PJM Just Made Grid Power Conditional for Data Centers

On July 27, PJM Interconnection's board directed staff to file a proposal with FERC that changes a basic assumption the data center industry has been building on for two years: that a signed interconnection agreement means firm power. It doesn't anymore, or at least it won't after June 2027, a shift tech and energy press picked up on within a day as a sign the AI buildout had hit a hard limit.

The proposal defines a "Large Load" as any customer with 50 MW or more of cumulative peak demand at a single site within a one-mile radius. Starting June 1, 2027, any new Large Load that hasn't secured its own generation or otherwise locked in supply will be first in line for curtailment during capacity shortages, ahead of PJM's existing pre-emergency load management steps. PJM is also building a mandatory registry to track these customers by location, service area, and generation status, and it plans to exclude uncommitted new large loads from the demand forecasts used in future capacity auctions. Affected customers would be compensated, similar to existing demand response programs, but compensation isn't the same as reliability.

The trigger was PJM's December capacity auction, where data center demand accounted for nearly 40% of total capacity costs, and where the forecast peak load for the 2027/2028 delivery year came in roughly 5,250 MW above the prior forecast, with about 5,100 MW of that increase attributable to data centers alone. PJM is also running a separate Reliability Backstop Procurement starting this September to cover an estimated 60 GW shortfall. Put simply: load grew faster than supply, someone has to absorb the risk when the system is tight, and PJM just decided it won't be everyone else on the grid.

What this PJM move means for data center offtakers

For the last two years, financing models, SLAs, and site valuations across the industry have treated grid interconnection as a firm asset. PJM's filing reclassifies it as conditional, at least for the segment of demand that shows up after mid-2027 without its own generation behind the meter. That reclassification has immediate underwriting consequences. Lenders and hyperscalers evaluating new PJM sites now need to treat curtailment exposure as a diligence line item alongside land, water, and fiber, not an afterthought. Existing sites with generation already secured are largely insulated; new sites betting on grid power alone are not.

This doesn't kill the PJM data center pipeline, but it does raise the cost and complexity of building there without a captive power source. Expect site selection to increasingly favor locations with either excess local generation capacity, direct access to co-located gas or fuel cells, or utility territories that haven't yet hit this kind of constraint, a dynamic LandGate has been tracking as subtransmission capacity, not transmission headlines, becomes the real gating factor in where data centers can site. It also raises the bar for what "shovel ready" means: a signed interconnection agreement is no longer sufficient proof of firm power, and buyers should expect that standard to spread beyond PJM as other ISOs watch how this plays out.

What this means for energy producers

The regulatory shift caps a year of capital moving toward behind-the-meter and co-located generation, and the sequence is worth laying out. In April, Oracle expanded its fuel cell commitment with Bloom Energy to as much as 2.8 GW, building on an initial system Bloom had installed in 55 days, well ahead of a 90-day target. On June 30, Brookfield expanded its financing framework with Bloom Energy from $5 billion to $25 billion, a fivefold increase, as part of Brookfield's broader $100 billion AI infrastructure fund. The next day, National Grid Ventures committed $1.75 billion for a 35% stake in Joulent to help fund Project Kilby, a 2.67 GW gas plant in West Texas built with Chevron that will deliver power to a Microsoft-operated data center under a 20-year PPA, without touching the public grid. Four weeks later, PJM's board made the case for all three deals retroactively.

Before PJM's announcement, these deals read as speed plays: a way to get power online faster than interconnection queues allow. After it, they read as insurance against a regulatory environment that now formally deprioritizes uncommitted grid-dependent load. That's a durable tailwind for producers building fast-to-deploy, co-located generation: natural gas peakers, fuel cells, and behind-the-meter renewables paired with storage. It also raises the value of sites where generation can be sited close to load, since proximity increasingly determines who gets treated as firm and who gets curtailed. Producers with permitted, buildable capacity near constrained grid nodes are sitting on an asset that just became more valuable, and the deal flow of the last four months is the market pricing that in real time.

The diligence question for both sides

The practical effect of PJM's move is that "firm megawatt" and "grid-connected megawatt" are no longer synonyms in this territory, and the gap between them is where the risk (and the opportunity) now lives. Offtakers need to know, at the parcel and substation level, whether a site's power is actually protected from curtailment or merely assumed to be. Producers need to know where load is concentrating and underpowered, since that's where the next co-located deal gets built. Both questions come down to the same underlying data: interconnection queue status, substation capacity, and generation proximity at the site level, which is exactly the kind of diligence LandGate's data center due diligence reports and power infrastructure data are built to answer before a term sheet gets signed, not after.

To learn more about LandGate’s tools and data for the data center and energy markets, book a demo with our dedicated energy & infrastructure team.