This week, PJM Interconnection owes the Federal Energy Regulatory Commission a piece of paperwork that sounds clerical and is anything but. Within 30 days of FERC’s October 9 order in Docket No. ER26-3515-000 — a deadline falling on Sunday, November 8, so effectively Monday, November 9 — PJM must file revised tariff language containing only the two elements of its large-load proposal that FERC accepted: the definition of a Large Load and the Large Load Registry. Everything else in the August 13 filing — the curtailment service, the bring-your-own-capacity rules, the capacity-market provisions — comes out.
The compliance filing will not add a megawatt of supply, cut a single bill, or curtail a single customer. It matters anyway, because what remains after the surgery is the instrument on which every future argument about large loads will depend. Before October 12, when the accepted provisions took effect, PJM’s visibility into its largest loads came largely through the annual load-forecast process, in which load-serving entities submit adjustments for loads they expect to arrive. The registry replaces that periodic picture with a standing, tariff-backed obligation to report.
The definition: 50 MW, one mile, every kind of load
The definition FERC accepted is short and deliberately broad. A Large Load is end-use customer load with a cumulative peak of 50 MW or greater at a single electrical site. Affiliated load facilities within a one-mile radius count as one site, an anti-gaming provision meant to stop an owner slipping under the threshold by building several sub-50 MW facilities close together. FERC declined to limit the definition to data centers, rejecting requests from FirstEnergy and others to carve out manufacturing and other industrial load. And the definition covers existing loads as well as future ones, regardless of in-service date.
That breadth is the point, and the first thing to understand about the registry’s politics. Industrial customers argued existing loads — already integrated into the grid and reflected in historical forecasts — should be excluded, at least where the registry served the rejected curtailment framework. PJM’s answer, which FERC adopted, is that the operator needs visibility into the full spectrum of large loads to manage its system and ensure reliability, regardless of whether any curtailment regime ever attaches to them. With that regime rejected, FERC added a qualifier: industrial customers will be included for informational purposes only.
What the registry actually collects
The registry lives in Schedule 11 of PJM’s Reliability Assurance Agreement. For each Large Load, the filing calls for the eligible customer and a description of the load; its location; the in-service date for each increment of load; any peak shaving adjustment value; the delivery point or point of interconnection; peak load quantity and ramp schedule; telemetry specifications; backup generation, including fuel type and permitting limitations; contracts or service agreements with its load-serving entity; points of contact at the transmission owner, electric distributor, and load-serving entity; and other details as necessary under the PJM Manuals.
Read that list as a planner would, and its logic becomes clear. Location and delivery point tell PJM where the load lands on the transmission system. In-service dates and ramp schedules tell PJM when it arrives and how fast it grows — the difference between a 50 MW load that appears overnight and one that ramps over four years is the difference between a reliability problem and a planning exercise. Backup generation and its fuel and permitting limits tell PJM what the load can do for itself in a pinch, and telemetry specifications tell PJM whether it can see the load in real time at all.
One field on the original list will not survive the compliance filing. FERC expects PJM to update the registry’s required information to remove items belonging to the rejected package, such as the amount of bring-your-own-new-capacity brought by a particular Large Load. The registry as filed was partly an administrative arm of the curtailment regime; as accepted, it is a planning instrument, and the compliance filing is where that distinction gets written into the tariff.
Who has to report, and who fills the gaps
The reporting obligation sits on the load-serving entity. PJM’s tariff language requires each entity bound by the Operating Agreement to submit, or cause to be submitted, the registry information for each Large Load it serves. That “or cause to be submitted” matters: the LSE is the obligated party, but it may not hold every data point itself. PJM acknowledged an LSE may not have all the information readily available, and the tariff contemplates that the electric distributor may provide it in those cases. PJM also proposes to obtain information from distributors to validate the registry’s contents or consult with relevant Large Loads.
FERC resolved the accuracy dispute in the LSE’s favor. Protesters objected that accuracy responsibility and compliance costs would fall on transmission owners and electric distributors; FERC disagreed, holding that the LSE bears the obligation and the distributor steps in only where the LSE lacks the information. Southern Maryland Electric Cooperative’s related requests — Manual provisions filed for review before the registry took effect, clarified distributor liability, and a cost-recovery mechanism — were not granted as conditions.
What remains unresolved is the machinery underneath the obligation. The detailed reporting mechanics were deferred to future PJM Manual revisions, which do not go through the same FERC review as tariff language. Equinix argued the registry lacks due-process protections — notice to affected customers, verification by those customers, a correction mechanism — and industrial customers asked for notice, correction, and appeal procedures. FERC declined to address the telemetry and accuracy protests tied to the rejected curtailment framework. With those obligations gone, the due-process questions move to the Manual process and PJM’s stakeholder committees, argued without a FERC order forcing the outcome.
Who sees it — and who does not
Confidentiality is where the registry’s design shows its sharpest edge. PJM maintains it under its existing obligations in Operating Agreement section 18.17, which FERC found adequate. Detailed data goes to Authorized Commissions — state public utility commissions and their associations — transmission owners, electric distributors, load-serving entities, PJM’s Independent Market Monitor, FERC, and other entities, subject to confidentiality safeguards. The public gets only aggregated data posted on PJM’s website.
That line excludes a group that fought hard to be on the inside of it. State consumer advocates — the offices representing residential customers in rate cases — argued they should have access because registry information will be central to their advocacy over cost allocation for exactly these loads. FERC sided with PJM’s existing rules, which do not grant consumer advocates access to this type of information, and accepted PJM’s argument that the registry may hold confidential, commercially sensitive, or infrastructure-security information, and so should reach “as small a universe of entities as possible.” Representative Frank Pallone argued the opposite — that as much as possible should be public, because it would be invaluable for improving demand forecasts. FERC’s compromise is aggregation: transparency about the total, secrecy about the parts.
The practical consequence is a two-tier information world. State commissions can see load-level detail and use it to police retail commitments and design retail rules. Consumer advocates, journalists, and the public will see totals — how much large-load capacity is coming, where in aggregate, and how fast — without knowing which customer sits behind which ramp schedule. PJM’s large-load page has pointed to a new registry tool planned for February 1, 2027, with aggregate public information following in March 2027. Until then, the registry operates largely out of public view, though its tariff provisions took effect October 12.
What the registry feeds — and what it does not do
The planning payoff runs through the load forecast. FERC called the registry valuable “separate and apart” from the rejected curtailment framework — not a consolation prize, but independently just and reasonable because an operator that cannot see its largest loads cannot plan for them. PJM’s annual forecast process has always depended on load-serving entities submitting large-load adjustments, with contract status distinguishing firm commitments from speculative requests. A tariff-backed registry with in-service dates, ramp schedules, and delivery points gives that process firmer inputs — inputs that matter most where the arguments are hottest, in the zones where large-load growth is fastest and where cost-allocation fights, like the backstop’s growth-based formula, now turn on measured growth rather than asserted growth. The registry will not settle those fights; it will narrow them, from arguments about whether a load exists to arguments about what its ramp schedule means.
The operational payoff is less abstract than it sounds. PJM told FERC there have been severe incidents of large loads tripping offline that could have led to significant outages had the operator not responded in time — the filing record cites a 1,500 MW drop from a Loudoun County data center tripping offline. A load that large disappearing in seconds is a contingency event; a load that large appearing without the operator knowing its ramp schedule is a planning failure waiting for a hot day.
But the registry’s limits deserve equal emphasis. It imposes no capacity obligation on any load, designates no new capacity, excludes no load from the capacity market’s demand curve, and authorizes no curtailment. Because FERC rejected the demand-curve exclusion, all forecast demand stays in the variable resource requirement curve for the December auction for the 2029/30 delivery year. The registry improves the forecast that feeds that curve; it does not change what the curve counts. And because the curtailment service was rejected without prejudice, registry data cannot be repurposed into curtailment obligations through a Manual revision; any such regime needs a new tariff filing and a new FERC order.
What to watch in the compliance filing itself
The filing due this week is, by design, a subtraction exercise. It must contain only the Large Load definition, with no definition of or references to New Large Load — the rejected category that would have captured loads arriving after June 1, 2027. It must carry the OATT and Reliability Assurance Agreement provisions for the registry with the curtailment and bring-your-own-capacity language removed, including rejected-package data fields like bring-your-own-capacity amounts. Because FERC rejected the proposed tariff sheets for including rejected language, the filing is effectively a clean rewrite of the surviving provisions rather than a redline of the original.
Watch whether the registry provisions are genuinely clean of cross-references to the rejected framework, whether PJM previews Manual-level reporting mechanics, and whether any party protests the filing itself.
What a replacement proposal would still have to answer
The compliance filing closes the tariff-cleanup chapter. It does not close the underlying question, and FERC’s order reads in places like a checklist for whatever PJM files next. A replacement would have to answer the jurisdictional objection first: obligations must run to wholesale entities, not directly to retail customers, and must not effectively set a retail rate or dictate which retail customers get curtailed. It would need a record showing why sorting loads by size, vintage, or capacity contribution is not unduly discriminatory, and a justification for any accreditation lock-in for resources bringing their own capacity. It would need to reconcile any curtailment design with Reliability Standard EOP-011-4, under which firm load may be involuntarily curtailed only after an Energy Emergency Alert Level 3 — not earlier, and not before voluntary demand response has been deployed. It would need to account for deliverability, so an offsetting resource is actually deliverable to the load’s zone, and for existing commercial arrangements, including FERC’s flagged edge case: a 49 MW load that grows by 1 MW and suddenly becomes a 50 MW load under a whole new regime. And it would need to explain why a load-serving entity should pay capacity charges for load the operator does not procure capacity for.
FERC also pointed to a different door. In the separate large-load show-cause proceeding, the Commission preliminarily found PJM’s tariff unjust and unreasonable for lacking transmission services for flexible large loads willing to limit their use of the system under certain conditions — specifically, a Non-Firm Contract Demand service. FERC suggested such a service might achieve some of what the rejected framework aimed at. The crucial difference is voluntariness: a non-firm service is a bargain a load chooses, in exchange for faster or cheaper connection, not an obligation imposed on it. PJM has told FERC it intends to make Section 205 filings in that proceeding by early to mid-November, including new transmission services for flexible large loads. If a voluntary flexibility product emerges there, the registry will be its plumbing.
That is the registry’s real significance, easy to miss beneath the louder fights it survived. The rejected framework tried to do three things at once — see the loads, price their impact, and control them in emergencies — and FERC accepted only the first. But seeing the loads is the precondition for the other two, under any design, voluntary or mandatory, state-led or federal. The states, the auctions, and the planners all need the same facts. The compliance filing due this week decides none of those questions. It establishes the record from which they will all be argued — and, for the first time, that record will be a tariff obligation rather than a forecast adjustment.