Who Pays for AI? Causal Evidence That Data Center Growth Widens the Residential-Commercial Electricity Price Gap

The rapid buildout of AI data centers has concentrated unprecedented electricity demand in a handful of U.S. states, yet no prior study has causally identified the effect of this concentration on retail electricity pricing across customer classes. We provide the first such estimate. Using a difference-indifferences (DiD) design with state and month fixed effects on 1,188 state-month observations (2015-2025), we find that the residential-commercial electricity price gap widened by +1.27 ¢/kWh (p = 0.025) in the five highest data-center-concentration states (Virginia, Texas, California, Nevada, Illinois) relative to matched controls after January 2022.

At average U.S. household consumption, this equals $133 per household per year-and growing. An event study validates parallel pre-trends and localizes the causal break to 2023-not 2022-ruling out the Russia-Ukraine energy shock and pointing instead to the 12-18 month regulatory lag between data center interconnection agreements and approved residential rate adjustments. Prophet time-series forecasts project this gap to continue widening through 2028, with Texas residential customers paying 114.6% more than commercial customers and Virginia reaching 61.1%, far exceeding the national average premium of approximately 29%.

These findings reveal a structural cross-subsidy in which hyperscale operators negotiate discounted commercial tariffs and the revenue shortfall is recovered from residential ratepayers-a mechanism that is policytractable through rate case reform independent of the pace of AI deployment.