Peak Demand Charges Are About to Get More Expensive Across the PJM Grid

Facility managers and energy managers have spent years learning when their buildings are most expensive to run, and shaping operations around those hours. That playbook is now getting harder to rely on. A new independent analysis prepared for the Pennsylvania Public Utility Commission (PUC) shows that the PJM grid, which serves Pennsylvania and twelve other states plus Washington, D.C., is tightening faster than expected, largely because of surging data center demand. The report was produced by Synapse Energy Economics, with support from Mondre Energy and Aspen Technologies, and it lays out in specific terms how supply and demand pressure is building through 2030 and beyond. For any organization managing electricity costs across a factory, warehouse, or commercial building portfolio, the findings point to a clear shift: the hours when the grid is stressed, and when demand charges bite hardest, are no longer fixed.

Why Peak Risk Hours Are Moving

PJM measures grid reliability using a metric called loss of load expectation, or LOLE, which estimates how often the system can be expected to experience a power shortfall over a given period. The industry standard target across North America, including PJM, is an LOLE of 0.1, corresponding to roughly one day of shortfall risk every ten years. Historically, PJM’s stress hours have clustered predictably in winter mornings, with January at 8 a.m. identified as the peak risk hour in 2027 across all scenarios studied.

That pattern starts to break down by 2030. In the report’s Reference scenario, winter mornings remain the top risk period, but a growing share of stress hours also appears in summer and winter evenings. In the more severe High Load and Low Supply scenario, the shift is more pronounced: summer evenings emerge as a major new risk window, with close to 15 percent of estimated unserved energy occurring in July at 8 p.m. The reason is straightforward. Data centers run at a largely flat load around the clock, so as they account for a growing share of total demand, they add stress to hours that used to be considered safe. For any facility that shifted load into evening hours to avoid winter morning peaks, that strategy may need a second look.

The Numbers Behind the Price Pressure

The reliability numbers translate directly into cost. Under the Reference scenario, modeled 2030 LOLE for PJM as a whole reaches 0.59, nearly six times worse than the 0.1 planning target. Under the High Load and Low Supply scenario, which assumes higher data center growth and tighter supply, 2030 LOLE jumps to 13.20, more than 130 times worse than target, equivalent to an expected 13 days of system stress somewhere in PJM that year. By contrast, a scenario with no new data center load added after 2027 keeps LOLE at 0.05, better than the target.

Capacity and energy prices move in tandem with these reliability figures. PJM’s current price collar, approved by federal regulators, caps capacity auction prices at 325 dollars per megawatt-day and floors them at 175 dollars per megawatt-day through the 2029/2030 delivery year. The report finds capacity prices sitting at or near that cap in the Reference scenario through 2030, and at or near the cap in every year modeled under the High Load and Low Supply scenario. Energy prices tell a similar story: in the more stressed scenario, average annual energy prices run in the range of 100 to 225 dollars per megawatt-hour, compared to lower and more gradually rising prices in the scenario without new data center load. Because energy costs make up a larger share of total system costs than capacity costs, these energy price differences carry a larger downstream effect on customer bills than capacity prices alone.

What Facility and Energy Managers Should Do Now

These findings suggest a few practical steps worth taking before rate structures and grid conditions shift further:

• Revisit load curve assumptions against the new risk windows identified in the report, rather than relying solely on historical winter morning peak patterns.
• Model demand response or curtailment options that cover summer evening hours, since that period is emerging as a new stress window in higher load scenarios.
• Evaluate on-site battery storage or generation as a hedge against price volatility. The report notes that solar and battery storage are the primary near-term resource additions expected in Pennsylvania, adding 3 to 7 gigawatts combined through 2030.
• Track large-load interconnection and tariff rules if planning any facility expansion. The PUC approved a model tariff in May 2026 that assigns infrastructure cost responsibility to customers with loads of 50 megawatts or greater, which could affect expansion economics.
• Use granular, near real-time energy monitoring to catch shifts in peak demand patterns as they develop, rather than discovering them after a bill arrives.

None of these steps require waiting for a crisis. The report’s own modeling shows that the sharpest changes in reliability metrics occur between 2029 and 2030, driven almost entirely by data center load additions, giving facility and energy managers a window to plan ahead of that shift rather than react to it.

Conclusion

The PUC-commissioned analysis makes a case that is hard to ignore: PJM’s grid stress is becoming less seasonal and less predictable, driven by the rapid growth of data center demand. For facility directors, factory managers, and CFOs overseeing electricity budgets, this means demand charges and time-of-use rate exposure are likely to extend into hours that were previously considered low-risk, particularly summer evenings. Reliability and cost pressure are converging into a year-round planning issue rather than a two-season concern. Organizations that build visibility into these shifting patterns now, through better monitoring and more flexible load management, will be in a stronger position than those that wait for the next bill to explain what changed.

Reference

• Pennsylvania Public Utility Commission: Assessing Electricity Resource Adequacy in Pennsylvania and PJM