PJM Interconnection, the largest grid operator in the U.S., is struggling to secure enough generation for households, businesses and data centers. This overwhelming power challenge has resulted in its board recently filing several actions with the Federal Energy Regulatory Commission (FERC) to mitigate this power demand from hyperscalers and AI data centers across 13 states.
On Monday, the PJM Board of Managers directed that two regulatory proposals be filed with FERC, the U.S. government agency that regulates the interstate transmission of electricity. Arising from PJM's Critical Issue Fast Path stakeholder solutions process, the filings aim to preserve grid reliability and mitigate rising electricity capacity costs across its service region.
PJM’s 13-state region includes residential, commercial and industrial customers in Virginia, Maryland, the District of Columbia, Pennsylvania, New Jersey, Michigan, North Carolina, West Virginia and six other Midwest states. Affordability pressures are reportedly squeezing these markets as the grid operator’s board steps in to seek FERC approval for large-load and data center reforms.
Decisive action on grid reliability
The first filing, labeled the Reliability Backstop Procurement, proposes a two-path solution to fast-track third-party power suppliers for data centers. These paths include facilitated bilateral matchmaking and a centralized backstop auction, which aim to strengthen reliability and manage affordability by addressing large grid load connections faster than the pace of new generation.
“The Board believes this reliability threat requires decisive action,” the letter stated. Members highlighted key issues and actions PJM is taking within its authority to resolve this challenge of balancing its grid.
“The region has reached a critical point in its resource adequacy outlook,” the PJM Board of Managers added.
Facilitated bilateral matchmaking, where interested buyers and sellers structure and negotiate directly via a third-party intermediary, is a path PJM is seeking to use to cure projected shortfalls while addressing long-term needs, according to the filing. PJM states this process helps avoid over-procurement within the reliability backstop framework and reduces the overall volume of new power generation capacity the grid operator must buy.
Ultimately, PJM states this approach allows another pathway for data centers to safely bring new supply online.
Hyperscalers fund VPPs to counter load growth
An example of this mechanism in action is Google's grid optimization strategy within PJM's territory. Virtual Power Plant (VPP) aggregator Voltus will deploy up to 100 MW of distributed energy resources under a three-year Bring Your Own Capacity (BYOC) agreement.
Voltus will aggregate this energy from local businesses and homes into a Google-funded VPP in the PJM grid to help offset data center load growth. It also establishes additional power capacity for Google's AI and data center infrastructure over the next two years.
The other step within this first proposal is for FERC to grant permission for PJM to hold private auctions through a third party for data centers to negotiate and purchase power generation from suppliers for distribution and transmission onto PJM’s grid. According to PJM, new large load requests are forecasted to increase by an estimated 70 GW by 2038. About 15 GW of existing electricity generating resources have been retired since 2022.
PJM highlights that its most recent capacity auction for the 2028/2029 Delivery Year left a 6.8 GW shortfall below the system’s Reliability Requirement. This metric measures the capacity that can reliably serve the grid during extreme conditions under a standard of one loss-of-load event every 10 years.
“The Board is also acutely aware of the affordability pressures facing consumers,” PJM’s letter stated. “The present trajectory of rapid load growth, tightening supply and rising capacity costs is not sustainable. The region needs substantial new investment in supply, and the central affordability question is how the costs of that investment should be allocated.”
The grid operator explains that the centralized backstop auction offers data center developers not only the ability to price alternative options among third-party suppliers, but also a streamlined pathway for grid interconnection amid mounting power generation concerns. PJM is proposing to conduct a one-time Reliability Backstop Procurement from Sept. 30 through Oct. 21, capping the total cost of accepted supply offers at $555/MW-day with results released in early December.
PJM’s virtual ‘pause button’ approach
The second filing PJM submitted aims to facilitate large load growth while preserving system reliability during capacity shortages through its Connect and Manage approach. This framework aims to establish a virtual "pause button" to temporarily cut back data center power consumption if PJM's grid becomes overwhelmed. The action would ensure PJM can prioritize keeping the lights on for the 67 million people it serves in the event system demand exceeds available generation capacity.
According to the Connect and Manage filing, the Board will instruct PJM staff to exclude any incremental new large load requests relative to the forecast utilized for the 2028/2029 Base Residual Auction from the demand used in future RPM Auctions starting with 2029/2030. This should ensure existing customers do not bear higher capacity costs from new large load requests that do not directly serve them.
To manage this, PJM is developing a “Large Load Registry” to improve load forecast accuracy and transparency by publicly tracking cumulative peak loads of at least 50 MW at a single site behind one or more points of interconnection within a one-mile radius.
PJM’s Board of Managers noted that both frameworks align with the ratepayer protection principles endorsed by the White House, data centers and PJM governors while respecting state authorities.
“The federal government, PJM states and large-load customers have emphasized that new large loads should bear the costs they cause,” the Board stated. “Because PJM does not have jurisdiction to allocate retail costs directly to individual data centers, state action will be essential. PJM will support those efforts by providing information from the new Large Load Registry and other data available through its settlement and billing processes.”
Facts Only
* PJM Board of Managers filed actions with FERC regarding data center power demand.
* The filings aim to mitigate power demand from hyperscalers and AI data centers across 13 states.
* The region includes customers in Virginia, Maryland, the District of Columbia, Pennsylvania, New Jersey, Michigan, North Carolina, West Virginia, and six other Midwest states.
* One proposal is the Reliability Backstop Procurement, proposing bilateral matchmaking and a centralized backstop auction for data center power suppliers.
* Facilitated bilateral matchmaking aims to address projected shortfalls while managing reliability.
* A mechanism involves Google's VPP aggregator Voltus deploying up to 100 MW of distributed energy resources under a Bring Your Own Capacity (BYOC) agreement within PJM territory.
* The proposal seeks FERC permission for PJM to hold private auctions for data center power purchase agreements.
* Forecasts estimate new large load requests could increase by an estimated 70 GW by 2038.
* The most recent capacity auction for the 2028/2029 Delivery Year showed a 6.8 GW shortfall below the system’s Reliability Requirement.
* The centralized backstop auction proposes capping total cost of accepted supply offers at $555/MW-day with results due in early December.
* The Connect and Manage approach aims to establish a virtual "pause button" for data center power consumption during grid overloads.
* PJM is developing a “Large Load Registry” to track cumulative peak loads of at least 50 MW at a single site within one mile.
Executive Summary
The PJM Board of Managers has directed the filing of two regulatory proposals with the Federal Energy Regulatory Commission (FERC) to address power demands from data centers and hyperscalers, driven by concerns over grid capacity. The filings aim to preserve grid reliability and mitigate rising electricity capacity costs across PJM’s 13-state region.
The first proposal, Reliability Backstop Procurement, suggests a two-path solution for third-party power suppliers: facilitated bilateral matchmaking and a centralized backstop auction. This is intended to expedite the management of large grid load connections relative to new generation availability. Facilitated bilateral matchmaking seeks to resolve shortfalls while avoiding over-procurement within reliability frameworks, allowing another pathway for new supply to connect.
The second filing utilizes a Connect and Manage approach, proposing a virtual "pause button" to temporarily reduce data center power consumption if the grid is overwhelmed. This framework also seeks to adjust future capacity auctions by excluding incremental large load requests from the 2028/2029 forecast, aiming to protect existing customer costs while managing growth. PJM is working to improve load forecasting through a "Large Load Registry" to enhance transparency and accuracy for these management strategies.
Full Take
The core tension in this situation involves the conflict between rapid technological load growth from data centers and the physical constraints of existing grid capacity, framed by affordability concerns for consumers. PJM's response attempts to resolve this tension through regulatory and operational mechanisms. The dual approach—market-based solutions (backstop procurement) and system-level management tools (pause button/registry)—suggests a recognition that traditional supply-side adjustments alone are insufficient against exponential demand shifts.
The structure of the proposals reveals a strategic attempt by PJM to shift the responsibility for managing future load growth onto the entities causing the demand, yet simultaneously implement safeguards for system reliability and consumer affordability. The proposal to have new large loads bear their costs is acknowledged as essential, but the execution requires state action since PJM lacks direct retail cost allocation authority. This introduces a layer of jurisdictional complexity: balancing regional grid management with state-level retail cost responsibilities.
The VPP example illustrates an attempt to introduce distributed flexibility into the system, leveraging local resources (homes and businesses) to offset centralized load demands, which is a pattern seen in broader grid modernization efforts. The focus on forecasting accuracy via the Large Load Registry suggests that the challenge is not just about finding generation, but about achieving reliable visibility into future consumption patterns to manage risk proactively rather than reactively. The pattern emerging is an attempt by centralized infrastructure operators to gain control over demand management and cost allocation when external forces—like AI scaling—outpace established regulatory frameworks.
Bridge Questions: If the system-level pause button is activated, what specific criteria will PJM use to prioritize load shedding across disparate customer types? How will the proposed mechanism for allocating costs based on large load requests interact with existing state-level rate-setting authority? What long-term structural changes are necessary to ensure that distributed flexibility measures translate into equitable reliability and affordability outcomes for all participants?
Sentinel — Human
The text functions as a detailed report synthesizing specific regulatory proposals from PJM regarding data center load and grid reliability, exhibiting the structural complexity of human policy reporting.