The biggest battery east of the Mississippi is being built to absorb AI's power appetite

In a former soybean field northeast of Columbus, a row of white battery cabinets is being wired into a substation that feeds one of the largest AI computing clusters in the country. By next June, the first phase of Eolian's Flint Grid project will push 200 megawatts into that substation for up to five hours at a stretch: enough stored energy, in a single discharge, to run tens of thousands of homes through an evening peak.

That single project is now the largest battery on PJM Interconnection, the multistate grid operator that manages power for much of the eastern U.S., and the biggest east of the Mississippi, with 1 gigawatt-hour of storage capacity when fully built. It is also the most concrete answer yet to a problem that has put both Republican and Democratic governors on the same side of a fight: AI data centers are pushing power prices up across the eastern United States, and the grid operator has not been able to add new supply fast enough to keep up.

The mechanism is the same one that has made batteries the fastest-growing piece of the U.S. grid. The system charges when power is cheap and abundant, often midday when solar production is at its peak, then discharges into the substation when demand is highest and prices are steepest. In California and Texas, that arbitrage pattern has already displaced new gas-fired "peaker" plants in the auctions grid operators use to keep capacity in reserve.

PJM is the test case. The operator has spent the last several years failing to bring new power plants online even as demand from data centers, electrification, and reshored manufacturing has climbed. That shortfall has already shown up in bills: a recent capacity auction cleared at prices several times higher than the prior round, drawing bipartisan criticism from statehouses in the region, and one the project is designed to interrupt.

The political pressure is no longer abstract. The White House is publicly pushing AI companies to "build, bring, or buy new power supply" rather than lean on the existing grid. The Intel semiconductor complex under construction at the north end of the same New Albany industrial bloc, a roughly $28 billion computer-chip factory, gives Flint Grid a neighbor that will draw heavy, sustained load from the same substation once it ramps.

PJM's structural problems, including a capacity market that until recently barely rewarded batteries at all, run deeper than any single project can fix. One 1-gigawatt-hour installation, large as it is, is still a small fraction of the new capacity the operator will need to keep pace with rising demand from data centers, electrification, and reshored manufacturing. The honest version of the bet being made is narrower: that batteries can now win the next marginal auction in capacity-constrained markets the way they already have in Texas and California, and that the political cost of doing nothing has gotten high enough to clear the way.

The first cabinets are scheduled to come online in June 2027.