Data Centers and the Challenge of Peak Electricity Demand
AI-driven data centers are growing rapidly, adding substantial electricity demand in many regions. Their system impacts depend not only on how much electricity they consume, but also on when they consume: a small number of peak-demand hours drive reliability risks and increase reliance on expensive, more carbon-intensive generators. This project provides empirical evidence on the extent to which data centers can shift electricity consumption away from these peak periods when economic incentives are strongest. Using ERCOT’s Four Coincident Peak (4CP) program—which bases large customers’ annual transmission charges on demand during the four highest system-load intervals each summer—the analysis estimates how data center electricity demand changes around 4CP events. These estimates are then combined with generator bid and dispatch data to quantify effects on peak-time wholesale electricity prices and identify which generators are operating on the margin. By measuring the real-world demand flexibility of large electricity consumers, the project informs broader discussions of how to integrate data centers while advancing electricity system reliability and decarbonization goals.
Martin Navarrete
Doctoral Student, Applied EconomicsMartin Navarrete is a doctoral student of applied economics in the Business Economics and Public Policy Department.
Susanna Berkouwer
Assistant Professor of Business Economics & Public PolicySusanna Berkouwer is an assistant professor of Business Economics & Public Policy at the Wharton School.
