Largest U.S. Power Grid Operator Plans Emergency Capacity Auction, Costs to Be Passed On to Data Centers
Claire Weston
PJM, the largest U.S. grid operator, plans a September emergency auction to secure 6.8 GW of power capacity for 2028, with all costs passed directly to data centers — the AI electricity bill is shifting from 'who builds the plants' to 'who pays.'
How big is the power gap?
PJM's routine capacity auction earlier this month fell short of system reliability targets. The shortfall: roughly 6.8 GW — equivalent to seven conventional nuclear plants.
This means → existing grid planning can no longer cover 2028 demand. The gap is large enough to trigger an emergency backstop procedure.
PJM calls the emergency round a "Reliability Backstop Procurement," aiming to close the entire deficit in a single auction.
Why are data centers paying?
PJM's territory includes northern Virginia's "data center corridor," the densest cluster of data centers in the United States.
AI compute demand has surged in recent years. Nearly all incremental power load comes from data center expansion.
In plain terms = whoever created the gap pays to fill it. PJM is sending the bill directly to the fastest-growing source of electricity demand.
Where did this auction come from?
In January, President Trump and a bipartisan group of governors proposed that PJM hold a one-time auction allowing tech companies to bid on 15-year capacity contracts — effectively funding new power plant construction.
PJM has already faced criticism from multiple governors, utilities, consumer advocacy groups, and the White House. Calls for reform or even a breakup are growing.
This reflects a shift: grid governance is becoming a political question, not just a technical one. When the biggest electricity consumers are tech giants, cost allocation is a policy choice.
Can one auction really close the gap?
Whether 6.8 GW can be effectively secured through a single emergency auction remains the market's central variable.
Fifteen-year contracts mean winning bidders must lock in capital for plants that do not yet exist — long investment horizons and significant risk.
This means → if participation is thin or bids come in too high, only part of the gap gets filled. The remainder stays unresolved, keeping pressure on electricity prices and grid reliability.
Content is for reference only, not financial advice.