A power-line failure outside Washington, DC, became a stress test for one of the world's densest data-center markets this week.

TechCrunch reported that more than 3 gigawatts of data-center load stopped drawing power from the PJM grid within about 30 seconds after the fault, as facilities switched to backup systems. The sudden demand drop pushed excess electricity onto the system, with the report citing PJM data showing a peak surplus of 3.49 gigawatts and an 11-minute recovery period.

The incident did not cause a blackout, but it was visible beyond the immediate fault area. Ting Labs, whose consumer plug-in sensors monitor home electrical conditions, collected data showing a voltage spike from Northern Virginia to Chicago, according to the report. Northern Virginia is a major hub for cloud and AI infrastructure, so clustered behavior by large loads can matter far beyond a single campus.

PJM's own materials describe the grid operator as managing high-voltage reliability for more than 67 million people across all or parts of 13 states and the District of Columbia. That scale makes the event notable even without customer outages: a local line fault appears to have been amplified by automated data-center protection systems responding in sync.

The policy question is whether large computing loads should be required to ride through brief disturbances instead of dropping off the grid immediately. Texas grid operator ERCOT is already moving in that direction for large loads. For AI builders, the episode is another reminder that compute capacity is increasingly constrained by electrical behavior, not just chips, land and permits.