The country's largest grid operator tried to hold an emergency power auction this week and got told no by its own regulator. That's the short version of what happened when FERC partially rejected PJM's reliability backstop procurement on September 29, forcing PJM to cancel an auction window that was supposed to open the very next day. The longer version is a story about who pays for the electricity data centers are devouring, and PJM still doesn't have an answer the commission will accept.
The Shortfall Is Real. The Billing Address Isn't Settled.
PJM needs roughly 6.8 GW of new capacity to cover a shortfall for the 2028/2029 delivery year, a gap driven largely by data center demand forecasts that have outpaced two consecutive base capacity auctions. The grid operator's fix — a Reliability Backstop Procurement — would lock in new generation under 15-year commitments rather than the usual one-year capacity contracts, an unusual structural shift meant to actually get steel in the ground instead of just moving money around.
FERC didn't kill the plan. It suspended it. The commission said pieces of PJM's proposal dealing with cost allocation, transmission owner exit rules, and load-serving entity collateral requirements "may be unjust and unreasonable" and sent PJM back to fix them, with FERC Chairman Laura Swett calling the filing a "mess" that arrived with no time for the agency "to rehabilitate" before the deadline. PJM has no new auction timeline. The window that was supposed to run September 30 through October 21 is simply gone.
That delay matters more than it sounds. A cost allocation dispute isn't a technicality — it's the entire question this topic is actually about. PJM says it intends for the large loads causing the shortfall to pay for the capacity the shortfall requires. Intent isn't a mechanism, and FERC just confirmed the mechanism isn't built yet.
Three Auctions at the Ceiling Is Not a Market Clearing
Capacity prices in PJM have been running at or near the administrative price cap for multiple consecutive auctions, a pattern that one industry analysis pegs at roughly $329/MW-day for the 2026/2027 delivery year and near the cap again for 2027/2028 and 2028/2029 (per lib.ink's industrial-cost reporting). I'd treat the precise dollar figures there with some caution — it's not a primary filing — but the structural point survives scrutiny: an auction that clears at its ceiling three times running isn't discovering a price, it's reporting that demand has permanently outrun the cap designed to contain it. The proposed 15-year backstop contracts carry a $555/MW-day cap specifically because PJM expects the standard auction mechanism can't attract enough new supply on its own.
That's the tell. When a market operator has to design a parallel, longer-duration, higher-ceiling procurement track just to get new generation built, the standard capacity market has stopped doing its one job — signaling scarcity cheaply enough that someone builds a plant in response.
Who Actually Picks Up the Tab
PJM has begun a separate, related process that bears directly on this: its annual review of proposed Large-Load adjustments to the 2027 Long-Term Load Forecast, with roughly a dozen utilities — including Dominion, NOVEC, FirstEnergy, and Duke — presenting data center and large-user demand projections to PJM's Load Analysis Subcommittee. PJM retained Charles River Associates to independently assess how much of that demand will actually materialize, with findings due in October and a final load adjustment published in January. That forecast exercise is the quiet upstream input to the cost allocation fight: overstate large-load demand and you procure capacity nobody needs at ratepayer expense; understate it and you're back here next year with another emergency auction.
Clean energy advocates have flagged that PJM's reforms can't succeed without state-level follow-through on implementation, a point raised in discussion of the reforms around NRDC's commentary on the filings. That's a fair structural observation, though the specifics of what states need to do remain thin in public reporting so far.
What to Watch
The real test isn't whether PJM resubmits a fixed backstop proposal — it will, probably within weeks, given Swett's public pressure. It's whether the revised cost allocation mechanism actually assigns the 15-year contract costs to the large loads driving the shortfall, or whether it gets diluted across the broader ratepayer base the way capacity costs so often do once the lobbying starts. Watch for PJM's refiling at FERC, the Charles River Associates load forecast due this month, and the January 2027 final load adjustment — each one will tell you whether "the loads causing the problem pay for it" survives contact with an actual tariff.
