FERC Approves PJM's 50MW Registry, Blocks Cutoffs [2026]

The Federal Energy Regulatory Commission handed the data center industry a split decision this week. On October 9, 2026, the agency approved PJM Interconnection’s plan to build a formal registry tracking large electricity users across the 13-state grid, but it struck down the companion rules that would have let PJM force those same facilities offline during power shortages. The ruling lands as data centers, driven largely by AI workloads, have become the single biggest new source of electricity demand on the PJM system, and it reshapes how the nation’s largest grid operator can manage that growth going into 2027.

For an industry that has spent much of 2026 bracing for mandatory power cuts tied to new data center construction, the PJM data center registry decision is a partial reprieve. PJM gets visibility into where big loads are going and how fast they plan to ramp up. But the harder question, who gets cut off first when the grid is stressed, goes back to the states and, PJM says, straight back to FERC’s docket.

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

FERC Approves the Registry, Rejects the Curtailment Plan

According to Utility Dive, FERC’s October 9 order approved PJM’s proposed large-load registry but rejected most of the companion framework PJM had built around it, known as the Interim Resource Adequacy Service, or IRAS. That framework would have let PJM curtail qualifying large loads, meaning data centers and other big industrial users, ahead of other emergency measures during a grid shortfall. FERC said no to that piece specifically, along with PJM’s related “bring your own new capacity” rules, collectively known as BYONC.

The registry itself survived intact. Per the fact record on this filing, it will capture a facility’s location, its ramp schedule (how quickly it plans to scale up electricity draw), its capacity supply arrangements, and other operational details. PJM’s proposal sets the threshold at large loads of at least 50 megawatts, roughly enough continuous demand to power tens of thousands of homes, a threshold confirmed by both PJM’s own regulatory filings and reporting from mGrid. Implementation of the registry was slated to begin October 12, 2026, the same day this story broke across financial wires.

That split outcome, registry yes, curtailment no, is the headline that Reuters ran with on wire services Monday: “US panel approves PJM’s registry of data centers, rejects plan to curtail power to them.” It’s a precise framing, though it understates how contested the curtailment fight has become behind the scenes at PJM and among the states inside its footprint.

Inside PJM’s New 50-Megawatt Large-Load Registry

PJM Interconnection runs the electric grid for all or part of 13 states and Washington, D.C., spanning a service territory of roughly 65 million people. It has watched its load forecasts climb repeatedly over the past three years as data center developers, chasing AI compute capacity, signed up for grid connections faster than PJM’s planners expected. The registry is PJM’s attempt to get ahead of that trend with better data rather than blunt-force rationing.

Under the approved rules, any new large load crossing the 50-megawatt threshold has to register with PJM and disclose its location, its planned ramp schedule, and details about what capacity supply it has lined up to meet its own demand. That last point matters: PJM has argued for over a year that a growing share of new data center interconnection requests arrive without a corresponding commitment of new generation to serve them, which is part of why the grid operator pushed so hard for the BYONC rules FERC just rejected.

The registry concept was first detailed publicly in a report from E&E News by POLITICO, which described PJM’s board proposing to track new customers with at least 50 megawatts of peak demand, including their location and ramp timelines, specifically so the grid operator could plan around the pace of the AI-driven buildout rather than reacting to it after the fact.

Why Regulators Killed the Interim Resource Adequacy Service

FERC’s objection to IRAS wasn’t about whether PJM needs tools to manage surging demand. It was about who has the legal authority to use those tools. In its order, FERC wrote that several provisions of PJM’s proposed tariff appeared to place obligations directly on retail customers, dictate which of those retail customers would be curtailed, and establish what amounted to a retail rate, three things that sit squarely inside state jurisdiction, not federal wholesale-market jurisdiction.

“We find that several provisions of the proposed tariff appear to place obligations directly on eligible new large loads, which are, for the most part, retail customers; that certain other provisions appear to dictate which retail customers taking retail service will be curtailed; and that the proposed compensation framework appears to establish a retail rate.”
Federal Energy Regulatory Commission, via Utility Dive

That’s a jurisdictional line FERC has drawn before, and it’s one of the thorniest fault lines in U.S. energy regulation: wholesale markets are federal turf, retail rates and retail curtailment priorities are state turf, and a plan that blurs the two tends to draw exactly this kind of rejection. PJM’s IRAS framework, in trying to get ahead of a capacity crunch by reaching past the wholesale market and into how individual retail customers get served, ran straight into that boundary.

FERC’s order also flagged a discrimination concern: singling out large loads, specifically data centers, for curtailment ahead of other classes of customers could be read as treating similarly situated retail customers unequally depending on their size or industry, another issue that cuts against how FERC has traditionally approached non-discrimination requirements in tariff design.

The “Bring Your Own New Capacity” Rule Also Gets Rejected

BYONC was PJM’s parallel attempt to solve the same underlying problem from a different angle. Instead of (or alongside) curtailment rights, BYONC would have required new large loads to either bring their own new generation capacity online or accept some form of curtailment exposure if they didn’t. It was PJM’s version of a “pay or play” rule aimed squarely at data center developers who want fast grid connections without necessarily financing the power plants needed to serve them.

FERC rejected BYONC alongside IRAS, for essentially the same jurisdictional reasons. The commission’s order treated both proposals as part of a single overreach: PJM, in trying to solve a supply-demand mismatch quickly, had drafted rules that reached past its wholesale-market lane and into retail-service territory that belongs to state utility commissions.

State Jurisdiction vs. Federal Authority

This tension, state control over retail electric service versus federal control over wholesale power markets, is not new, but the AI data center boom has pushed it to the front of the regulatory agenda faster than almost anyone expected two years ago. PJM’s 13-state footprint includes states with very different political appetites for data center growth, from Virginia’s “Data Center Alley” boosterism to growing pushback in states where residential ratepayers worry they’ll subsidize the grid upgrades data centers require. A one-size-fits-all curtailment rule written by PJM and blessed by FERC would have effectively pre-empted those state-level fights. FERC’s order keeps that fight at the state level, at least for now.

What Counts as a “Large Load”

The 50-megawatt threshold is a meaningful line in practice. It’s well below the size of the largest hyperscale AI training campuses now being planned, some of which are reportedly targeting demand in the multiple-gigawatt range over time, but it’s large enough to sweep in most standalone data center buildings and co-location campuses. That means the registry will likely capture a wide swath of the current data center development pipeline inside PJM’s territory, giving the grid operator its first systematic, mandatory view into where that pipeline actually sits.

Timeline: How the PJM Data Center Fight Reached FERC

The dispute didn’t appear overnight. PJM’s board had been signaling for most of 2026 that it needed new tools to manage data center-driven load growth, and the proposal went through several iterations before landing at FERC’s door in its final form.

Date	Event	Source
Late July 2026	PJM’s board signals plans for a large-load registry tracking new customers above 50 MW	E&E News by POLITICO
August 13, 2026	PJM formally proposes forcing data centers to secure their own power or face curtailment ahead of other customers	Reuters
August 17, 2026	Reporting details PJM’s push to curtail new 50 MW-plus data centers before other emergency load cuts	mGrid
September 23, 2026	PJM files its formal large-load interconnection tariff revisions with FERC under docket ER26-3515	PJM FERC filing
October 9, 2026	FERC approves the registry, rejects most of IRAS and all of BYONC	Utility Dive
October 12, 2026	PJM’s data center registry becomes available for implementation	Reuters, via TradingView

What stands out in that sequence is how fast PJM moved from “we’re thinking about a registry” in July to a fully litigated FERC order in October, barely ten weeks. That pace reflects how urgent PJM’s board considers the load-growth problem, and how much pressure the grid operator is under from both data center developers wanting fast interconnection and incumbent ratepayers worried about cost allocation.

Expert and Commissioner Reactions

The FERC order wasn’t unanimous in spirit even where commissioners agreed on the outcome. A concurring statement attributed to Commissioner Rosner, cited by Utility Dive, warned about what happens if PJM’s capacity market sends the wrong signal while large amounts of demand get shed involuntarily elsewhere in the system.

“The result could be a capacity market signaling that PJM has more than enough capacity while the region sheds thousands of megawatts of demand on an involuntary basis.”
Commissioner Rosner, FERC, via Utility Dive

Rosner’s statement also pointed toward where the responsibility should now land: with the states, not with PJM acting unilaterally.

“We need state regulators to establish clear rules for which customers should be curtailed first when circumstances require it. And we need PJM to be ready to adjust their operations to this evolving paradigm.”
Commissioner Rosner, FERC, via Utility Dive

PJM’s own board of managers, in filings tied to the proposal, made clear it isn’t trying to slow down data center development itself, only to manage how that growth interacts with grid reliability and existing ratepayers.

“The Board does not support approaches that seek to limit or eliminate this growth.”
PJM Board of Managers, FERC filing

Read together, those statements sketch the actual fight inside PJM’s territory: nobody with real regulatory authority wants to stop data center growth outright. The argument is entirely about mechanics, who decides when load gets cut, under what rules, and who bears the cost of making sure the lights stay on for everyone else.

How PJM’s Approach Compares to Other Grid Operators

PJM isn’t the only grid operator wrestling with data center load growth, but its structure makes this fight distinctive. PJM is a FERC-jurisdictional regional transmission organization, which is exactly why this dispute landed at FERC in the first place. ERCOT, the grid operator covering most of Texas, sits largely outside FERC’s wholesale jurisdiction because it doesn’t significantly cross state lines, giving Texas regulators and its own legislature more direct control over how large loads get treated during shortages. MISO and CAISO, like PJM, are FERC-jurisdictional RTOs and face similar structural constraints on how far their tariffs can reach into retail service.

That structural difference matters for data center site selection. A developer choosing between a PJM state, a MISO state, and Texas isn’t just comparing power prices and tax incentives anymore; they’re comparing how much regulatory uncertainty each grid carries around future curtailment rules. PJM’s registry, now that it has survived FERC review, gives data center developers at least a known set of disclosure obligations. The curtailment question, left unresolved by this order, remains the bigger unknown and will likely keep getting litigated state by state.

PJM Proposal Component	FERC Outcome (Oct. 9, 2026)	Who Governs Next
Large-load registry (50 MW+ disclosure)	Approved	PJM, effective Oct. 12, 2026
Interim Resource Adequacy Service (IRAS curtailment)	Rejected (most provisions)	Sent back toward state jurisdiction
Bring Your Own New Capacity (BYONC)	Rejected	Sent back toward state jurisdiction
Facility location & ramp-schedule disclosure	Approved as part of registry	PJM
Retail curtailment priority rules	Not authorized under this filing	State utility commissions

Historical Context: PJM’s Years-Long Capacity Crunch

PJM’s capacity market has been under strain for several years, well before this specific registry fight. Rising interconnection queues, slow retirements-to-replacement transitions for aging generation, and a historically long wait time to bring new power plants online all predate the current AI data center boom. What changed in 2025 and 2026 is the speed and scale of new demand arriving on top of an already tight supply picture. PJM’s board has pointed to data centers specifically as the fastest-growing category of new large-load interconnection requests it has ever processed, which is part of why the grid operator moved to build a dedicated tracking mechanism rather than relying on the slower, case-by-case interconnection studies it has traditionally used.

Ohio has been a particular flashpoint in this broader fight. Reporting from the Ohio Capital Journal characterized PJM’s broader plan as leaning heavily on individual states and their utilities to force data centers to secure their own power supply, or risk having that power cut during grid emergencies, a dynamic that previews exactly the kind of state-by-state patchwork FERC’s rejection of IRAS and BYONC now makes more likely rather than less.

Market Impact for Cloud Providers and Data Center Developers

For hyperscale cloud operators and AI infrastructure developers with projects inside PJM’s 13-state footprint, this ruling changes the compliance calculus without changing the underlying risk. The registry requirement is now a known, fixed cost of doing business: disclose your location, your ramp schedule, and your capacity arrangements once you cross 50 megawatts. That’s manageable paperwork compared to outright curtailment exposure.

What developers don’t get is certainty about the curtailment question. Because FERC pushed that decision back toward the states, a data center project in Virginia, Ohio, and Pennsylvania (all PJM states) could end up facing three different sets of rules about when and how it might get cut off during a grid emergency, rather than one uniform PJM-wide standard. That fragmentation could become a real factor in site selection for the next wave of AI data center investment, favoring states that move quickly to write clear, developer-friendly curtailment rules over states that leave the question open.

It’s also worth noting that PJM’s data center fight is unfolding alongside other data center cost and resource pressures already making headlines, from scrutiny over produced-water use in cooling systems, a subject TechRadar reported on recently, to the broader AI infrastructure buildout straining both power and water resources simultaneously.

Industry and Investor Reaction

Markets treated the ruling as a mixed but net-neutral outcome for data center developers rather than a clear win or loss. The registry adds disclosure burden without adding direct cost. The rejection of mandatory curtailment removes, for now, a worst-case scenario that data center operators and their lenders had been pricing into risk models for 2027 construction timelines. At the same time, the lack of a settled curtailment framework means that risk hasn’t disappeared, it’s just been deferred to a messier, slower, state-by-state process.

PJM, for its part, signaled through its board filings that it intends to keep pushing on the underlying problem rather than treat FERC’s rejection as the final word. Given the pace at which PJM moved from its initial July 2026 registry concept to a fully adjudicated FERC order by October, a revised curtailment proposal, redrafted to stay inside federal wholesale-market jurisdiction this time, looks more likely than not within the next filing cycle.

Predictions: Where the PJM Data Center Fight Goes Next

PJM refiles a narrower curtailment framework. Expect PJM to come back to FERC with a version of IRAS that avoids directly dictating retail curtailment order, likely routing that decision through state utility commissions instead of PJM’s own tariff.
States move unevenly on curtailment rules. Data-center-heavy states like Virginia will likely move faster and more favorably toward developers than states already skeptical of data center growth, creating the patchwork FERC’s order makes possible.
The registry becomes a model other RTOs copy. Because the registry survived FERC review relatively cleanly, expect MISO, SPP, and other FERC-jurisdictional grid operators to propose similar large-load disclosure registries rather than risk a jurisdictional fight over curtailment rights.
BYONC-style capacity requirements resurface in a different form. The underlying problem, data centers connecting faster than new generation comes online, doesn’t go away because BYONC was rejected. Expect PJM or individual states to attach new-generation requirements to interconnection agreements directly rather than through a PJM-wide tariff rule.
Litigation and appeals follow. Given how much is riding on this decision for both data center developers and incumbent ratepayers, expect at least one party, likely a state commission or a consumer advocate group, to seek rehearing or appeal portions of FERC’s October 9 order.

Why This Decision Matters Beyond PJM’s Footprint

PJM serves roughly a fifth of the U.S. population and sits at the center of the country’s data center corridor, anchored by Virginia’s enormous co-location cluster. How FERC resolves the tension between grid reliability and data center growth here will function as a template, intentionally or not, for every other FERC-jurisdictional grid operator facing the same AI-driven demand surge. The core finding, that curtailment and retail-rate decisions belong with states rather than with a regional grid operator acting alone, is a legal principle that will outlast this specific PJM docket.

It also underscores a broader truth about the current AI infrastructure boom: the bottleneck isn’t just chips, land, or capital anymore. It’s the regulatory architecture governing who gets electricity, when, and under what terms, and that architecture was largely built decades before anyone anticipated a single data center campus could draw as much power as a mid-sized city.

Frequently Asked Questions

What did FERC actually approve on October 9, 2026?
FERC approved PJM Interconnection’s proposed large-load registry, which requires data centers and other large electricity users of at least 50 megawatts to disclose their location, ramp schedule, capacity supply arrangements, and other operational details.

What did FERC reject in the same order?
FERC rejected most of PJM’s proposed Interim Resource Adequacy Service, which would have let PJM curtail large loads ahead of other emergency measures, and it rejected PJM’s “bring your own new capacity” rules in their entirety.

Why did FERC reject the curtailment plan specifically?
FERC found that the proposed tariff placed obligations directly on retail customers, dictated which retail customers would be curtailed, and effectively established a retail rate, all matters that fall under state jurisdiction rather than FERC’s wholesale-market authority.

Does this mean data centers in PJM’s territory can never be forced to go offline during a power shortage?
No. The order rejects PJM’s specific proposed framework for mandatory curtailment, not the general possibility of curtailment. Any future curtailment priority rules would likely need to come through state utility commissions rather than through PJM’s own tariff.

When does the PJM data center registry take effect?
PJM’s registry was reported as available for implementation beginning October 12, 2026.

What size data center has to register with PJM?
PJM’s registry proposal applies to large loads of at least 50 megawatts of demand.

How does PJM’s situation compare to Texas’s ERCOT grid?
ERCOT largely sits outside FERC’s wholesale jurisdiction because it doesn’t significantly cross state lines, which gives Texas regulators more direct control over large-load rules than PJM, a FERC-jurisdictional RTO spanning 13 states, currently has.

What happens next for PJM’s rejected curtailment plan?
PJM is expected to revise and refile elements of its proposal, likely in a form that routes curtailment-priority decisions through state regulators rather than PJM’s own tariff, in order to address FERC’s jurisdictional concerns.

FERC Approves PJM’s 50MW Data Center Registry [2026]