PJM and the capacity crunch

The capacity market exists because of a bet made when electricity markets were restructured: that discovered prices would pull private investors into building generation. That produced two signals. Energy prices clear every five minutes and pay for electricity actually delivered. Capacity prices, quoted in dollars per megawatt-day, were meant to be the durable, longer-horizon signal that new plants are needed. PJM historically ran that auction four years ahead precisely so a developer had time to respond: price turbines, line up financing, build. Stakeholder challenges and rule revisions forced PJM to pause auctions for several years, and it fell so far behind that it now procures only for the next reliability year. That sounds procedural and is not. The entire theory of the design was lead time, and the lead time is gone.

The price move comes from a confluence rather than a single cause. For years capacity cleared low, falling as far as about $30 per megawatt-day against roughly $100 in the early 2010s, which PJM and other stakeholders came to worry was too weak to sustain investment and was encouraging early retirements. Demand then grew quickly, with data centers the lead story but electrification and electric vehicles named alongside them, and development spilling out of Northern Virginia into western Pennsylvania and North Carolina as the original hub congested. On top of that PJM revised how it accredits resources, so a nameplate megawatt now earns less reliability credit. Mechanically that is what lifts the price: with each bid counting for less, more bids stack up the supply curve before it crosses the demand curve, and the crossing point sits higher. Last year’s auction cleared near $270 per megawatt-day and hit its cap; this one cleared at a new cap of about $329 negotiated in a settlement with Pennsylvania, roughly 20% higher in round numbers. Both auctions therefore tell you less than they might, because each time the market wanted to clear higher than it was permitted to.

Piper is deliberate about what this does and does not signal. Resource adequacy gets reported as blackout risk; he pushes that framing away and calls this an economic event, though a significant one. Customers in Ohio, Pennsylvania and Maryland are receiving notices that bills rise $20 to $30 a month this year from last year’s auction alone, with this year’s higher price flowing through after that. What a record price cannot do is summon supply on the schedule the auction now runs on. It is a strong signal to build gas, and the turbine manufacturers are ramping factories, but you cannot deploy many machines next year or the year after and a combined-cycle plant is a 2030 to 2032 conversation. Solar, wind and storage can be built inside a year or two, exactly the timeframe needed, but they are credited with far less reliability value, so they answer the capacity question much less than their megawatts suggest.

That points at demand response, which is fast to deploy, qualifies in PJM and is accredited relatively well at roughly 60% of nameplate against 8% to 10% for solar. It should have been the resource that responded. It brought about a gigawatt to the 2024 auction and, despite a far stronger price, slightly less this time, which surprised both speakers. Piper’s explanation is not price but risk: accreditation rules are uncertain and evolving on the fly, with open questions about whether data center load flexibility counts as demand aggregation and whether distributed storage competes with demand response or can be folded into it. He expects a period of adaptation before growth. The pattern generalizes beyond PJM. Every major grid operator has done two things at once, lowering the credited reliability contribution of wind, solar and storage while raising target reserve margins, with conventional gas also scrutinized after winter storms Uri in Texas and Elliott in PJM. Wanting more while counting less can only push prices one way, and they pass straight through to retail rates because the utilities serving customers have to pay the bills.