One Power Line, 3.49 GW Surplus: The PJM Wake-Up Call

ERCOT has announced plans to require large loads, including data centers, to ride through disruptions rather than instantly disconnect

Decision Focus

Last week, a power line failure outside Washington, DC triggered a synchronized mass-disconnect across the PJM grid. More than 3 gigawatts of data centers stopped drawing power within roughly 30 seconds, leaving the grid with a 3.49 GW surplus and driving voltage fluctuations from Northern Virginia to Chicago. Recovery took more than 10 minutes. The operational signal for global heads of data center energy is direct: coordinated mass disconnection by geographically concentrated loads is now a documented grid stability event — and at least one major RTO is already drafting a response.

90-Second Brief

Today, a power line fault caused Northern Virginia data centers to switch to backup power nearly simultaneously, removing approximately 3.1 GW from PJM’s grid within seconds. The resulting surplus peaked at 3.49 GW, roughly 3% of total PJM demand at the time, and required more than 10 minutes to stabilize. ERCOT has announced plans to require large loads, including data centers, to ride through disruptions rather than instantly disconnect. The PJM incident mirrors a similar event two years prior, indicating the problem is structural rather than incidental.

What Is Really Happening?

Northern Virginia is home to the highest concentration of data centers in the world, all sitting inside PJM’s control area. When a fault signal reaches these facilities simultaneously, their protection logic responds within seconds — decoupling thousands of megawatts from the grid in near-unison. The grid, designed to absorb gradual load shifts, cannot accommodate that speed of change without triggering voltage instability across a multi-state zone.

The scale dynamic compounds over time. Data centers represented roughly 6% of PJM’s electrical load as of 2024; that share is projected to reach 24% by 2040. Last week’s event involved approximately 3% of PJM’s instantaneous demand. As the base load share quadruples, the absolute magnitude of any synchronized disconnection scales with it — while the grid’s tolerance for simultaneous mass-switching events does not.

The root mechanism is architectural. Individual data center UPS systems are designed to protect IT equipment by instantly decoupling from the grid when a voltage anomaly appears. That logic is sound at the facility level. When dozens of facilities in a single dense zone apply identical logic at the same moment, each site’s protective response becomes a collective aggravating force at the system level..

Why It Matters for Global Heads of Data Center Energy

The immediate exposure is regulatory. ERCOT has announced it will require data centers and other large loads to ride through disruptions rather than disconnect. PJM has not yet formalized an equivalent rule, but the incident has drawn direct attention from grid operators. If PJM moves toward a comparable standard — which grid operators are reportedly reviewing — facilities in Northern Virginia face a compliance redesign of their power protection architecture. The lead time for the infrastructure required to meet such a standard is measured in years.

The capital implications are material. Ride-through capability at campus scale requires grid-forming battery systems or advanced power conversion equipment capable of absorbing fluctuations rather than rejecting them. One vendor, ON.Energy, reports it is currently installing 3 gigawatts of such systems across four data center campuses. That deployment scale suggests a segment of operators is already treating ride-through as an engineering specification, not a policy watch item — a gap in preparation timing that matters for procurement and budget cycles.

There is also a site selection implication that extends beyond compliance. Northern Virginia’s density creates a unique exposure: individual protection decisions aggregate into collective grid events. A new interconnection request that looks operationally sound in isolation may carry compounding compliance costs once ride-through requirements are formalized and apply to the full concentrated load profile in the zone. Operators modeling new capacity in the region need to factor in not just their own site’s requirements but their interaction with the broader Northern Virginia load cluster.

Forward View

Three fronts warrant active monitoring over the next 12 to 18 months.

On regulation, the ERCOT ride-through announcement signals that RTOs are treating this as a reliability mandate, not a voluntary best practice. The critical question is whether PJM formalizes equivalent language — and on what timeline and to what technical standard. Because infrastructure lead times for compliant systems run long, the compliance clock may effectively start before formal rulemaking completes.

On technology, the architectural response is not yet standardized. Grid-forming battery systems represent one path; coordinated sequential load reconnection protocols represent another. Grid operators have noted that orderly, sequential disconnection and reconnection — rather than simultaneous switching — could substantially reduce instability magnitude without requiring hardware changes. Whether RTOs will pursue protocol-based solutions before mandating hardware investment remains an open and consequential question for capital planning.

On interconnection economics, if ride-through becomes a connection condition rather than a performance incentive in PJM, it changes the cost baseline for large interconnection requests in the affected markets. Operators currently modeling Northern Virginia interconnection economics should treat ride-through compliance as an emerging line item in pro forma analysis.

What Is Still Uncertain

PJM has not announced a formal rule change. The trajectory toward a ride-through requirement in its territory is inferred from the incident record and the ERCOT precedent, not from any confirmed regulatory filing or PJM board action. The specific technical standard, compliance timeline, and enforcement structure for any future PJM requirement remain undefined. The geography of the four campuses where ON.Energy is deploying its systems is not specified in available reporting, so it is not confirmed whether that work is concentrated in PJM territory. Operators should treat the ERCOT signal as directional and the PJM risk as credible but unconfirmed pending formal regulatory communication.

One Question for Your Team

If PJM issues a ride-through requirement for large loads in Northern Virginia within the next 18 months, what is your current compliance position across each affected campus — and what is the realistic capital and procurement lead-time exposure to close the gap?