PJM Interconnection is moving toward mandatory ride-through requirements for large data center and cryptocurrency loads after 3.8 GW of demand – roughly the output of three large nuclear plants – disconnected from the grid in Northern Virginia on July 22, the largest single load-loss event in the RTO’s history. The episode exposed a reliability gap: hyperscale facilities currently trip offline during voltage disturbances that generators are already required to withstand, amplifying frequency swings and forcing other resources to compensate instantly. PJM’s response will set a precedent for every balancing authority facing surging data center demand.
Why Northern Virginia’s Data Center Cluster Creates Unique Grid Risk
Northern Virginia’s “Data Center Alley” hosts the world’s highest concentration of hyperscale facilities, drawing an estimated 3.5-4 GW of continuous load today with another 4-6 GW in active development queues. Unlike traditional industrial loads, these campuses combine extreme density – individual buildings often exceed 100 MW – with power supply designs that prioritize server uptime over grid stability. Most hyperscalers deploy uninterruptible power supply (UPS) systems backed by diesel generators sized for full-load carry, but their low-voltage ride-through settings are typically calibrated to protect IT equipment, not to support the bulk power system. When a transmission fault depresses voltage, thousands of UPS units can transfer to battery or generator power within cycles, shedding gigawatts of grid demand almost simultaneously.
The July 22 event followed a transmission line fault that caused a voltage sag across the Dominion Energy territory feeding Data Center Alley. Approximately 3.8 GW of load – about 7% of PJM’s peak summer demand – disconnected within seconds. Generators across the RTO accelerated as the sudden load loss created a frequency overshoot, triggering automatic generation control responses and, in some cases, over-frequency tripping of renewable resources. No customer outages resulted, but the disturbance violated NERC BAL-003 frequency response obligations and demonstrated that coordinated load tripping can destabilize the grid as severely as a large generator loss.
PJM’s existing tariff treats these facilities as firm load with no ride-through obligation. Generators, by contrast, must meet NERC PRC-024 voltage and frequency ride-through curves under FERC Order 901. The asymmetry means the largest new load class on the system operates under looser reliability rules than the resources serving it. PJM’s July 29 stakeholder presentation framed the problem explicitly: “Load that trips during disturbances increases the burden on remaining resources and undermines system reliability.”
Generator Ride-Through Rules Offer a Template – But Load Is Different
FERC Order 901, issued in October 2023, mandated that inverter-based resources (IBRs) – solar, wind, and battery storage – meet specific ride-through performance standards by 2026. That rulemaking took three years and produced detailed technical requirements for voltage and frequency tolerances, momentary cessation limits, and post-disturbance recovery. Applying a similar framework to data centers is conceptually straightforward but practically distinct in three ways.
First, data center loads are not electrically homogeneous. A campus may mix 480 V UPS-fed server racks, medium-voltage chiller plants, and on-site generation with different ride-through capabilities. A single standard must accommodate legacy facilities retrofitted with newer UPS firmware alongside greenfield designs. Second, the economic calculus differs: generator ride-through adds marginal capital cost (roughly 1-2% of project capex for IBRs) to avoid curtailment revenue loss. For data centers, ride-through may require UPS battery upsizing, flywheel additions, or generator control upgrades that could cost $50-150/kW of protected load – a meaningful fraction of the $8-12/W all-in cost of a new hyperscale campus. Third, enforcement jurisdiction is unsettled. NERC reliability standards apply to “users, owners, and operators of the bulk power system,” but large load customers have historically been regulated through interconnection agreements and tariff provisions, not directly through NERC standards. PJM may need a tariff change approved by FERC, a process that typically takes 12-18 months.
That points to a phased approach: an initial PJM tariff revision requiring new interconnection requests to demonstrate ride-through capability per a defined curve (likely aligned with PRC-024 voltage bands), followed by a NERC standard development project for existing loads. The latter would face stronger industry resistance and longer timelines – NERC’s Project 2020-02 on generator ride-through took four years from SAR to FERC approval.
Cross-Cutting Pressure: Capacity Market Signals and Resource Adequacy
The ride-through debate intersects with PJM’s capacity market reform. The RTO’s 2025/26 Base Residual Auction cleared at $269.92/MW-day – a tenfold increase from the prior year – driven by thermal retirements, load growth, and stricter capacity accreditation for intermittent resources. Data centers represent the largest source of load growth in PJM’s queue: over 30 GW of new demand interconnection requests, though many are speculative. If hyperscale facilities trip offline during disturbances, they effectively reduce the diversity benefit that load provides to the capacity market. A 100 MW data center that disconnects at 0.9 per-unit voltage contributes zero capacity value during the very events – hot summer afternoons with transmission congestion – when the system needs it most.
By comparison, a 100 MW battery with proper ride-through settings can inject power during the same voltage sag, earning capacity and ancillary service revenue. That asymmetry distorts investment signals. Developers already factor ride-through capability into storage and solar-plus-storage bids; extending the requirement to large loads levels the playing field. My rough estimate: if 20 GW of PJM’s projected data center load adopted generator-grade ride-through, the avoided frequency response procurement could save the RTO $50-100 million annually in regulation and frequency response reserves, based on current PJM ancillary service prices.
Cryptocurrency mining loads add another dimension. Unlike hyperscalers with multi-year power purchase agreements and ESG commitments, many miners operate on interruptible tariffs or behind-the-meter arrangements with minimal grid visibility. Their load can appear or disappear based on coin price and difficulty adjustments, making planning harder. PJM’s stakeholder deck noted that crypto facilities were among the loads that tripped on July 22. Requiring ride-through for new crypto interconnections – and potentially for existing ones above a threshold like 50 MW – would improve situational awareness for operators.
Who This Affects
- Utility transmission planner: Must model data center campuses as dynamic loads with defined ride-through curves rather than static P/Q injections, requiring updated load modeling tools and coordination with customer engineering teams.
- Storage and hybrid developer: Gains clearer revenue certainty for frequency response and regulation services if large loads no longer exacerbate frequency excursions; can bid more aggressively into ancillary service markets.
- State utility commissioner: Faces pressure to approve cost recovery for utility-side upgrades (dynamic VAR support, STATCOMs) needed to maintain voltage at data center interconnection points if customer-side ride-through is insufficient.
- Hyperscale data center operator: Confronts $10-30 million per campus in potential UPS and controls upgrades for existing facilities; new builds must bake ride-through into electrical design from day one, affecting site selection and power procurement timelines.
- FERC policy analyst: Will evaluate whether PJM’s tariff filing uses the least-cost reliability approach or imposes undue discrimination against large load customers; precedent will guide other RTOs (ERCOT, CAISO, MISO) facing similar load growth.
What to Watch Next
- PJM stakeholder process timeline: The Markets & Reliability Committee (MRC) and Members Committee (MC) votes on a tariff filing – likely by Q1 2025 if the current pace holds – will signal whether the RTO pursues a narrow new-interconnection rule or a broader retrofit mandate.
- NERC Standards Committee action: A Standards Authorization Request (SAR) for load ride-through, if filed by PJM or NERC staff, would launch a 3-4 year standard development cycle; watch for the SAR posting and technical conference announcements.
- Dominion Energy interconnection queue data: Track how many new data center requests in the Dominion zone withdraw or modify designs after ride-through requirements are formalized – an early indicator of compliance cost impact.
- FERC Order 901 compliance filings: Generator IBR owners must submit ride-through capability demonstrations by mid-2025; the technical details approved there will likely become the de facto benchmark for load-side requirements.
Bottom line: The July 22 event forced PJM to confront a structural mismatch: the grid’s fastest-growing load class operates without the ride-through discipline required of every generator connected to the same system. Closing that gap will cost hyperscalers real money but reduces systemic risk for everyone else – and the RTO that moves first sets the standard for the rest of North America.
Read the full report at Utility Dive
Note: facts and figures attributed above to Utility Dive reflect that outlet's original reporting. Broader context, cross-sector connections, and forward-looking scenarios reflect independent analysis by our editorial team.
About this article: Drafted by Energy Ai with AI-assisted research and writing based on public reporting, then reviewed under our editorial process before publication.
Leave a Reply