PJM's December capacity auction will clear at its price cap for a fourth straight time

Bottom line.

1. Three PJM capacity auctions in a row have cleared at the regulator-approved cap, and the latest fell 6,831 MW short of the reliability requirement.
2. The December auction faces a requirement 2,699 MW higher, and PJM's demand curve holds the price at the cap until procured capacity is about 11,370 MW above July's total.
3. FERC's October 9 rejection of PJM's large-load rule closed the main route to a lower price; the question of who pays and who is curtailed first now moves to state regulators and PJM's next filing.

Planning assumption. By December 31, 2026, PJM's 2029/2030 capacity auction clears at its $337.83/MW-day cap across the whole footprint and procures less capacity than the 158,712 MW reliability requirement (85 percent: very likely).

PJM Interconnection, the grid operator for 13 states and Washington, D.C., runs its next capacity auction from December 9 to 15 and posts results on December 22. The last three cleared at the regulator-approved ceiling, and the most recent fell 6,831 MW short of what PJM says reliability requires. This forecast puts an 85 percent probability on the 2029/2030 auction clearing at its $337.83 per MW-day cap while again falling short.

What the auction is and why it stopped working as a price signal

A capacity auction buys promises rather than electricity: generators and demand-response providers commit to be available years ahead (normally about three; this auction is for the year starting June 2029), and PJM pays them a clearing price per megawatt per day, which can differ by zone when transmission limits bind. The price is meant to rise when supply runs short, pulling in new plants, and fall when they arrive.

For three auctions running, the price has instead hit its ceiling. The auction for the 2025/2026 delivery year cleared at $269.92/MW-day for most of PJM, up from $28.92 a year earlier, with the BGE and Dominion zones higher at $466.35 and $444.26, according to PJM's results report. Pennsylvania's governor filed a complaint, and FERC approved a negotiated price "collar" for the next two auctions in April 2025 (Utility Dive). Both auctions under it cleared at the cap: $329.17/MW-day for 2026/2027 (PJM) and $333.44/MW-day for 2027/2028, which came in 6,623 MW short of the reliability requirement (PJM). In April 2026 FERC accepted extending the collar to the 2028/2029 and 2029/2030 auctions, with support from all 13 PJM governors, the White House National Energy Dominance Council and the Department of Energy (PJM).

The 2028/2029 auction, posted on July 14, 2026, cleared at the cap of $325/MW-day. It procured 149,181.6 MW of capacity, including utilities that supply themselves outside the auction, against a requirement of 156,013 MW, a shortfall of 6,831 MW. The auction alone cleared at a 14.4 percent reserve margin, or 14.7 percent including self-supplying utilities, against a 20 percent target, and only 317.1 MW of new generation cleared (PJM results report).

PJM and its independent market monitor both point to data-center demand as a major driver. For the 2027/2028 auction, PJM said "nearly 5,100 MW" of a roughly 5,250 MW forecast peak increase was "attributable to data center demand" (PJM). The market monitor puts $6.3 billion of the $16.4 billion in 2028/2029 capacity charges, and $29.4 billion or 46.2 percent across the last four auctions, on data-center load (market monitor presentation). PJM's own report adds a second cause on the supply side: "A price cap below the established VRR Curve can reduce the amount of investment, and therefore supply, in the PJM region" (PJM results report).

The base rate: capped scarcity markets stay capped until steel arrives

The class of situation here is a scarcity market in which a political price cap sits below the price the market's own rules would set, while new supply takes years to build. The cap holds the price down but does nothing to speed construction, so it is reasonable to expect the capped price to persist until physical supply catches up. The nearest comparison is the neighbouring operator: MISO's summer capacity price fell from $666.50/MW-day for 2025–26 to between $384 and $424 for 2026–27 once 5.6 GW of new accredited capacity arrived (Utility Dive). The price fell only after the new plants actually existed.

PJM's December auction does not have that supply. Three numbers set the arithmetic.

First, the target moved up. PJM's planning parameters for 2029/2030 set the reliability requirement at 158,712 MW, 2,699 MW higher than the year before, and the forecast peak at 171,916 MW, 3.6 percent above the prior auction's 165,954 MW (PJM planning parameters). The same document sets the collar at a $337.83/MW-day cap and a $181.91 floor.

Second, the gap to close is large. To meet the new requirement, procured capacity would have to rise about 9,530 MW above the 2028/2029 total (158,712 minus 149,182). Between the last two auctions, offered supply rose 3,447 MW.

Third, the equipment that would close it is booked. GE Vernova's gas-turbine backlog and slot reservations reached 116 GW in the second quarter of 2026 (GE Vernova, SEC filing); Siemens Energy reports a 69 GW gas backlog and lead times of "three years or more" (Utility Dive). The one nuclear restart that would add firm capacity inside PJM, Crane, the former Three Mile Island Unit 1, is not running yet; Constellation still expects to restart it in 2027 (World Nuclear News).

The most likely course, December 2026 to mid-2027

Bidding opens on December 9 and closes on December 15 (PJM auction schedule). Offered supply grows by a few gigawatts, much of it from demand response, uprates and resources that were offered above the cap last time and did not clear (1,276.8 MW was offered in July but did not clear). That is not enough to cover a requirement that has itself risen 2,699 MW. On December 22 PJM posts a result at the $337.83 cap for the whole footprint, with a shortfall in roughly the same range as the last two auctions.

The headline number then feeds three processes already in motion.

The first is the emergency procurement. In July, PJM's Board directed a "Reliability Backstop Procurement" sized to the 2028/2029 shortfall (Board letter). On September 29, FERC accepted it but suspended it for five months, to February 28, 2027, setting a paper hearing on cost allocation, transmission-owner exit rules and collateral (196 FERC ¶ 61,245, Docket ER26-3380). FERC Chairman Swett said the commission "will not be forced into accepting a deeply flawed, eleventh-hour procurement mechanism" (Utility Dive). PJM plans a new filing on October 29, with bidding proposed for March 1 to 22, 2027 and results on May 3 (PJM stakeholder presentation). A fourth capped shortfall strengthens the case for approving it quickly.

The second is who pays. The Department of Energy told FERC on October 7 that "large electric energy loads — not American households or other business ratepayers — must fund" the new supply (Utility Dive). States have already moved the same way. Virginia regulators created a Dominion rate class for loads of 25 MW and above, effective January 1, 2027, with 14-year contracts and minimum charges (Virginia SCC order). Ohio's regulator approved an AEP Ohio tariff requiring new data centers over 25 MW to pay for at least 85 percent of subscribed capacity (POWER). Each capped auction adds to the pressure to push the next increment of cost onto the data centers themselves.

The third is a rule PJM proposed and FERC turned down. The same July Board letter directed PJM to file an "Interim Resource Adequacy Service," under which new large loads without enough capacity behind them as of June 1, 2027 would be curtailed before PJM called on its pre-emergency demand response, and incremental new large loads would be left out of auction demand from the 2029/2030 auction onward. The planning parameters posted on August 31 said the demand curves were "subject to adjustment pending FERC acceptance of PJM's IRAS filing in Docket No. ER26-3515." On October 9, FERC accepted only PJM's definition of a large load and a large-load registry and rejected the rest, including the interim service, saying several provisions "appear to place obligations directly on Eligible New Large Loads" (197 FERC ¶ 61,019, Docket ER26-3515; Utility Dive). PJM says it is reviewing the order. Unless a revised proposal is approved before bidding opens, no approved rule removes those loads from December's demand. FERC also said the provisions intrude on state jurisdiction, so the question of who is curtailed first, and who pays, moves back toward state regulators and PJM's next filing. Either way, developers have a strong reason to bring their own generation, and that generation draws on the same booked turbines and slowly restarting reactor listed above.

The cost reaches retail bills slowly but visibly. The US average residential price was 18.16 cents per kWh in August 2026, up 3.4 percent on a year earlier (EIA), and EIA's October outlook expects 18.2 cents for 2026 and 18.7 cents for 2027 (EIA STEO). EIA notes that retail prices also reflect natural gas prices, weather, transmission and distribution upgrades and, in some regions, capacity auction prices, so the rise is not due to data centers alone.

The rivals, and what would let them win

A clearing price below the cap across the footprint is the main rival, at 7 percent. PJM's posted demand curve holds the price at the cap up to 160,551 MW (PJM planning parameters workbook), so a below-cap price needs procured capacity to rise about 11,370 MW above July's total, more than the 9,530 MW needed just to meet the requirement. PJM's chief executive wrote on September 1 that "there will need to be substantial load reduction for the auction to clear at a price lower than the 'collar cap'" (PJM). The one route that did not depend on new plants has largely closed: with the large-load exclusion rejected on October 9, it now needs a refiled exclusion approved before bidding opens. Some signals still point down: PJM's 2026 forecast cut its 2028 peak by 4,414 MW from the 2025 forecast, citing updates to electric-vehicle forecasts, economics and large-load adjustments, including a new requirement that near-term large loads show firm commitments (PJM 2026 load report), and PJM was not on NERC's 2026 summer list of elevated-risk areas (GridBeyond, summarising NERC).

A delay that pushes the results past December 31 is the second rival, at 5 percent. The schedule was published in August and the collar is already approved, so a delay would need a fresh FERC or court intervention. The remaining 3 percent is something else, such as a result at the cap with no shortfall.

The main claim is narrow by design: one number on one date. The scenario around it, of costs shifting to large loads and developers building their own power, is what the dated register below tests.

Dated register

These secondary predictions are published with the forecast and will be checked against the sources named; only the headline claim is scored automatically.

1. The 2029/2030 auction's shortfall against the reliability requirement is between 3,000 and 9,000 MW (65 percent; PJM results, December 22, 2026).
2. FERC accepts PJM's refiled backstop procurement without suspending it beyond February 28, 2027, by that date (55 percent; FERC eLibrary).
3. Backstop procurement results are posted by May 31, 2027 (45 percent; PJM).
4. Constellation states by December 31, 2027 that Crane has returned to service (60 percent; Constellation, NRC).
5. PJM's 2030/2031 planning parameters, due February 1, 2027, show a price cap or collar in force for the May 2027 auction (60 percent; PJM).
6. EIA's December 2027 outlook puts the 2027 US average residential price at or above 18.7 cents per kWh (60 percent; EIA STEO).

What to watch

Four things tell an outsider whether this course is on track. Before December 9, look for any refiled large-load exclusion at FERC; without one, the demand curve keeps the 2,699 MW higher requirement. On December 22, look at the press release's headline and opening paragraphs, where PJM stated the price and the shortfall in each of the last two auctions. Around the end of December, look for FERC action on the refiled backstop procurement, which a shortfall makes more urgent. On February 1, 2027, look at the 2030/2031 planning parameters.

That last date carries the fact that would most change the longer story. FERC's April 2026 acceptance covered the collar only through 2029/2030. Whether any cap applies to the auction in May 2027, the first back on PJM's normal schedule, is not settled in any public document we found. Without one, the demand curve's own ceiling would apply; for 2026/2027 that ceiling would have been about $500/MW-day, and PJM's own no-cap simulation of the July 2026 auction cleared at $554.72/MW-day across most of the footprint and $776.69 in the ComEd zone (PJM results report).

What this means

For data-center developers in PJM, the planning case is capacity at the cap for every delivery year through 2029/2030, with the price beyond that open. Projects that arrive without their own firm supply carry both cost and curtailment risk, so on-site or contracted generation belongs in the base plan rather than the contingency plan, and equipment queues are the binding constraint on that plan.

For utilities and state regulators, the October 9 order puts the design of large-load tariffs back in their hands. The Virginia and Ohio tariffs are the existing templates; the open choice is whether new loads pay for capacity directly or are curtailed first when it runs short.

For investors, the signal to watch is not the December price, which is close to fixed, but the size of the shortfall and whether a cap is extended to the May 2027 auction. Those two numbers set the revenue ceiling for new capacity in PJM.