On September 30, 2026, PJM Interconnection was supposed to open bidding on the most unusual capacity procurement in its history. It did not. The day before, the Federal Energy Regulatory Commission issued an order accepting PJM’s one-time Reliability Backstop Procurement — and then suspended it for five months, setting an effective date of February 28, 2027, subject to refund and further proceedings. The auction designed to close a 6,831 MW reliability shortfall never started, and PJM has published no revised timeline.
The order, 196 FERC ¶ 61,245 in Docket No. ER26-3380-000, is best read as a split decision. FERC affirmed the core of the procurement: the eligibility rules, the selection process, and the $555/MW-day price cap all survived as just and reasonable. But the Commission found that PJM had not demonstrated that three elements — cost allocation, transmission-owner exit rules, and collateral requirements for load-serving entities — met the Federal Power Act’s just-and-reasonable standard. It set those issues for a paper hearing, opened a companion Section 206 investigation in Docket No. EL26-108-000, and strongly encouraged PJM to come back with a new Section 205 filing that fixes the defects. Until then, the backstop is frozen.
What the July auction revealed
To understand why PJM proposed a backstop at all, start with the regular capacity market — the forward auction for committed availability described earlier in this series, where load-serving entities collectively buy enough capacity commitments to cover forecasted peak demand plus a reserve margin. PJM’s Base Residual Auction for the 2028/29 delivery year, with results announced July 14, 2026, did something it had only done once before: it failed to procure enough capacity to meet the reliability requirement.
The numbers tell the story. The auction procured 138,318 MW of unforced capacity (UCAP) and cleared at $325/MW-day — the FERC-approved cap — across the region, for roughly $16.4 billion in total capacity commitments. Even so, procured capacity plus Fixed Resource Requirement resources fell 6,831 MW short of the reliability requirement. The prior auction, for 2027/28, had missed by roughly 6,500 MW. Those are the first two auctions in PJM history in which the entire region missed its reliability requirement.
The more damning figure is what the auction bought that was new. Of the capacity that cleared, only about 317 MW came from new generation and about 208 MW from upgrades to existing resources — 525 MW of new supply combined, or less than 8 percent of the shortfall. A market clearing at its price cap should be screaming for new entry. Instead it delivered a rounding error. That gap between the price signal and the supply response is the entire case for the backstop.
It is also worth noting what the cap concealed. Without the FERC-approved price collar, PJM said the auction would have cleared at nearly $555/MW-day across most of the footprint and $777/MW-day in the Commonwealth Edison zone, pushing total costs to roughly $29.7 billion instead of $16.4 billion. The collar protected consumers from that bill. But a cap that binds two auctions running is also a confession: at the price the market is allowed to pay, new supply is not arriving fast enough.
What the backstop was designed to do
The Reliability Backstop Procurement was PJM’s answer to that failure — a one-time, transitional procurement conducted outside the regular Base Residual Auction, aimed squarely at new capacity. Where the regular auction buys one-year commitments from the existing fleet plus whatever new projects happen to clear, the backstop was designed to sign long-term deals that make new projects financeable. Selected resources could receive fixed commitments of up to 15 delivery years, with a commercial-operation deadline of June 1, 2032.
The central design choice was the price. The backstop set a volume-weighted maximum willingness to pay of $555/MW-day in UCAP terms — roughly 71 percent above the $325/MW-day cap in the July auction, and essentially the level the regular auction would have reached without its collar. That is not a coincidence. PJM was, in effect, offering new resources the price the capped market could not: a level its staff judged necessary to attract new steel in the ground rather than reshuffle existing megawatts. The $555 figure is a weighted-average ceiling across accepted offers, not a single clearing price, and FERC explicitly found the price cap just and reasonable.
The initial central-procurement target was the shortfall itself: 6,831.3 MW UCAP. Eligibility was limited to new resources and resources that did not receive a commitment for the 2028/29 delivery year — the point was additionality, not a second payday for capacity the market had already bought. PJM’s board directed the filing on July 27, and PJM filed at FERC on July 31, asking for approval by September 29 so the bid window could open September 30 and run through October 21, with results before the December capacity auction for the 2029/30 delivery year. FERC met that deadline. It just did not give PJM the answer PJM wanted.
The bilateral track: the first door
The backstop was never only an auction. It was designed with two doors, and the first one — bilateral contracting — may matter more than the second in the long run. Beginning with a request for proposals launched June 9, PJM ran a facilitated matchmaking process connecting large electricity users, above all data centers, directly with developers of new supply projects. Loads that sign qualifying bilateral contracts or self-supply commitments are excluded from the central auction’s procurement target: every megawatt placed privately shrinks what the auction has to buy. Projects that enter matchmaking but fail to secure a bilateral deal remain free to bid into the central procurement.
The logic is cost causation, the principle that runs through this entire dispute: the costs of new capacity should fall on the loads driving the need for it. If the largest new loads contract for their own supply, the residual socialized procurement shrinks, and so does the bill everyone else pays. PJM has said bilateral contracting is the preferred path for new capacity to enter the market, and the matchmaking process was designed to run in parallel with the central procurement so developers had both routes open.
The bilateral track also connects to a broader PJM reform: starting with the 2029/30 delivery year, PJM plans to exclude new large loads from the demand forecasts used in general capacity auctions, so existing consumers do not bear higher capacity costs caused by new large loads that do not bring — or otherwise contract for — the new supply needed to serve them. A Large Load Registry is being developed to improve forecast accuracy and transparency. Taken together, the bilateral track and the forecast exclusion amount to a new doctrine: large-load growth should arrive with its own supply attached, or pay for it directly, rather than socializing the cost through the capacity market.
Why FERC suspended it
FERC’s objections were not to the procurement’s existence but to its plumbing — specifically, to whether the costs would land on the right parties. The Commission found three defects, all variations on the same theme.
First, cost allocation. PJM proposed to allocate backstop costs based on a snapshot of current load, and FERC found that PJM had not shown the method consistent with cost causation. A snapshot of today’s load does not capture the forecast growth from large new users — the very loads driving the shortfall. FERC offered an alternative: allocate costs using updated load forecasts, so the allocation “reasonably captures all forecasted load growth,” and use the most up-to-date load and resource forecasts to set the initial procurement target. The Commission also rejected PJM’s proposal to let qualifying cooperatives and municipal utilities opt out of the procurement, finding that carve-out could be unduly discriminatory against other load-serving entities with data centers in their footprints.
Second, transmission-owner exit rules. If a transmission owner leaves PJM before the backstop’s costs are fully recouped, PJM proposed assigning the remaining costs to the exiting transmission owner. FERC said PJM had not demonstrated that those costs would be directed to the entities causing them, and directed instead that such costs be assigned to the load-serving entities within the transmission zone. The principle is the same: costs follow causation, not corporate form.
Third, collateral. FERC found PJM’s collateral requirements for power suppliers appropriate, but its proposed collateral requirements for load-serving entities potentially flawed — the agency said they did not strike a reasonable balance “between protecting PJM’s markets and its members against the risk of default and ensuring that market participants have adequate flexibility to manage their collateral requirements.” The stakes were concrete: Northern Virginia Electric Cooperative said it would have faced roughly $2 billion in collateral postings under the proposal. A requirement that large, applied to a cooperative serving one of the fastest-growing load pockets in the country, illustrates why FERC hesitated.
FERC Chair Laura Swett’s concurrence was blunter than the order itself, saying the Commission would not be forced into accepting what she characterized as a deeply flawed, eleventh-hour procurement mechanism with billion-dollar implications for consumers. PJM, for its part, welcomed the acceptance, noted that FERC affirmed the substantial majority of the proposal, and said it intends to work quickly to address the remaining concerns. Both readings are accurate. That is what makes the order a genuine split decision rather than a polite rejection.
What the freeze means for new capacity
The immediate consequence is delay in a market where delay is expensive. The 2028/29 delivery year begins June 1, 2028. The original schedule — bids September 30 to October 21, results in early December, commitments finalized before the December 2029/30 auction — was already compressed. A five-month suspension pushes any procurement into early 2027 at the earliest, and February 28 is an effective date, not a confirmed bid-window opening. PJM has said only that previously communicated milestones are under review and that it will provide an administrative update at its Market Implementation Committee. The December 2029/30 Base Residual Auction, PJM has said, remains on schedule.
For developers, the freeze extends uncertainty about the single most important question: what price and contract length can a new project actually count on? The backstop’s 15-year commitments at up to $555/MW-day were designed to solve the financing problem that one-year capacity commitments at a capped $325 cannot. Every month without that mechanism is a month in which projects that might have been built for 2028/29 or soon after sit without a revenue anchor. The July auction’s lesson — that capped one-year prices do not call forth new supply at the needed scale — has not changed. Neither has the load growth behind it: PJM’s footprint includes the Northern Virginia data-center cluster, the largest concentration of data-center load in the United States, and data-center demand is widely cited as the dominant driver of the region’s forecast growth.
For consumers, the suspension changes nothing about bills already set. Capacity costs for already-cleared delivery years remain in place, and the suspension does not reduce current electricity costs. What it does is leave the future cost allocation unresolved: the same 6.8 GW hole, the same question of who pays, and a procurement mechanism that cannot open until FERC is satisfied with the answer. FERC also used the order to remind everyone that states have a key role in managing data-center load growth — a signal that the Commission does not believe federal market design alone can resolve the pressures the AI buildout is placing on the grid.
What to watch
Three things will determine what happens next. The first is whether PJM files a new Section 205 proposal curing the three defects — revised cost allocation based on updated forecasts, corrected exit rules, and rebalanced load-serving-entity collateral — and how quickly. FERC said such a filing could prompt it to hold the paper hearing in abeyance, which makes the refiling, not the hearing, the likely path forward. The speed of that filing is the single best indicator of when, if ever, the backstop opens.
The second is the bilateral track. Matchmaking continues independently of the suspended auction, and every qualifying bilateral deal shrinks the eventual central target. If large loads contract privately at scale, the backstop that eventually opens could be materially smaller than 6,831 MW — which would be the cost-causation principle working as designed. If bilateral deals stall, the full weight falls back on the central procurement and its unresolved allocation fight.
The third is the December 2029/30 auction. It will show whether the July shortfall was a two-auction anomaly or the new normal — and whether the market’s regular machinery, even with the large-load forecast exclusion, can begin closing a gap that two consecutive auctions at the price cap could not. The backstop was conceived as a bridge: a one-time fix to buy time for a broader review of market fundamentals and investment incentives. FERC’s suspension has lengthened the bridge’s construction while the gap it was meant to span stays exactly where it was.