Impacts of Co-Locating Data Centers with Designated Energy Resources in the PJM Interconnection

Abstract

Rapid data center development has been causing rising capacity prices, and this effect is especially pronounced in the PJM Interconnection. Solutions like self-procuring behind-the-meter (BTM) capacity, demand response, and clean-only mandates have become part of real-world conversations to mitigate economic and emissions impacts, yet publicly available modeling methods for planning amid continued data center growth remain scarce. We develop a methodology using the GenX open-source capacity expansion model to analyze PJM-wide impacts when 1 GW of incremental flat data center load connects to the grid, implementing a custom demand response module to test that policy scenario alongside partial grid connection and clean-only policies. We find that fully firm interconnection of incremental data center load adds an estimated $920.6M in total system costs and increases capacity prices by about 3%. Reducing the data center's firm grid connection to 500 MW (50%) and 250 MW (25%) through BTM co-located resources with demand response participation reduces incremental system costs by 81% and 97%, respectively. We also find that BTM co-location is less resource efficient relative to direct grid connection with full cost internalization, suggesting that restricting data centers to co-located resources — rather than grid-optimal siting — is not an ideal solution. Environmentally, co-located resources and demand response participation shift the optimal capacity mix away from natural gas and coal in favor of renewables and battery storage. These findings corroborate expectations that partially firm grid connection with BTM resources and demand response participation are powerful solutions to address capacity constraints.