The Bill Arrives — One Household Inside Virginia's Data Center Expansion

Vicky Hu bought her place in Loudoun Valley Estates about twenty-five years ago. She bought it for the trees. The lot runs 1.34 acres near Broad Run in Ashburn, Virginia, which, if you haven't been keeping up, is the county where something like seventy percent of the world's internet traffic passes through on a given day, humming away inside windowless concrete buildings the size of aircraft hangars.

On June 13, 2025, Dominion Energy called Hu at home to tell her a proposed transmission project might cross her property. Under the route Virginia's State Corporation Commission would later approve as its fallback, a tower roughly 185 feet tall would stand about 126 feet from her house.

The SCC is Virginia's utility regulator, the body that decides whether power lines get built and where they go. The project, called Golden-Mars, is the last segment of a three-part transmission loop meant to bring bulk electricity into Loudoun's data center corridor. Loudoun County itself, in a May 2026 filing, argued the project was driven "almost entirely by private data center demand." Hu put it plainer than that: the company told her the line was for data centers, then went to the Commission and called it a public necessity, which is the legal standard a utility has to satisfy before it can run steel through your yard.

That phone call started her on a year-long education in how Virginia governs its own power system. What she found was an arrangement spread across so many institutions that not one of them could tell her who was responsible for what was about to happen to her property.

Whose decision is it

The docket number is PUR-2025-00056. Two local hearings in September 2025, a telephone hearing in December, more than 600 registered speakers, hundreds of pages of written comment. The Commission characterized the public response as almost universally favoring underground construction and raising concerns about property values, landscape, and health.

On April 9, 2026, the SCC found the project necessary and approved overhead Route 4. Route 4, though, required the Loudoun County School Board to authorize use of school property. If the School Board didn't act by July 20, the approval reverted to Route 3A, the one that puts a tower 126 feet from Hu's house. The Commission said in its own order that Route 3A was, by almost any objective measure, inferior to Route 4 and to another alternative. It approved 3A as the fallback anyway.

So the fate of a homeowner's property passed from a utility regulator to a school board. On June 30, the Loudoun Board of Supervisors and Planning Commission jointly urged the School Board to authorize a Route 4 variant so that 3A wouldn't become final. On July 13, Hu appeared at a special School Board meeting, calling the dispute "thousands of homeowners' fight" and warning that the precedent would reach past her property line.

The School Board voted against the Dominion project.

Look at that sequence. The SCC decides the line is necessary. Dominion picks the routes. The School Board controls access to school land. The Board of Supervisors can write letters. Hu watches from 126 feet away, and she is the only party in the whole arrangement who has to live with every one of those decisions at once. Each institution has authority over one part of the sequence, and none of them is responsible for the outcome.

What $879 million buys and doesn't buy

Loudoun County's FY2027 fiscal projections put data center computer-equipment tax revenue at roughly $879 million, about 77 percent of projected personal-property tax revenue once you set aside state reimbursements. The county's FY2026 budget presentation figured data center revenue was worth 66 cents on the general real-property tax rate, and said that without it the rate would likely run above $1 per $100 of assessed value.

Which means the same county that told the SCC this transmission project was driven "almost entirely by private data center demand" depends on that demand for better than three-quarters of its personal-property tax base. The county keeps a Revenue Stabilization Fund targeting 10 percent of data center revenue, so somebody in the budget office is lying awake thinking about what happens if the industry moves on or the tax structure changes. But no county document in the public record sets that revenue against what the growth costs: roads, water and sewer, schools, emergency services, planning, permitting, legal work, or the damage to property borne by residents who happen to live along a transmission corridor.

Virginia also imposed a temporary statewide data center electricity-consumption tax of $0.011 per kilowatt-hour, running July 1, 2026 through June 30, 2028, with collections above $600 million refunded pro rata. That's a state tax. None of it flows into Loudoun's budget. The county's dependency rests on equipment taxes it levies itself, and those depend on the industry continuing to build in Loudoun.

JLARC, Virginia's legislative audit agency, reported in 2024 that staff in five major data center localities described the facilities as imposing relatively low direct costs on local government. Few employees, no school enrollment the way a subdivision brings school enrollment. That finding is about what the county government itself spends. It says nothing about electric transmission, utility upgrades recovered through customer rates, or what a 185-foot tower does to the value of a house somebody bought for the trees.

The line that may cross Hu's property doesn't show up anywhere in Loudoun's budget as a data center cost. It's carried by a homeowner who also enjoys the lower tax rate made possible by the industry that put the tower there.

The order that hasn't become a bill yet

Hu has been a Dominion residential customer for a quarter century. The transmission line is one way data center growth reaches her. Her electric bill is the other. After a year of reading SCC dockets and county filings, what she's found is that the two exposures, property and rates, travel through separate institutional processes on separate clocks with separate decision-makers, and land on the same household.

On August 5, 2026, the Virginia Mercury reported that the SCC had ordered Dominion to develop a tariff assigning more transmission costs directly to data centers and other large-load customers. The docket is PUR-2026-00056, Dominion's annual application covering the regional transmission-service charges it passes through to customers.

The operative word is develop. What the SCC ordered was a tariff-development exercise. Environmental groups intervening in the case supported what the trade calls a "but-for" analysis: if the transmission facility wouldn't have been needed but for the data center, the data center pays for it. Dominion and the data center interests disputed either the timing or the mandatory character of that approach.

The full text of the August order wasn't publicly retrievable as of this writing, and several questions stay open. When must Dominion file the new tariff. How does the order define a "dedicated" or "directly attributable" facility. Does it reach existing large loads or only new ones. And do the regional backbone projects triggered by data center growth in the aggregate stay allocated the ordinary way, spread across all customer classes. The governor's office has said the change could save other customers hundreds of millions of dollars. That's an executive-branch estimate, not a residential-bill methodology anybody can check against the order.

Separately, in the 2025 biennial review, the periodic proceeding where the Commission examines a utility's earnings and rates, the SCC created a new rate class called GS-5 for customers drawing at least 25 megawatts. That means data centers. It takes effect January 1, 2027, and carries minimum billing obligations of 85 percent for transmission and distribution demand, so a data center pays for most of the capacity it reserves whether or not it uses it. When Dominion recalculated its transmission charges with GS-5 folded in, the projected monthly residential increase dropped from $2.90 to about $0.94. That $0.94 reflects the current allocation method. It isn't savings from a tariff nobody has written yet.

I've spent a fair amount of ink on the distance between a fact about a document and a fact about a room. An SCC order directing Dominion to develop a tariff is a fact about a regulatory docket. A lower residential bill is a fact about a household. Between them sit tariff filings, cost-of-service studies, hearings, appeals, and however many years all of that consumes. Hu can read the docket. She can't yet read the bill it's supposed to produce.

What a residential bill won't tell you

The SCC publishes a standardized bill comparison for a residential customer using 1,000 kilowatt-hours a month. That bill was $116.18 in May 2020 and $149.92 in July 2025, a 29 percent increase, $33.74 in nominal dollars. An interim fuel factor effective July 2025 added $8.95 to the standard monthly bill, more than the $8.75 year-over-year increase, which means the immediate change came predominantly out of fuel costs, not any identified data-center charge.

For Hu, that $33.74 is one line in a longer arithmetic. She's also weighing what a 185-foot tower does to the value of a property she's held for twenty-five years, in a county whose low tax rate depends on the industry that put the tower there. The rate increase and the property exposure come out of the same demand. No public document connects them, because they move through institutions that don't share any books.

A year of reading SCC filings teaches you this much: no Dominion residential customer can determine, from any public document, how much of that $33.74 goes toward serving data center load. The published bill combines base rates, fuel, riders, transmission, distribution, and other adjustments, with no data-center subtotal anywhere in it.

JLARC's independent cost-of-service model found Dominion's existing class allocations broadly in line with its own. Residential customers were assigned 55 percent of transmission costs, against 53 percent in the independent model. JLARC didn't conclude that residential rates were already being materially overcharged for the system as it stands. It concluded that existing rate structures didn't adequately protect ordinary customers against the scale and risk of the expansion still to come.

The forward numbers are less comfortable. JLARC estimated data center growth could add roughly $14 a month to a residential bill by 2040 under a half-growth scenario, and $33 to $37 a month under unconstrained growth, in constant 2024 dollars. For a household already absorbing a 29 percent increase across five years while facing possible loss of property value, those projections describe a direction of travel, and it only runs the one way.

An intervenor's expert witness identified 203 planned Dominion transmission projects with a combined estimated cost near $7.6 billion, classifying about $2.4 billion as primarily driven by data centers and another $3.3 billion as partially driven. Those are one expert's classifications, not Commission findings, and they don't convert cleanly into a residential-rate share. What they describe is the pipeline of steel and easements that will eventually turn up in somebody's bill, or somebody's backyard.

The pipeline

The question Hu can't answer from the public record is whether Golden-Mars is the last line through her area or the first of several. That depends on forecasts that don't agree with each other.

Dominion's own documentation submitted through PJM, the regional operator that coordinates the mid-Atlantic grid, projects data center billing demand rising from about 4.1 gigawatts in 2025 to 10.7 GW by 2036 and 17.6 GW by 2046. PJM's 2026 forecast has Dominion-zone summer peak growth at 5.4 percent a year over ten years, still the fastest absolute growth in the region but under the previous forecast's 6.3 percent. PJM revised down after tightening how it counts large-load projects, separating firm commitments from speculative requests. JLARC modeled both an unconstrained case, where statewide demand could more than triple by 2040, and a half-growth case reflecting limits on how much proposed development could actually be served.

The forecasts diverge partly because they measure different things, billing demand versus coincident peak versus total system load, and partly because nobody knows how much of the contractual pipeline is real. Dominion reported roughly 47 gigawatts of large-load contractual requests as of July 2025. Its own 2046 forecast sits far below that. A request queue is a line of people who would like electricity, not a prediction of how much they'll get. Every gigawatt that does materialize needs transmission, though, and transmission runs through somebody's neighborhood.

Even the lower forecasts have demand roughly doubling in a decade. The infrastructure to serve it is already drawn up: 203 projects, $7.6 billion. Golden-Mars is one. The next could be elsewhere in Loudoun, or Prince William, or out on the Coastal Plain. Whoever lives there will find out when the phone rings.

126 feet

Hu spent more than a year working through SCC dockets, county filings, and School Board agendas. The public record she found is extensive. She can read the Commission's reasoning. She can see the county's revenue figures. She can track the governor's claims about ratepayer savings.

What she can't find is a number telling her how much of her electric bill pays for data center growth, or an accounting that would tell her county what that growth actually costs once you include the infrastructure that never enters the county budget.

The institutions have produced orders, rate classes, and directives. GS-5 takes effect in January. The tariff-development exercise proceeds on whatever timeline the Commission sets. These are real actions and they may eventually move costs around.

Meanwhile Vicky Hu lives in a house she bought for the trees, 126 feet from where a tower may go up to feed buildings that employ almost nobody in her community and account for three-quarters of her county's personal-property tax revenue. She has spent a year putting the same question to every institution that touches her situation, and what she's learned is that the question doesn't fit on any of their forms.