Twenty-three billion is the cost-allocation default, not the data-center bill

In fourteen states, the cost of building grid capacity for a new data-center campus lands on the residential ratepayer, not on the company that triggered the build. That is a rule written by state utility commissions, not a law of physics, and the commissions can rewrite it whenever a rate case opens. Substations upgraded "for the grid" and transmission reinforced "for everyone" are socialized across every customer class; the hyperscaler's "fair share" pledge sits beside that ledger as a voluntary line item, not a binding allocation.

Monitoring Analytics, the PJM market monitor, attributes twenty-three billion dollars in cumulative customer price increases through the end of 2028 to expected data-center demand — all of it inside the 14-state PJM footprint, all of it the running total of that default. Read as a Big Tech problem, the figure is a casualty count. Read as a rate-case problem, it is a deadline.

The fight is happening this year. State commissions across the PJM footprint are writing the cost-allocation rule for the next round of data-center loads, with public comment windows open or imminent in 2026. A cost-causation study is the ask that flips the default from socialized to named — turning "the grid" back into a specific customer, with a specific bill. The mechanism repeats every time a new large load triggers a build; the only variable is whether the commission names the trigger or spreads it across the rate class. The hyperscalers don't pay when pledges are voluntary. The default is what they don't pay.