PJM Capacity Auction Explained: Why Prices Jumped Elevenfold

The PJM capacity auction is how the largest US grid operator pays power plants, batteries and flexible customers to be available when the grid is under stress, and its price jumped from about $28.92 per megawatt-day for 2024/25 to about $269.92 for 2025/26 and about $329.17 for 2026/27. Three things drove the rise: forecasts of data-center demand, retirements of older plants faster than new ones could connect, and rule changes that credited many plants with less capacity. The cost reaches every one of the roughly 65 million people PJM serves through their electricity bills.

On 30 July 2024 PJM, which runs the grid for thirteen states and Washington, D.C., from New Jersey to Illinois and including the data-center corridor of northern Virginia, published the result for 2025/26. The price of being available had risen roughly ninefold in a single auction. Within weeks, investors were repricing companies that owned existing plants. Vistra, which had lost about $1.6 billion in the 2021 Texas freeze, became one of the best-performing stocks in the S&P 500 in 2024 (reported). Governors in Maryland, New Jersey, Pennsylvania, Virginia and Illinois started attacking PJM's governance.

What is a capacity market?

A wholesale electricity market pays generators for the energy they produce, at a price set every five minutes by the most expensive plant needed. That works for plants that run most of the time. It fails for the plants a grid needs on its few worst days, such as gas peakers that run a few hundred hours a year. In ordinary years they cannot earn enough from energy sales to pay for being there. This shortfall is called the "missing money."

Markets answer it in one of two ways. Texas's ERCOT relies on scarcity prices, letting energy prices spike, up to $5,000 per MWh since 2022, in the hours when supply is tight. PJM and the other Eastern markets run capacity markets, which pay separately for a commitment to be available. The trade-offs between those designs are set out in our comparison of regulated vs deregulated electricity markets, and the book's chapter on power markets works through both.

How does the PJM capacity auction work?

PJM's capacity market is called the Reliability Pricing Model. Its main event is the Base Residual Auction, normally held about three years before each delivery year, which runs from June to May.

The price is set where supply offers cross the demand curve. Because the curve is steep near the requirement, a few gigawatts of extra demand, retirements or accreditation cuts can move the price by hundreds of dollars per MW-day. The penalties are real: during Winter Storm Elliott on 23–24 December 2022, penalties of about $1.8 billion were assessed (reported).

Accreditation matters as much as the price. A plant's nameplate rating is not what it is paid on. The book's chapter on reserves and the evidence problem gives a stylized comparison in which 1,000 MW of nameplate nuclear might be accredited at about 950 MW, gas at about 800 MW, onshore wind at about 150 MW and solar at about 100 MW. Those are illustrative values, not PJM's ratings, but the gap is the point.

What were the PJM capacity auction results?

To translate the units, $329 per MW-day is about $120,000 per MW-year, or about $10 per kilowatt-month. In PJM's 2027/28 auction, about 5.1 GW of the roughly 5.25 GW forecast load increase was attributed to data centers (PJM, reported).

Why did PJM capacity prices rise so much?

Three causes overlapped, and the steep demand curve amplified each.

Data-center load. PJM's load forecasts rose sharply as technology companies planned data centers measured in gigawatts. Dominion Energy, the utility serving northern Virginia, reported a data-center pipeline of about 54 GW in stages of contracting, with about 12 GW under signed service agreements, in the third quarter of 2026. The gap between those two numbers is itself a lesson: only a minority of large-load requests become signed agreements.

Retirements faster than new entry. Coal and older gas plants were retiring, and new plants could not replace them quickly. The US interconnection queue held about 2,060 GW of projects waiting to connect, with a median of about 61 months from request to operation (Lawrence Berkeley National Laboratory, reported), and gas turbines were sold out for years.

Accreditation reforms. Changes to how PJM credited capacity at times of stress cut the megawatts many plants could offer, shrinking supply on paper even where nothing physical had changed.

The 2025 price collar, negotiated after Pennsylvania Governor Josh Shapiro challenged the market design (reported), limited the cost to consumers. It also limited the signal to build. The uncapped 2027/28 price of about $529.80 shows what the market thought scarcity was worth.

How do capacity prices affect electricity bills?

Capacity costs flow through retail rates to every customer. The book uses a stylized household whose share of system peak, including the reserve margin, is 2.5 kW, and a nuclear unit with 950 MW of accredited capacity. The numbers are round, not PJM's tariff.

For the household, a line that cost about $2 a month became one costing $20–25, with no change in usage. Many households saw smaller increases spread over time, depending on how their utility passed costs through. For the owner of the nuclear plant, a rounding error became about $100 million a year for a plant that already existed and needed no new investment. That transfer, from customers to existing generators, is why the auctions became a political fight. A capacity price is a signal to build, but it is also a bill, and the bill arrives first. The book's chapter on following the money shows where that line sits in a full electricity bill.

Who benefits from high capacity prices?

The owners of plants that already have a grid connection and accredited capacity. Investors re-rated merchant generators from declining businesses into owners of the scarcest product in American infrastructure: firm power with a connection. Consolidation followed. Constellation completed its purchase of Calpine on 7 January 2026, creating a fleet of about 55 GW. NRG bought about 13 GW from LS Power, closing on 30 January 2026. Vistra agreed to buy Cogentrix's roughly 5.5 GW for about $4.0 billion net in January 2026.

Developers tried to bypass the grid by building data centers next to power plants. FERC rejected an expansion of co-located load at Talen's Susquehanna nuclear plant on 1 November 2024, by two votes to one, after utilities and consumer advocates argued it would let a single buyer use grid services without paying and take supply out of the market as prices spiked. The deal was rebuilt as a grid-delivered contract, explained in our article on how power purchase agreements work. FERC set co-location rules for PJM in December 2025, PJM adopted rules for loads of 50 MW and more on 16 January 2026, and FERC issued show-cause orders on large loads to all six RTOs and ISOs on 18 June 2026, without a final rule (reported).

What capacity markets do and don't do

Capacity markets are better at keeping existing plants open than at getting new ones built. Three years is shorter than the time needed to permit, connect and build most new plants in 2025–26, and a one-year price gives lenders little certainty. The 2024–26 price shock therefore rewarded mainly the owners of plants that already existed.

They also struggle to value performance in the hours that matter. Much of the eastern US's winter reliability risk is correlated gas-supply failure, which a megawatt of gas capacity on paper does not capture. Performance penalties and ELCC accreditation are attempts to price that risk. Texas learned the cost of correlated failure in February 2021, as our case study of the Texas power crisis describes.

Other markets have their own versions. MISO's summer 2025/26 capacity auction cleared at about $666.50 per MW-day, up from about $30 the year before, before falling to about $424.30 for summer 2026/27 in its North and Central zones as surplus rose (reported). Great Britain holds capacity auctions four years and one year ahead.

How the PJM shock fits the power equation

The Power Equation holds that an energy company creates durable value only when four factors hold at once: supply at a cost the market will pay, a route to market, capital matched to the asset, and a licence to operate. The PJM shock turned on the second factor. The plants that gained most had no new technology and little new cost advantage. What they had was a grid connection and accredited capacity at the moment those became scarce, in a market whose rules turned scarcity into a price.

The same case shows how quickly the route can be contested. The co-location ruling, the price collar and the governors' interventions were all arguments about the fourth factor, the licence, and specifically about who should pay. Utilities facing the same load are writing large-load tariffs with minimum bills and exit fees so that data centers, not other customers, carry the risk of demand that does not arrive; our guide on how to read a utility rate case covers that side. For the wider industry picture, see our guide to how the energy industry makes money and the book's study of power markets, wholesale electricity and capacity.

The takeaway

PJM's capacity auction pays resources to be available at times of stress, and its price rose about elevenfold between 2024/25 and 2026/27 because data-center forecasts, retirements and accreditation changes hit a steep demand curve at the same time. The money flowed mainly from households and businesses to the owners of plants that already existed, which is why the auctions became a political fight and why the price has sat at a negotiated cap since 2026/27. A capacity market sells a commodity, but its price is a set of rules, and the rules get rewritten whenever the bill becomes political. Chapter 8 of The Power Equation explains how wholesale power and capacity are priced.