Demand Shock and Institutional Constraint: AI Datacenters and Grid Governance in the PJM Interconnection

Abstract

The rise of AI and the demand-heavy datacenters needed to power it have reshaped American electricity markets over the past five years. The case of PJM Interconnection and its capacity market crisis demonstrates how. The entry of datacenter demand into electricity markets exposed the mismatch between 21st-century demand shocks and a 20th-century governance architecture designed for slow, incremental, and predictable load growth. This thesis draws on scholarship on electricity restructuring, energy federalism, and regulatory capture, alongside analysis of market reports, regulatory actions, policy documents, and industry evidence. I argue that the crisis emerged through three linked dynamics. (1) The original design of PJM's market is poorly suited for datacenter load. (2) The fragmented authority structures within PJM's governance limit its ability to coordinate reforms. (3) The costs of attempted adaptations by market actors are distributed unevenly across market actors. More broadly, this thesis shows that when private capital moves faster than outdated governance structures, institutions rely on temporary political fixes and workarounds rather than on true structural reform.