Data Centers Initially Lowered Electricity Costs, Now AI Buildout Threatens Trend

Data centers, often associated with rising electricity prices, have historically contributed to lower retail electricity costs, according to a working paper from the Electric Power Research Institute (EPRI). Between 2015 and 2024, data center operations defied public anxiety and actually caused electricity prices to decrease. Researchers utilized data from the Federal Energy Regulatory Commission (FERC) and retail revenue from the U.S. Energy Information Administration to establish a causal link. Their findings revealed that for every doubling of data center capacity, average retail electricity prices saw a reduction of 3.5%. On a statewide level, this decrease was approximately 6%.

This counterintuitive trend is largely attributable to economies of scale within electricity markets. EPRI researcher Asa Watten, a coauthor of the study, explained to Fortune that electricity pricing differs from commodity markets like soybeans or gasoline. Instead of being solely determined by production costs, electricity prices are based on cost recovery, meaning they are influenced by consumption levels. As fixed costs are distributed across a larger number of consumers and kilowatt hours (kWh), increased consumption leads to a greater division of these fixed costs, thereby lowering prices per unit. Furthermore, the increased load from data center usage can incentivize the activation of more generators, many of which are becoming increasingly energy-efficient, further contributing to cost reductions.

However, this pattern of decreasing electricity costs is not guaranteed to persist. The substantial buildout of AI infrastructure, projected to be a $7 trillion endeavor, introduces a significant variable that could disrupt the established trend. Goldman Sachs had previously projected that the AI infrastructure buildout would increase electricity costs by 6% between 2026 and 2027, with an additional 3% increase by 2028. This projection contrasts with the historical data analyzed by EPRI, which demonstrated a downward price pressure from data centers up to 2024. The critical factor is whether the demand for electricity driven by AI applications will outpace the efficiencies and economies of scale previously leveraged by data centers.

The YouGov poll administered last year to 1,000 Americans highlighted a prevalent public concern, with over two-thirds expecting electricity prices to rise if a data center were built in their vicinity. This widespread perception underscores the potential for public backlash and regulatory scrutiny if electricity costs do indeed escalate due to the AI boom. The EPRI research complicates this narrative by showing that, until recently, the opposite was true. The future trajectory of electricity prices will depend on the interplay between the massive investment in AI-related data centers, the ongoing development of energy-efficient technologies, and the overall demand for power across all sectors.