Sachs report that the US will use an estimated 8% of its power to fuel AI by 2030. That’s a 160% increase.

It’s not just crypto and AI draining the US grid

NERC’s 2024 report also highlighted that more electric vehicles are adding to the stresses on power grids. Texas, which runs its own grid, is facing the possibility of shortages due to damage to the grid from storms and the excess needed from tech infrastructure.

In October, the Texas Public Utility Commission reported that its grid won’t be able to fully support any more data centers on the Texas grid.

NERC’s report found that the US could suck up as much as 132 gigawatts of power. This is up by 80 gigawatts since last year.

A major part of NERC’s concerns is the lack of progress on installing alternative energy sources. Options like solar panels are constantly hit with delays, which could knee-cap energy reserves as big tech companies shift more and more into energy-guzzling AI projects.

Featured image: Pexels (1, 2)