Three times now, PJM’s capacity auction has gone to its ceiling and stopped. The Base Residual Auction for the 2028/29 delivery year, with results announced July 14, 2026, cleared at $325/MW-day across the entire PJM footprint — the FERC-approved cap — for the third consecutive auction. The two before it did the same: $329.17/MW-day for 2026/27 and $333.44/MW-day for 2027/28, each at its own cap. A market that clears at its ceiling three times running is not discovering a price. It is reporting that the price it wants to discover is not permitted.
PJM has told us, in its own arithmetic, what that forbidden price is. Without the cap, the grid operator estimates, the July auction would have cleared at $554.72/MW-day across most of the footprint — roughly 71 percent higher — and at $777/MW-day in the Commonwealth Edison zone. Total capacity commitments would have come to roughly $29.7 billion instead of $16.4 billion. The cap, by PJM’s simulation, kept about $13.3 billion out of a single delivery year’s bill. That is the largest fact in this story, and it cuts both ways: the cap is doing exactly what its authors intended, and it is concealing exactly what a market is for.
What the cap is, and why it exists
The cap is one half of a “collar” — a ceiling of about $325/MW-day and a floor of $175/MW-day — that now governs four consecutive PJM capacity auctions. It was born in politics, not in market design. In December 2024, Pennsylvania Governor Josh Shapiro filed a complaint at FERC arguing that the existing cap would leave consumers paying up to $20.4 billion in added costs over two years without commensurate gains in new capacity or reliability. The governors of Illinois, New Jersey, Maryland, and Delaware backed him. FERC approved the collar 4-0 for the 2026/27 and 2027/28 auctions, and this spring extended it to the 2028/29 and 2029/30 auctions with the support of all 13 PJM-state governors, the White House National Energy Dominance Council, and the Department of Energy. Without the collar, FERC’s own decision noted, the cap for the next auction would have been about $500/MW-day — and the floor would have been zero.
The politics were responding to a genuine shock. The 2024/25 auction had cleared at $28.92/MW-day in the unconstrained region. The next one, for 2025/26, cleared at $269.92 — more than nine times higher in a single year — driven by surging load forecasts, retirements, and tighter accreditation of what counts as capacity. Bills jumped by double-digit percentages across the region. The collar was a tourniquet applied to a market whose price had gone from an afterthought to a line item voters could see. FERC has always described it as time-limited — a bridge, not a permanent feature. A cap is a confession of a particular kind: a capacity market exists to translate scarcity into a price that calls forth supply, and a cap says that when scarcity gets severe enough, the translation will be deliberately falsified.
What three straight at-the-cap clearings tell the market
The first thing they tell the market is that the demand curve has hit a wall. In the design described earlier in this series, a sloped demand curve is supposed to produce a different price for every balance of supply and demand: a little short, a little above the cost of new entry; very short, much higher. Three consecutive clearings at the cap mean the curve’s slope has been irrelevant for three years. Whether the system is modestly short or severely short, the answer is the same number. The auction has stopped discriminating between degrees of scarcity, which is another way of saying it has stopped transmitting information.
The second lesson is geographic: the map went flat. In the 2024/25 auction, five locational delivery areas were constrained and cleared at different prices — DPL-South at $426.17, for example, against $28.92 in the unconstrained region. In the July auction, every zone, from MAAC to ComEd to Dominion, cleared at exactly $325. Zero constrained areas. Zero price separation. A flat price map says nothing about where capacity is most needed. Locational signals, one of the capacity market’s most valuable products, have been zeroed out by the same instrument that capped the system-wide price.
The third lesson concerns the number everyone now quotes. The $554.72 figure is PJM’s own estimate of the uncapped clearing price — a simulation, not a market print. No buyer ever actually paid it; no seller ever actually received it. The honest statement is that the market is trying to say something above $500, and the most careful available estimate puts it near $555. Precision beyond that is false precision.
What the cap hides
What the cap hides, above all, is the shortfall. The July auction procured 138,318 MW of unforced capacity, plus 10,864 MW under Fixed Resource Requirement arrangements — and still fell 6,831 MW short of the reliability requirement, leaving PJM at a 14.7 percent reserve margin. It was the second consecutive auction in PJM history in which the region as a whole missed its reliability requirement; the 2027/28 auction had missed by roughly 6,500 MW. A market clearing at its cap while missing its target two years running is not a market in equilibrium. It is a market rationing a good it cannot supply at the permitted price.
The cap also hides how little new supply the price is producing. Of the capacity that cleared in July, only about 317 MW came from new generation and about 208 MW from upgrades to existing resources — 525 MW of new supply combined, less than 8 percent of the shortfall. The price has risen roughly elevenfold since the 2024/25 auction. It is sitting at the maximum the rules allow. And it is calling forth a rounding error’s worth of new steel. The textbook mechanism — high prices summon new entry — is failing, not because the price is too low in absolute terms, but because a one-year commitment at a capped price cannot finance a power plant. Developers build on revenue streams they can take to a lender, not a single year’s payment.
One more thing the cap hides: the cap itself has been falling while the market tries to rise. The 2027/28 cap was $333.44; the 2028/29 cap was $325, down 2.5 percent. The ceiling came down as the pressure went up — because the cap is set administratively, by negotiation and regulatory order, not by the market.
Who gains
Start with the obvious winners: consumers, in the narrow and immediate sense. The cap kept roughly $13.3 billion out of the 2028/29 delivery year’s capacity bill — the difference between a $16.4 billion commitment and a $29.7 billion one — and it is the reason every governor in the footprint supported the collar. The governors are winners too: the collar is a political instrument that delivered a political result, letting elected officials show voters they had done something about bills without taking responsibility for the supply shortfall the ceiling papers over. A quieter set of winners are the load-serving entities that hedged — PJM has noted that load hedged through self-supply or bilateral contracts is not exposed to the clearing price. In a rationed market, those with private arrangements eat first.
Who loses
The most visible losers are the owners of existing generation, who forwent roughly $13.3 billion in a single delivery year relative to the uncapped outcome. That figure deserves a caveat: at $325/MW-day, the existing fleet is still earning roughly eleven times what it earned in the 2024/25 auction. Nobody in the PJM fleet is being impoverished by the cap. When the price cannot rise to the market-clearing level, the first casualties are the marginal investments — the turbine overhaul, the life extension, the incremental upgrade — that never show up in an auction report but determine how much capacity actually shows up on a hot afternoon.
The deeper losers are the developers who would build new capacity if the price were real. The July auction’s 525 MW of new supply is the verdict. A one-year commitment at $325 cannot underwrite a new plant, and the cap ensures that no one-year commitment can ever be better than $325. New entrants do not lose a payday. They lose the ability to enter at all.
And then there are the ultimate losers, who are the same people as the winners, on a longer clock. The 6,831 MW shortfall did not disappear because the auction cleared at $325. It must still be filled — and the mechanism designed to fill it, the Reliability Backstop Procurement, carries a price cap of $555/MW-day, essentially the level the regular auction was not allowed to reach. If the backstop eventually buys its full target at something near that level, consumers will pay the uncapped price anyway, for the marginal megawatts, on top of the capped price for the rest. The cap may not have saved $13.3 billion so much as deferred part of the bill to a second, more expensive checkout lane — with 15-year contracts attached.
The backstop and the suspended bridge
That is what makes FERC’s September 29 order on the backstop so consequential. The Commission accepted the core of PJM’s one-time procurement — including the $555/MW-day volume-weighted maximum willingness to pay — but suspended it for five months, setting an effective date of February 28, 2027, and opening a paper hearing on three defects: cost allocation, transmission-owner exit rules, and collateral requirements for load-serving entities. The order, 196 FERC ¶ 61,245, was a split decision: the price survived, the plumbing did not.
The juxtaposition is the story. The same system that caps the regular auction at $325 has blessed $555 for the backstop — for the same underlying product, committed availability, differentiated only by contract length and administrative channel. The $555 figure is the system’s admission that the regular market’s price is wrong: PJM is offering new resources, outside the auction, the price the auction cannot offer inside it. A market that needs a second, parallel market to say what it actually thinks has a primary price that has become decorative.
What the refile must resolve
PJM has told Utility Dive, through spokesman Jeffrey Shields, that it intends to file a revised backstop proposal at FERC by October 29. That is a stated intention, not a deadline FERC imposed: the Commission set a five-month suspension and a paper hearing, not a calendar for PJM’s response.
Substantively, the refile must cure the three defects FERC identified, and each one is a version of the same question the cap raises: who pays. Cost allocation must be rebuilt around updated load forecasts that capture the growth from large new loads, rather than a snapshot of today’s load. Transmission-owner exit rules must assign stranded backstop costs to the load-serving entities within the zone, not to the exiting corporate entity. Collateral requirements for load-serving entities must be rebalanced so they protect the market without imposing postings so large they price smaller participants out of the process. FERC has signaled that a satisfactory refiling could prompt it to hold the paper hearing in abeyance — which makes the refiling, not the hearing, the path forward.
But the refile must also resolve something FERC’s order left implicit: whether a $555 backstop can coexist, indefinitely, with a $325 regular auction. If the answer is yes, then PJM has effectively split its capacity market into two tiers, and the honest debate is about whether that two-tier system is a sensible design or an admission of failure. If the answer is no, then either the collar must go or the backstop must be repriced down to a level that cannot finance the entry it exists to buy. There is no third option.
The price the market is not allowed to find
Former FERC Commissioner Mark Christie once called the capacity construct “an administrative construct, not a market.” Three auctions at the cap give that critique its sharpest illustration. A market whose price is set by settlement among governors, whose ceiling falls while scarcity rises, whose map is flat, and whose real price must be sought in a parallel procurement is not obviously a market at all. It is a negotiated allocation wearing a market’s clothes.
None of this means the cap was a mistake. The counterfactual — a $555 print in July, landing on bills across 13 states while the supply it was supposed to summon could not possibly arrive in time — was a genuine policy disaster that the collar averted. The cap bought time. The question is what the time was bought for. If it was bought to build the supply, reform the cost allocation, and design a market that can price scarcity honestly, then $13.3 billion in deferred costs was a reasonable price. If it was bought to postpone that reckoning, then the bill is still coming — at $555, in a backstop, with interest. The December auction for the 2029/30 delivery year, closing December 15, is the fourth and final collar auction; it will presumably clear at its cap too. Watch the refile, watch December, and watch what anyone proposes for the first auction with no collar.