Date of Award

Spring 5-1-2026

Document Type

Thesis

Degree Name

Bachelor of Science (BS)

Department

Economics

First Advisor

Monica Das

Abstract

This thesis examines whether the expansion of AI-related computational infrastructure has contributed to measurable changes in state-level commercial electricity consumption in the United States. While most economic research evaluates artificial intelligence through productivity, firm behavior, or labor market outcomes, the physical energy requirements of large-scale computation have received comparatively little attention. Drawing on production theory, this paper argues that AI adoption increases derived demand for electricity by expanding the utilization of electricity-intensive computational capital—primarily data centers—and that this effect should be geographically concentrated in states with high data-center exposure. Using state-level commercial electricity sales data from the U.S. Energy Information Administration over the period 2010–2024, I estimate a difference-in-differences model that compares cumulative electricity growth in high data-center states against a control group after 2020, the period in which AI adoption and cloud-based compute intensity accelerated sharply. The preferred specification controls for a common time trend and a treatment-specific trend to account for potential violations of the parallel trends assumption. The main finding is that treated states experienced approximately 10.5 percentage points greater cumulative growth in commercial electricity consumption relative to control states after 2020, a result that is statistically significant at the 5 percent level. This result is robust to alternative specifications and is consistent with the mechanism proposed: AI-driven infrastructure scaling generates a gradual, geographically concentrated increase in commercial electricity demand that is not visible in national aggregates but emerges clearly at the state level.

Download

Included in

Economics Commons

Share

COinS