PJM Interconnection faces a 6.8 GW power shortfall, bringing intense scrutiny to data center growth and grid planning.
PJM’s initial large-load stakeholder process yielded 12 proposals but reached no consensus. An accelerated procurement is now underway to address a 6.8 GW capacity shortfall.
The region’s most recent capacity auction left the RTO 6.8 GW short of its Reliability Requirement for the 2028/29 delivery year. This gap intensifies pressure on the grid operator to establish rules for rapidly expanding large-load demand—particularly from data centers—and to secure additional power supply. In response, PJM has proposed a Reliability Backstop Procurement (RBP), an accelerated mechanism designed to close the deficit. Following board approval, PJM filed the proposal with the Federal Energy Regulatory Commission (FERC) on July 31.
For data center developers, PJM’s large-load policies directly dictate project viability, the volume of generation they must build or contract, and the long-term certainty of electricity costs. The strategy also reflects lessons learned from PJM’s first major effort to manage large-load growth through its stakeholder framework. Launched in 2025, that initiative generated 12 proposals without achieving consensus. Lacking sufficient stakeholder backing for any single option, PJM’s board leveraged the record to chart its own policy direction and mandate next steps. The central challenge now is whether PJM can accelerate decision-making while maintaining clear boundaries between high-level policy direction, stakeholder-led implementation, and the evidentiary basis for major planning assumptions.
A recently released transcript from a July 23 FERC technical conference on PJM governance underscores the sharp divisions among participants regarding the stakeholder process. During PJM’s inaugural Critical Issue Fast Path (CIFP) process for large loads, all 12 submitted proposals were rejected by stakeholders, with none advancing to the PJM Board of Managers. Pennsylvania Deputy Secretary of Policy Jacob Finkel recounted the aftermath at PJM’s Valley Forge headquarters. “The members went around patting themselves on the back for what a thoughtful, deliberative, creative process they had just gone through,” Finkel said. “This is a failure. You have not advanced the ball.”
Neil Osnato, founder of Persistence Analytics Group, argued the bottleneck was never a shortage of ideas. Instead, the stakeholder process was tasked with resolving a complex policy question pitting participants with divergent economic interests, downside exposures, and cost-allocation preferences against one another. “At some point, more alternatives do not create consensus,” Osnato said. “They create more ways to avoid making the underlying decision.”
Joseph Bowring, president of Monitoring Analytics and PJM’s independent market monitor, noted that while the stakeholder process retains value, it requires greater efficiency. “Stakeholders were generally pursuing their narrow self-interest and ignoring the broader public interest in efficient, competitive markets that assign costs and risks to those that cause them—in this case, the data center load,” Bowring told Data Center Knowledge. Collectively, the proposals tackled fragmented aspects of the large-load challenge, spanning load forecasting, new generation development, interconnection queues, reliability standards, and the treatment of customers falling outside PJM’s traditional resource adequacy framework.
Ultimately, PJM’s board synthesized the CIFP record into a comprehensive large-load framework, drawing concepts from multiple stakeholder submissions. The resulting January policy addressed load forecasting, “bring-your-own” generation arrangements, expedited interconnection pathways, and a reliability backstop mechanism. Osnato emphasized that the outcome does not negate the process’s utility. It surfaced critical data, clarified competing viewpoints, and provided the board with a structured menu of options. However, it stopped short of delivering a definitive policy ruling. “The stakeaker process functioned as an input mechanism, but failed as the final decision mechanism on this particular issue,” Osnato said.
Abe Silverman, assistant research scholar at Johns Hopkins University’s Ralph O’Connor Sustainable Energy Institute, echoed this distinction during the FERC conference. He argued the core issue was not the proliferation of committees, but rather assigning them policymaking authority. “It’s when we ask committees to decide policy,” Silverman said. He advocated for PJM, its board, and state regulators to establish overarching policy direction upfront, allowing stakeholders to focus exclusively on technical execution and implementation details. “We would have saved so many hours of meetings,” he added. Pamela Quinlan, executive director of the New Jersey Energy Resilience Bank, reinforced this view while warning against discarding PJM’s technical stakeholder framework entirely. She stressed that strategic direction must originate from PJM leadership and the board, while stakeholders supply the specialized technical expertise required for implementation.
PJM’s July Base Residual Auction secured 138,318 MW via the Reliability Pricing Model. When adjusted for fixed-resource-requirement capacity, total RTO-procured capacity stood at 149,181.6 MW UCAP—falling 6,831.3 MW UCAP short of the Reliability Requirement for the 2028/29 delivery year. The proposed RBP aims to bridge this deficit. Pending FERC approval, the procurement is scheduled to launch on September 30. It would lock in resources for terms of up to 15 years, capping the weighted-average willingness to pay at $555 per MW-day. Any qualifying new generation secured through bilateral agreements or otherwise integrated into the system would directly reduce the volume PJM must procure through the backstop.
This reliability deficit carries immediate implications for the entities fueling PJM’s demand surge. Brian George, Google’s senior lead for U.S. energy markets, informed FERC that the tech giant has committed billions of dollars to PJM infrastructure and operates numerous regional data centers. Although not a formal PJM member, Google actively engaged in the large-load process alongside members and nonmembers to propose solutions for resource adequacy. Nevertheless, George emphasized that the company requires greater regulatory certainty before deploying additional capital into power infrastructure. Google anticipates funding new generation assets as part of a broader federal ratepayer protection initiative. “We can’t do that without the certainty that the market rules are going to exist in such a way that we know that investment is going to be worthwhile,” George said.
The RBP is designed to facilitate rapid generation deployment while PJM refines its overarching large-load framework. PJM outlined a centralized procurement targeting the 6.8 GW shortfall identified in the 2028/29 Base Residual Auction. Scheduled for September and October, the process would deliver results by December and guarantee new generation resources a fixed price over 15-year contracts. Additionally, PJM proposed lowering the procurement target for large loads already secured through bilateral supply, approved integrated resource plans, or demand-side participation. Critics warn the mechanism may distort cost allocation. Joseph Bowring cautioned that the RBP would effectively socialize the expenses of serving data center demand. “The RBP process would require other customers to pay for data center load, despite PJM’s assertions to the contrary,” Bowring said. “The RBP approach is going in the wrong direction.”
The 6,831 MW auction deficit now anchors a procurement that could trigger massive new generation investments and long-term infrastructure commitments. Given the scale of these decisions, Osnato stressed that rigorous evidentiary standards are paramount. Before the shortfall metric dictates procurement volumes, cost allocations, or infrastructure builds, PJM must transparently document the underlying assumptions, validate the supporting evidence, and verify which large-load additions are both executable and durable. Furthermore, the operator must define which variables must hold constant and what shifts would necessitate recalculating the projection. “Speed matters,” Osnato said. “But speed and verification should not be treated as opposites.”
This principle extends to PJM’s wider large-load planning architecture. “Before a large load representation becomes an infrastructure planning assumption,” Osnato said, “someone should be able to answer: Who proved it?” As PJM navigates a projected influx of approximately 70 GW of new large-load demand by 2038 while simultaneously closing the immediate reliability gap, the tension between urgency and rigor grows increasingly consequential. The grid operator is now pivoting toward an accelerated procurement pathway to stabilize near-term supply, even as fundamental disagreements persist over long-term policy design. The disagreement over these priorities continues to shape the debate.