A single fallen power line drops more than 3 gigawatts of data-center demand off PJM's grid at once — exposing a failure mode the grid was not designed for
More than 3 GW of data-center load disappeared from the PJM interconnection simultaneously after a fallen power line, revealing a new category of infrastructure risk: AI compute is now concentrated enough that a single fault can swing grid demand by gigawatts in seconds.
Grids are engineered around loads that fail gracefully and independently. Data-center load does neither — protective equipment drops it fast and it drops together. A multi-gigawatt instantaneous swing is a frequency-stability event, and the grid operator did not choose to have that risk on its system; it arrived with the buildout.
The consequence for compute planning is that grid interconnection stops being a procurement detail and becomes a design constraint. With 190 GW of hyperscale capacity announced across 777 projects, the question of how that load behaves during a fault is no longer hypothetical. Expect interconnection agreements to start specifying ride-through behaviour the way generators have always had to.
The Neuron — Everything That Happened in AI Today → · Quartz — AI data centers pass 1 gigawatt and strain the U.S. power grid →