Nic Carter Says Data Shows No Clear Link Between AI Data Centers and Higher Residential Power Rates

Castle Ventures partner Nic Carter stated that recent headlines linking AI data centers to higher household electricity bills lack supporting data. Carter wrote in a May 17 blog post that some regions with high data-center density have recorded modest rate increases, but overall trends show real electricity prices have declined over the past decade.

U.S. Energy Information Agency, the Bureau of Labor Statistics, and the Electric Power Research Institute.

States with the largest concentrations of data centers—Virginia, Texas, and Nevada—currently post the lowest residential power rates, Carter said. He explained that hyperscale operators locate facilities in areas that already offer lower electricity costs and established transmission capacity.

The states with the highest residential rates—California, Massachusetts, and Connecticut—have comparatively few data centers, he added. Carter listed blocked pipelines, restrictions on nuclear power, union labor costs, renewable-energy mandates, and local opposition to new infrastructure as contributing factors.

Carter noted that the electricity required per AI query has fallen by a factor of 200 since 2022, according to a 2025 Stanford University study. He also said hyperscalers are increasingly building on-site generation, which reduces pressure on the shared grid.

When data-center demand occurs outside peak hours, Carter said, the added load can spread fixed generation costs across more kilowatt-hours and thereby lower per-unit rates for all customers. He added that operators possess capital to expand capacity if regulators require them to do so.