Summary: Maryland’s Office of People’s Counsel complained to FERC on May 7 that ratepayers owe about $2 billion of PJM transmission spending approved over three years, mostly for data center demand located elsewhere. Tom’s Hardware put the decade-long added cost at about $1.6 billion for Marylanders. The office asks FERC to assign costs to data center zones or large data center customers rather than spread them broadly.
Maryland’s Office of People’s Counsel (OPC) — the state body that speaks for utility customers — took its case to the Federal Energy Regulatory Commission (FERC) on May 7. At issue is the slice of PJM Interconnection’s transmission build-out that Marylanders must fund: of the $22 billion in competitive regional transmission projects PJM approved over the past three years, most of it to meet data center demand, roughly $2 billion is billed to Maryland ratepayers under current rules. Tom’s Hardware picked up the filing on May 9 and translated it into household terms: Marylanders face about $1.6 billion in added costs over the coming decade.
Nobody in this fight disputes that the grid needs work; the argument is over who pays. Virginia, Ohio, Pennsylvania, and Illinois are where data center load growth is concentrated. Maryland’s own forecast is far more modest, yet its residents are being invoiced for construction that serves demand somewhere else. For the first time, the question of who absorbs AI infrastructure costs sits formally on a federal regulator’s desk. And the implications stretch past the Mid-Atlantic: other regional grids copy PJM’s allocation approach in spirit, so the answer reached here will echo far beyond it.
Where Maryland’s $2 Billion Share Comes From
Start with scale: PJM runs the largest transmission operation in the United States, covering all or part of 13 states plus Washington, D.C. — about 65 million people, or one in five Americans. The ten-year breakdown OPC cites runs: $823 million charged to residential customers (about $345 apiece, just under $3 a month), $146 million to commercial customers (about $673 each), and $629 million to industrial customers (about $15,074 each). Most of the Marylanders affected are customers of two utilities, BGE and Pepco. Three dollars a month reads as pocket change; what it actually buys is a pointed question: why should Maryland residents be the ones subsidizing Virginia’s data center corridor?
The Flaw Inside PJM’s “Hybrid” Cost-Splitting Formula
The complaint takes aim at PJM’s “hybrid” method for dividing the cost of projects above 500kV: one half is spread across zones in proportion to how much power they draw, the other half is set by power-flow analysis. When load was climbing at a similar pace everywhere, that design held together. Data centers broke its premise. Growth is now packed into a handful of zones, and proportional spreading quietly recasts slow-growth states as subsidy donors. Maryland People’s Counsel David S. Lapp said it plainly: “PJM’s cost allocation rules are broken,” warning further that “without FERC action, Maryland customers face paying billions for transmission infrastructure” that serves data centers. The fix OPC is asking for is equally plain: pin the costs to the zones where the data centers actually sit — Dominion’s Virginia zone, for example — or bill the large data center customers themselves.
The Voluntary Pledge That Can’t Hold This Line
The complaint makes a point of naming the White House’s Ratepayer Protection Pledge of March 4, 2026, under which hyperscale data center developers committed to covering the grid costs they cause. The weakness is baked in: the pledge is voluntary, so an operator that skips it leaves the costs right where they were — with existing customers. The OPC raises a second problem: data-center load forecasts carry “extreme uncertainty” — once a transmission project is approved, the utility earns on the investment whether or not the projected demand ever arrives. Large-load tariffs and transmission security agreements at the state level can govern projects within a state, but regional cost allocation sits beyond their reach. That gap is precisely what carries this complaint to FERC — where a pledge has no enforcement teeth, a regulator has to supply them.
Three Takeaways for Compute Planning and Site Selection
Teams sizing AI capacity should take three signals from this case. One, transmission cost now belongs inside AI unit economics: an identical GPU-hour inherits different grid obligations depending on which PJM zone it occupies, and those obligations may soon be priced on their own. Two, pressure on the demand side keeps building — AMD’s Q1 report showed data center revenue up 57% year over year — at the same moment the grid-side backlash hardens: Tom’s Hardware counts 69 jurisdictions that have imposed data center moratoriums, and Virginia has been shaken by a 76% price-increase controversy. “Bring your own power” is no longer a nice-to-have; it is edging toward a precondition of siting. Three, FERC’s ruling will decide whether grid costs follow load or follow people — a more fundamental variable in compute pricing than any individual chip-supply story.