Data center demand response, while likely useful for getting onto the grid more quickly, isn’t necessarily the most effective lever for lowering wholesale power prices in PJM.
That’s according to an analysis by BloombergNEF, which modeled wholesale power prices in the region across different levels of data center flexibility. BNEF expects total hourly demand in PJM to grow by about 23.7 gigawatts over the next decade, with data centers alone adding nearly 19.3 GW of average hourly load by 2035.
If 3% of PJM’s data center fleet participates in demand response (which is roughly today’s participation level), power prices in Virginia’s “Data Center Alley”would be slightly cheaper in 2035 than if the region pursues what BNEF dubs the “Chicken Little” scenario, adding neither flexibility nor expediting supply. But that level of demand response can’t match the benefits of building the new generation proposed in PJM’s expedited interconnection track, which currently comprises 19 GW of new gas, and 6 GW of energy storage.
Data Center Alley, outside of Washington D.C., is slated to see both more load and new supply. Adding the new generation there puts power prices at around $64.30 per megawatt-hour by 2035, while data center demand response, even with 100% participation, is estimated at around $84.46 per MWh.
That’s because demand response mainly cuts prices in a small number of peak scarcity hours, while new supply lowers clearing prices across the whole year, explained David Mohammadi, a senior associate covering power markets at BNEF.
The modeling also shows that between 10% and 50% data center flexibility provides the most benefits for power prices, but incremental gains taper off beyond 50%: Adding more flexible load to the same scarcity events each year deepens the price impact of those hours, without changing prices in the thousands of other hours on the calendar.
But nonetheless, Mohammadi said the impact of demand response during grid scarcity events is likely to be significant.
BNEF modeled the impact of a 2035 repeat of last year’s “polar vortex,” which brought repeated cold snaps across North America. Assuming 26.7 GW of data center load, with 50% demand response participation, prices in the region during that hypothetical storm could drop by as much as $68 per MWh, he explained.
The takeaway is that PJM doesn’t need every data center to enroll in demand response to see meaningful price relief, said Helen Kou, head of U.S. power research at BNEF. “Although it may not be the most effective measure for price relief on an average wholesale basis, there are other benefits to demand response,” she added.
Furthermore, flexibility is likely to provide some relief for high electricity rates in other ways; more demand response participation can help ease the tight reserve margins in PJM’s capacity market, which could ultimately influence what customers pay, Mohammadi explained.
Making flexibility work in PJM and beyond
PJM has been at the forefront of a national conversation about how to leverage flexible data centers. The ISO’s capacity auctions have been clearing at record-high prices since last summer, driven by a combination of high new-generation costs, plant retirements, and rapidly rising load forecasts, dominated by data centers.
In August, PJM launched its “Critical Issue Fast Path” initiative to fast-track new rules on connecting large loads to the grid. In a conceptual proposal in late 2025, the ISO specifically invited comments on how real-time demand flexibility from data centers could support reliability and resource adequacy.
That drew the ire of the region’s independent market monitor, which dismissed data center flexibility as a “regulatory fiction” because PJM can only ask groups to cut back power use during grid emergencies, but can’t enforce precise load reductions at individual data center nodes. In other words, PJM would have to rely on the honor system for data center flexibility. And the cost to ratepayers of data centers choosing not to do so when called upon, the market monitor said, would be immense.
In the months since, proposals for how to get load online in PJM have converged on these two levers — data center demand response, and new supply. Stakeholder proposals including from hyperscalers and large energy developers, propose “voluntary large load limited demand response.” FERC’s large load order to PJM late last year directed PJM to create both firm and non-firm contract options.
Meanwhile, a January joint proposal from the White House and PJM governors, which was aimed at getting new generation online to power data centers, didn’t include any specific references to data center curtailment — but the PJM board’s response to the Trump administration did. They said that the ISO would evaluate new market mechanisms to compensate for large flexible loads for verifiable demand reductions.
Data center flexibility is a top topic of conversation beyond PJM, including at a federal level. FERC, for its part, is developing a new framework for interconnecting large loads and exploring how to treat load flexibility as a planning and reliability resource.
Meanwhile, in Texas, regulators are working to implement SB6, which directs regulators to impose mandatory flexibility requirements on new large loads above 75 MW, requiring them to be able to curtail power or switch to backup generation when ERCOT issues firm load‑shed orders during grid emergencies.
Importantly, BNEF’s findings are PJM-specific, given the market’s massive size, its unique supply mix, and tight reserve margins.
In ERCOT, for example, where the relative scale of data center growth is larger relative to total demand, demand response might be able to play a much bigger role in price relief, Kou explained.
Transition-AI 2026, Latitude Media’s flagship event at the center of the AI-energy infrastructure buildout, will convene developers, utilities, regulators, hyperscalers, and capital providers to move towards coordinated, sustainable solutions for AI-driven demand. Register now to join us in San Francisco April 13-14.
