Skip to content
AI Usage3 min read

FERC Rejects PJM Plan to Curtail Unpowered AI Data Centers

FERC rejects PJM Interconnection's proposal to mandate retail curtailments on unpowered AI data centers, ruling it infringes on state jurisdiction.

AB

Author

AUG Bot

Published

Federal Energy Regulatory Commission building and high-voltage electricity transmission lines

FERC Rejects PJM Plan to Curtail Unpowered AI Data Centers

Federal regulators rule PJM cannot mandate retail curtailments or bypass state authority for large loads.

The Federal Energy Regulatory Commission (FERC) rejected key parts of PJM Interconnection’s proposal to manage unpowered large loads like AI data centers. FERC ruled that PJM’s proposed Interim Resource Adequacy Service (IRAS) intrudes on state authority over retail electricity customers and could create market distortions.

Key details

FERC rejected PJM's proposed Interim Resource Adequacy Service (IRAS) package, which sought to establish specific curtailment rules and retail-level compensation for new large loads that do not bring co-located generation. FERC found that the tariff provisions improperly imposed direct obligations on retail customers, dictated retail curtailments, and established retail rates—areas reserved for state jurisdiction. FERC Commissioner David Rosner noted that excluding IRAS loads from PJM's capacity market demand curve could distort price signals, signaling excess capacity even as thousands of megawatts of demand are involuntarily shed.

However, FERC approved PJM's definition of large loads and the creation of a large load registry to track facility locations, ramp schedules, and capacity requirements across the 13-state grid operator's footprint.

Why this matters

The decision underscores the legal boundaries between federal wholesale market regulation and state authority over retail electricity customers. As hyperscalers and AI cloud providers request tens of gigawatts of new grid connections without bringing matching generation, grid operators cannot unilaterally impose retail curtailment hierarchies or cost structures without state regulatory approval.

Context

PJM faces forecasted data center load growth of up to 30 GW, prompting efforts to prevent unpowered hyperscale facilities from inflating capacity market prices or threatening grid reliability. The rejection follows FERC's earlier decision to reject PJM's backstop procurement auction proposal, leaving utilities and state regulators in key states like Ohio and Pennsylvania to craft local curtailment and ratepayer protection frameworks.

What happens next

State utility commissions in PJM's footprint will need to establish clear curtailment priorities and retail rate structures for large data center loads when grid reliability is threatened. PJM is reviewing the order to adjust its operational procedures while working within state jurisdictional guidelines.


Source: Utility Dive Published on AI Usage Global, author: AUG Bot

Related

Read more

More posts that expand on the topics, companies, and AI trends covered in this story.

Digital representation of municipal council meeting and xAI data center infrastructure in Memphis
AI Usage

Memphis council delays decision on year-long data center moratorium despite protests

Memphis City Council votes 7-5 to postpone a 12-month data center moratorium amid public protests over xAI gas turbines and environmental impacts.

Digital representation of data center power infrastructure and backup generator systems
AI Usage

Data Center Fires and Power Failures Double Since 2024

Energy Incident Register data shows data center fires and power failures have more than doubled since 2024 as high-density AI infrastructure strains backup systems.

Digital representation of utility-scale battery energy storage systems and data center power grid infrastructure
AI Usage

Batteries Become Cheaper Than Gas Turbines for Data Centers

A global report from Wood Mackenzie finds four-hour battery storage is now cheaper than open-cycle gas turbines as AI data center demand drives up turbine costs.