What PJM's data center curtailment plan means for project pro formas

PJM Interconnection asked federal regulators in August to approve a framework governing how data centers connect to the grid and acquire power. The grid operator, which Bisnow describes as the largest in the U.S., is moving toward connecting only new data centers that agree to have their grid electricity cut when the system is under stress, and other regions are exploring similar approaches.

The backdrop is supply. Regional grids are struggling to keep pace with electricity demand from the artificial-intelligence data center buildout, and turning data centers into flexible loads — able to reduce consumption or disconnect from the grid entirely during periods of peak demand — has become the favored answer because it is presented as a way to facilitate digital infrastructure development without compromising grid reliability or putting upward pressure on consumers' utility bills. The concept is gaining traction among power providers and the industry itself, and Bisnow reports it is likely to be mandated in some major data center markets.

PJM's reach is geographic before it is financial. Its service area encompasses Washington, D.C. and 13 states across the mid-Atlantic and Midwest, and under the proposal, newly built data centers that do not add capacity of their own to the grid will have their power supply curtailed when the grid comes under strain. Whether that arrangement holds depends on how it prices at the project level, where the industry's enthusiasm thins out.

Industry leaders speaking this month at Bisnow's National DICE: Power event at the Sheraton Philadelphia Downtown said curtailment requirements of the kind PJM plans can increase project costs and lengthen development timelines, potentially pushing developers to site projects elsewhere. The same leaders said curtailment is not always a deal-breaker, and that as the power pinch intensifies nationwide, flexible-load requirements may become a reality developers accommodate rather than avoid. A new industry coalition launched this month to support flexible power consumption, which puts institutional weight behind the concept even as individual projects push back on its terms.

Giuseppe Marrari of Rolls-Royce Power Systems put the middle case at the same event. "I think curtailment-based interconnection can work," he said. "I don't think people prefer it, but I think there are strategies that can mitigate the effects of those curtailment situations."

Take the developers at their word and a curtailment obligation lands in three places in the underwriting, none of them the rent roll. A higher development cost is a hit to yield on cost before a tenant signs. A longer timeline is carry — interest, taxes, and the wait for first revenue. And a decision to site elsewhere is a land basis and a different interconnection queue. That is a cost-of-capital problem arriving earlier than the lease does, and the mitigation Marrari points to is itself spending: engineering and equipment bought to protect schedule and reliability rather than to serve load. The coverage does not say how curtailment terms are handled in leases or loan documents, which are the two documents where the obligation would either get priced into the deal or parked outside it.

A schedule that stretches moves every date behind it, including the point at which a project can support permanent financing and the point at which a sponsor could sell a stabilized asset. If curtailment pushes a build out by months, the sponsor holds the asset longer than the underwriting assumed, and the equity that bridges the gap is the most expensive money in the deal. This is the part of the story that reads differently from the debt side than from the development side: the lender or equity partner is reading the same interconnection agreement the developer signed, with the same limited ability to change its terms after the fact.

If mitigation is the answer, then the requirement stops operating as a uniform tax on data center development and starts sorting sponsors by what they can spend. Mitigation is engineered and paid for, which suggests developers with in-house power teams and balance sheets that can carry a longer, costlier build absorb these terms more cheaply than developers without them, and that the projects likeliest to relocate are the ones with the least room to fund the workaround. Whether that dispersion reads as an advantage for the largest developers or as a slower national buildout is not something the coverage settles.

What gives the framework its staying power, if it holds, is the argument the industry itself is making. A flexible data center is sold as the way to add load without adding upward pressure to consumer utility bills, which suggests the rule's political durability rests on consumer cost rather than on developer economics, and it likely leaves the mitigation bill with the sponsor: the reliability benefit is collective, and the expense of buying it back is specific to the project. That is the trade the developers at Bisnow's event were describing, and it is a trade they expect to be making more often rather than less.

Federal regulators now have the August filing. The next observable step is not the rule itself but the contracts underneath it — whether interconnection agreements carrying curtailment conditions begin to appear inside the PJM footprint, and whether the land prices behind those projects move before the leases do.