For most of the modern grid’s history, the reliability conversation has centered on supply. Operators planned reserves, engineered ride-through capabilities, and counted on generators to remain synchronized when the system wobbled, while treating demand as a passive quantity to be forecast and served. That assumption is now breaking down, and the clearest evidence arrived on July 22, 2026, when a routine transmission disturbance in the Dominion zone caused nearly 3.9 gigawatts of data-center load to abruptly disconnect from the grid and transfer to on-site generation almost instantaneously. It was the largest load loss of its kind in PJM’s history, and it crystallized a problem that planners had been warning about for two years: the fastest-growing category of electricity demand in the country behaves, during a fault, less like a factory and more like a tripping relay.
On Monday, November 16, 2026, PJM Interconnection responded. The nation’s largest grid operator filed two linked proposals at the Federal Energy Regulatory Commission that together amount to a major rethinking of how very large electricity users connect to and operate on the bulk power system. The first establishes mandatory voltage and frequency ride-through requirements for large computational loads. The second, branded the Fast Energization Process, reworks the interconnection queue so that gigawatt-scale load and the generation meant to serve it can be studied together and brought online in roughly half the time. Together, they try to balance two competing goals: serving an unprecedented wave of demand quickly, and preserving the stability of a system never designed for loads that can disappear by the gigawatt in moments.
A Routine Fault and an Unprecedented Load Loss
The July 22 event began with an ordinary occurrence. A fault on a transmission line produced a brief voltage sag of the sort the grid experiences regularly and is built to withstand. Generators rode through it. The surrounding network rode through it. But across Northern Virginia’s dense cluster of hyperscale facilities, the power-electronic equipment that conditions and converts incoming electricity reacted within milliseconds, sensing the dip and switching the load to uninterruptible power supplies and on-site generators. System load fell by several thousand megawatts almost at once, and PJM dispatchers spent the following minutes managing the resulting frequency and voltage excursions before the system settled. PJM’s Control Center managed the event safely, but the margin was thinner than operators would like, and the episode was not isolated. A smaller event in July 2024 had seen roughly 1,500 megawatts of data-center load drop in a cascade of successive faults, prompting a NERC review that first put the phenomenon on the industry’s radar.
The underlying physics is straightforward and, from a reliability standpoint, unforgiving. The switch-mode power supplies and rectifiers that feed modern computing equipment are engineered to protect sensitive silicon, and their default instinct when they detect an abnormal voltage is to disconnect. Multiplied across tens of thousands of servers inside a campus, and across multiple campuses sharing a transmission corridor, that instinct turns a localized disturbance into a system-wide imbalance: frequency spikes, voltages swing, and the generators still online must absorb the shock. The very feature that makes a data center resilient for its owner, the seamless failover to backup power, makes it a liability for everyone else sharing the network.
Teaching Large Loads to Ride Through
The remedy PJM proposes borrows from the rulebook that has long governed generators. For decades, power plants have been required to stay connected and operating through defined ranges of voltage and frequency rather than tripping at the first sign of trouble, because a resource that drops offline during a disturbance deepens the emergency it is reacting to. PJM’s filing extends that logic to demand by creating a new category it calls the Large Computational Load: a facility with aggregate demand of 50 megawatts or more, at least 25 megawatts of which is served through power-electronic conversion equipment. The definition is meant to capture data centers, cryptocurrency mining operations, and the artificial-intelligence training campuses now reshaping the interconnection landscape, and it tracks the thresholds that FERC and the North American Electric Reliability Corporation are developing in parallel.
Under the proposed data center ride-through requirements, these loads would be obligated to stay connected through a defined envelope of voltage sags and swings and frequency deviations, remaining energized through cleared transmission faults and restoring their pre-disturbance consumption within seconds of recovery rather than islanding themselves at the first dip. During deeper sags, facilities would be permitted to scale their power draw down in proportion to the voltage drop, a pragmatic concession that protects equipment without severing the connection entirely. The requirements would take effect in mid-January 2027 for newly interconnecting facilities as a condition of operation, while existing large computational loads would have one year from the effective date, or one year from commercial operation if later, to comply. PJM has been explicit that the standard it now seeks would have blunted the July 22 event had it been in force and followed, and the operator has said it will work with stakeholders on ramp-rate limits and other performance characteristics, with that work expected to begin in early 2027.
Compliance will not be trivial. Meeting a ride-through envelope will require testing and, in many cases, firmware and control-system changes across the uninterruptible power supplies, static transfer switches, and battery systems that sit between the grid and the server floor, which will require hyperscale operators and their equipment vendors to work together in ways few have had reason to until now.
Fast Energization and the Speed-to-Power Bargain
If the ride-through rules govern how large loads behave once connected, the companion Fast Energization Process addresses how they get connected in the first place. PJM’s interconnection pipeline has been strained by a volume of requests unlike anything in its history, with speculative proposals, duplicative studies, and a persistent resource-adequacy shortfall all competing for the same analytical capacity. PJM’s answer is to study large load and the generation intended to serve it in a single, consolidated process projected to take about one year rather than two. That would nearly double the pace of PJM large load interconnection while keeping the first-ready, first-served principle that keeps the queue orderly. The relief will not be immediate. According to the filing, the first cycle would not open until late November 2027 and would finish in 2029. Data-center groups have also warned that studying load and generation together could lengthen waits for projects that miss an annual cutoff.
The redesign also emphasizes commercial readiness. Readiness and financial requirements move earlier and grow more stringent, a deliberate filter meant to clear out projects that exist mainly on paper and to reserve scarce study resources for developers prepared to build, while an initial screening round gives those developers quick, actionable feedback before they sink years into the queue. New transmission service constructs, including interim service that can energize a load while permanent network upgrades are still under construction, offer a path to earlier operation. The process is also built to generate transparent data on which transmission upgrades are being driven by which loads, information state commissions have been demanding as they wrestle with the politically charged question of cost responsibility. PJM’s transmission owners filed a separate, coordinated set of tariff revisions the same day, covering the rates, cost allocation, financial protections, and revenue-crediting mechanisms that will determine how the expense of accommodating these loads is shared between the data centers themselves and the broader body of ratepayers.
A Coordinated Response Across the Markets
PJM did not act alone. The November 16 filing is the centerpiece of a coordinated response to show-cause orders that FERC issued across the organized markets earlier in the year, directing each operator to justify or reform the tariffs governing how it interconnects very large loads. The California Independent System Operator and other grid operators filed their own reform packages the same Monday, ISO New England filed the preceding Friday, and the Southwest Power Pool was slated to follow by the end of the week. The regional approaches differ in their particulars, reflecting divergent market structures and load profiles, but they share a common thread: each seeks to strengthen accountability for large loads and to protect existing customers from bearing costs and risks they did not create.
Running alongside the regional filings is a national standard-setting effort that gives the whole exercise its reliability backbone. In July 2026, FERC directed NERC to develop mandatory reliability standards governing the behavior of large computational loads, with an initial set of standards due by the end of December 2026 and a work plan for further standards due March 1, 2027. The minimum voltage and frequency ride-through thresholds are expected to come in that later work, in 2027. PJM has chosen to move ahead of that national timeline with its own regional standard, a decision that reflects both the urgency it perceives after July 22 and the reality that a continental standard, once finalized, will take time to propagate into enforceable obligations. The question of how PJM’s regionally tailored thresholds will ultimately align with NERC’s continental ones is one of the main open questions in the filings, and FERC will have to resolve it as the proceedings mature.
Who Pays, and What Comes Next
The reliability case for the ride-through rules is strong enough that the sharpest disputes are likely to form around cost and precedent rather than engineering. The broader data-center buildout has already provoked a pronounced public backlash rooted in the fear that existing utility customers will end up subsidizing the transmission and generation infrastructure built to serve hyperscale computing, and the cost-allocation tariffs filed by PJM’s transmission owners drop squarely into that debate. The grid reliability and data centers conversation has become inseparable from the affordability conversation, and the transparency provisions in the Fast Energization Process are, in part, an attempt to give regulators the granular evidence they need to assign costs defensibly rather than by rough approximation. How FERC threads that needle, requiring large loads to internalize the costs they impose while not erecting barriers so steep that economically valuable projects stall, will shape the economics of the sector for the remainder of the decade.
The procedural clock is now running. PJM has asked FERC for a 60-day comment period and an order within 120 days, which would put a decision in the first half of 2027. PJM wants its ride-through rules to take effect on January 16, 2027, before that order would arrive, and NERC’s work plan is due around the same time. Stakeholders from hyperscale operators to state consumer advocates and incumbent utilities will use the comment window to press their positions, and the filings are better understood as the opening of a negotiation than a finished settlement. The premise behind them is clear. Large computational load, which PJM already serves at roughly ten gigawatts and expects to grow by as much as thirty gigawatts by 2030, has made demand an active participant in system stability. In the emerging framework, the privilege of connecting at that scale comes with a duty to help hold the system together when it stumbles. That is the core of the FERC interconnection reform now unfolding: a new definition of what it means to be a responsible grid participant, written for an era in which a single customer can match a city’s demand and a nuclear unit’s sudden swings.
Conclusion
The filings PJM submitted on November 16 will be litigated and amended for months, and the ride-through thresholds may well shift before any order is final. But the shift in approach is unlikely to reverse. For the first time, the fastest-growing loads on the system are being asked to carry a share of the reliability responsibility that demand has historically been excused from. Paired with an interconnection process rebuilt for speed, the reforms point toward a grid that integrates gigawatt-scale demand and gives it obligations that match its influence. Whether the organized markets can implement that vision fast enough to keep pace with growing load will be answered in the FERC dockets opened this week.
From Vedeni Energy
Rules like PJM’s ride-through and Fast Energization proposals are decided not in headlines but in the committee rooms and stakeholder processes where the thresholds, tariffs, and compliance timelines are actually negotiated. Vedeni Energy represents clients virtually in stakeholder meetings across North American wholesale power markets, tracking these proceedings as they progress and ensuring our clients’ interests are on the record before deadlines close. If you would like to see where the decisions that affect you are headed, we offer a complimentary look-ahead brief on what is up for a decision at the committee that matters most to you. Message Patrick on LinkedIn or email info@vedeni.energy.