A failed power line near Washington, DC, briefly revealed how much strain clustered AI data centers can place on the electric grid. TechCrunch reports that more than 3 gigawatts of data center load disappeared almost simultaneously after the disruption, causing voltage spikes across the PJM grid from Northern Virginia to Chicago.
The event did not cause a blackout, but recovery took more than 10 minutes. PJM data cited by TechCrunch showed the vanished load peaked at 3.49 gigawatts, about 3% of demand on the grid at the time. In a power system that has to keep supply and demand closely balanced, that is enough to make lights flicker and force operators to stabilize the network.
Experts told TechCrunch that data centers need more orderly ways to disconnect and reconnect. One proposed fix is campus-scale battery and power-conversion equipment that lets a data center ride through voltage changes instead of abruptly switching away from the grid. The issue is becoming more urgent as AI training and inference clusters concentrate large, fast-moving electrical loads in a few regions.