As the grid transitions from decades of surplus to a new era of shortage, PJM has become ground zero for a debate over whether data center load is uniquely responsible for the current capacity crunch — or simply a forcing function that’s accelerating a shift the region has long seen coming.
Last week, PJM’s independent market monitor filed an emergency complaint with FERC, seeking to block the grid operator from interconnecting data center loads until it can guarantee it has the supply to serve them. The filing framed data centers as “the primary reason” for PJM’s recent price spikes, and described the region’s efforts to fast-track data centers that agree to be curtailed as “clearing the capacity market at the maximum price and at a quantity less than the reliability requirement.”
Blocking data centers or requiring them to bring their own matching generation will prevent “a massive wealth transfer” from ratepayers to generation owners of roughly $16.6 billion, the complaint added, pointing to increase in capacity market revenues over the course of the last two auctions.
But observers argue that that lens misses the fundamental mechanics of capacity markets, and the reality that skyrocketing prices in PJM were inevitable. “Even without data centers, load growth from other causes would still eventually cause the same capacity market cost increases,” explained Brent Nelson, managing director of markets and strategy at Ascend Analytics. That’s because capacity prices reflect the price of getting new generation to market, not the cost of supply.
PJM and its market monitor have been predicting capacity shortages for several years, Nelson pointed out. The data center boom didn’t break the capacity market construct; it simply sped up the timeline of those concerns.
“The IMM talking about capacity market costs as a data center problem appears to me to largely be a red herring,” Nelson said. “New entry inherently requires either high capacity prices or some other mechanism that flows revenue directly to new entry.”
For Nelson, another issue is the effort to single out data centers as the singular source of all grid-related problems.
“I have broader concerns that we’re starting to identify ‘good’ load and ‘bad’ load,” he explained. The market monitor, in its complaint to FERC, essentially argues that PJM should be able to choose not to serve a specific class of load: “I wonder how this would sound if the IMM were arguing that PJM shouldn’t add other forms of new load that might drive a shortage in capacity markets [such as] a gigafactory or an EV charging network or a heat pump adoption program.”
From excess to shortage
For years, PJM operated with excess capacity. The grid was flush with older, fully depreciated power plants that only needed to cover their operating costs. In that environment, capacity markets cleared at rock-bottom prices, as low as $50 per megawatt-day, because the market didn’t need to incentivize any new generation.
But as soon as the market reaches a capacity shortage and needs to spur new generation, the math changes entirely.
New entry requires prices more along the lines of $200-$300 per MWd to justify putting steel in the ground. At $50 per MWd, nothing would get built.
In a uniform clearing price market like PJM, all generation gets paid the highest price — including existing plants that were previously happy with the lower clearing prices. The result is that total costs explode, even to secure just a small amount of new capacity. Which explains why, in the market monitor’s view, even flexible data centers might not save the market.
In early November, the PJM market monitor sharply criticized PJM’s efforts to design a pathway by which data centers that agreed to be flexible could gain priority access to the grid. In its annual report the monitor found that if just 10% of the data centers in PJM’s queue opted out of flexibility, the extra cost for consumers would still be around $396 million per year.
The report inspired discussion and debate, but according to Nelson, it wasn’t a surprise; that’s simply the reality of a capacity market. If that 10% is enough to push the market from a surplus condition to a shortage condition, then the market price for everyone is set by the cost of new entry. It’s essentially an on/off switch, and once the market is in shortage conditions, prices will go up, regardless of how much new supply is needed and what type of load it’s powering.
The only way to get around that reality is to funnel revenue specifically to new sources, rather than to the entire existing fleet. That means relying less on the general capacity auction and more on targeted mechanisms like state-sponsored solicitations or direct bilateral contracts between retailers and developers, he explained.
Those conversations are happening, and some states functionally have that type of mechanism, but the concept is still very nascent. “I don’t think people have really thought through the implications of capacity markets,” Nelson said. “Everybody is freaking out about how high the prices are, but two years ago they were freaking out about how the prices were too low and we were going to have retirements.”
Speed sticker shock
That said, there are a few key differentiators that make the incoming data center load particularly challenging.
“The challenge that we have right now is that all of this is unprecedented,” said Julia Hoos, who covers the eastern U.S. at Aurora Energy Research. If the current load growth stemmed from residential load, “we would not be having this conversation,” she explained on a recent Latitude Dispatch.
Traditional utility planning is designed to handle incremental load growth: adding electric vehicles or heat pumps to the grid over the course of a decade, for example. When a customer requests 500 MW overnight, it starts to create reliability problems that the market can’t move fast enough to fix.
And no matter who pays for the infrastructure or generation, the reality is that very large, fast-growing loads like data centers inevitably cause complex ripple effects across the grid.
Even in parts of the country where utilities are experimenting with new retail tariffs, requiring large power users to commit to buying set quantities of power or covering infrastructure upgrade costs, the size and timeline of the data center buildout will inevitably impact ratepayers.
Hoos points to Dominion Energy’s territory in Virginia, long known as the epicenter of the data center industry, as a warning sign about how widespread the impacts may be. Dominion has thousands of pricing nodes, and at any given time around half of those have above-average pricing. This year, capacity pricing in Dominion’s zone reached record-high levels. However, just 100 individual nodes had an above-average price. In other words, those nodes saw such dramatic price spikes that they pulled up the average for the entire region, Hoos explained.
“When you start to deal with loads that are this large, it has all of these knock-on effects and costs down the line that can really impact [consumers] in ways that I think we’re not fully capturing yet.”
