Data Centers Caused 3.5 GW Grid Surge After Virginia Power Outage

The incident

A fallen power line near Washington, DC, this week triggered an unexpected cascade: more than 3 gigawatts of data center capacity vanished from the electrical grid in roughly 30 seconds. The sudden loss of demand sent voltage surging across the PJM Interconnection's territory, which stretches from Northern Virginia to Chicago and serves 67 million customers.

According to data from Ting Labs, an IoT sensor company monitoring electrical networks, the voltage spike caused lights to flicker across multiple states. While the grid avoided a blackout, it took more than 10 minutes to stabilize—far longer than the few seconds typically needed to recover from a downed line. At peak disruption, PJM's grid carried an extra 3.49 gigawatts of electricity with nowhere to go, as first reported by TechCrunch.

The disconnected facilities represented roughly 3% of total PJM demand at the time, Reuters noted. Northern Virginia hosts the world's densest concentration of data centers, making it particularly vulnerable to this type of cascading event.

Why it matters

This incident reveals a fundamental mismatch between how data centers protect themselves and what the grid needs to remain stable. As AI training and cloud computing drive exponential growth in data center construction—PJM expects these facilities to jump from 6% of its load in 2024 to 24% by 2040—the problem will intensify unless operators adopt grid-friendly designs. The alternative is more frequent voltage disruptions and potential blackouts affecting millions.

The technical problem

Electrical grids require near-perfect balance between supply and demand. When the power line failed, data centers detected the resulting voltage dip and switched to backup power within seconds of each other—a rational self-protection measure that became a collective threat.

"We need to figure a way for these loads that are located next to each other to sequentially either disconnect or reconnect," Ali Zain Banatwala, senior market models specialist at the Independent Electricity System Operator, told TechCrunch. Sequential disconnection would give grid operators time to respond rather than forcing them to absorb massive instantaneous swings.

Ricardo de Azevedo, CTO at ON.Energy, called the event "the canary in the coal mine," noting that similar incidents involving large loads "are happening more and more." A comparable event on the PJM grid in 2024 saw 60 data centers simultaneously drop 1.5 gigawatts—half the scale of this week's disruption.

Emerging solutions

ON.Energy has developed campus-scale uninterruptible power systems that place battery banks and power conversion equipment between data centers and the grid. Rather than disconnecting during voltage fluctuations, these systems can absorb excess power into batteries or dispatch stored energy when supply dips, responding within milliseconds.

The approach essentially presents the grid with a single, stable load instead of the erratic demand patterns from individual servers and cooling systems. De Azevedo said ON.Energy is currently installing 3 gigawatts of capacity across four data center campuses.

Grid operators are also responding. ERCOT, which manages Texas's grid, now requires large loads including data centers to "ride through" disruptions rather than immediately disconnecting, de Azevedo noted.

The details of this week's incident and the technical context were first reported by TechCrunch.