The More Wind and Solar We Add, the Less They Deliver

Booming electricity demand for artificial intelligence (AI) is reshaping the electric grid. Bloomberg New Energy Finance estimates 63% of global power plants built to satiate AI’s thirst for reliable, around-the-clock power will be coal, nuclear, and natural gas plants.The reason is simple: the reliability value of wind and solar falls dramatically as more of these facilities are placed into service.

Power demand for AI data centers is skyrocketing. In 2023, data centers only accounted for approximately 4.4% of U.S. electricity consumption, but the U.S. Department of Energy estimates that data centers could consume 6.7 to 12% of total U.S. electricity by 2028. Meeting this massive increase in consumption will require a significant number of new power plants to be built in the coming years. The question then becomes, what should we build?

Wind and solar advocates argue that solar panels, wind turbines, and grid-scale batteries are the best way to meet the rising demand for power because these facilities can be built quickly, and that building natural gas and nuclear power will take too long. 

However, there is a good reason why U.S. utilities plan to add 120 gigawatts of new fossil fuel capacity to the grid by 2030, enough to continuously power 97 million American homes for a year: wind and solar aren’t reliable enough to do the job of powering data centers.