Last week we looked at Google’s answer to PJM’s proposal that data centers bring their own power or risk being cut off during shortages: a nuclear deal with Constellation that is real, but narrow and years from delivering. On October 9, FERC weighed in on the proposal itself.
PJM’s plan was a response to a real shortage, with its latest capacity auction clearing well short of what the grid needs, according to POWER Magazine. The fix was blunt: new large loads could connect without new generation, as long as they accepted curtailment first.
FERC rejected that framework, the Interim Resource Adequacy Service, along with PJM’s “bring your own new capacity” rules. Two pieces survived: a definition of large loads as facilities of 50 megawatts or more, and a registry where those facilities report their location, ramp schedule, and capacity supply to PJM.
Why FERC Drew the Line
The objection was about jurisdiction, not engineering. Several provisions placed obligations directly on new large loads, which are mostly retail customers. Others effectively chose which retail customers would be curtailed and what they would be paid. That is state territory, and FERC said so: states “have a crucial role to play,” and the commission “strongly encourage[s] states to continue their important work on this issue.”
Commissioner David Rosner, concurring, was more pointed: “We need state regulators to establish clear rules for which customers should be curtailed first.” He also warned that PJM’s design could leave the capacity market signaling surplus supply while the region involuntarily sheds “thousands of megawatts” of demand.
Notice what FERC did not say. It did not find that PJM has enough power. It found that PJM was the wrong body to decide who goes without. Five weeks earlier, FERC had already rejected key elements of the backstop procurement PJM designed to close the shortfall. Two attempts to manage the gap have now stalled at the federal level. The gap itself has not moved.
Where That Leaves Developers
The curtailment question now goes to state commissions, with Ohio and Pennsylvania expected to lead, according to Jefferies analysts. Each state will move on its own schedule and reach its own answer, which means a developer siting a campus today is planning against rules that do not exist yet.
Google’s route works for a small set of buyers: companies large enough to sign multi-decade contracts with a nuclear operator that has uprate potential in the right place. Most developers are not in that set. For them, the standard interconnection queue is still the default, and it is still measured in years.
The rejected design carried a lesson of its own. Even PJM’s faster path came with a condition: connect now, accept curtailment first. For an AI training cluster or a mission-critical tenant, uptime is the product. Speed bought with uptime is not a trade those operators can make.
Taking the Grid Off the Critical Path
PJM’s instinct was right, and FERC’s ruling does not change it: large loads need a dedicated source of power. What remains unsettled is who sets the terms, and when. Next Century-Power removes that dependency. Phase 1 is a 1GW+ dedicated gas turbine microgrid, islanded and single-tenant, with redundancy designed to a 99.99% uptime target. It does not wait on an interconnection study or a state rulemaking, and it does not appear on anyone’s curtailment list. Phase 2 transitions the site to an integrated pod of four 300MW small modular reactors for long-term cost stability, with on-site carbon capture along the way.
The new registry will give PJM a clear view of where large loads are and how fast they plan to ramp. It will not bring power to any of them sooner. The question for any developer is how much of their schedule they are willing to hand to a rulemaking. More on the NCP model at nextcentury-power.com.
