12 pages · tap a page to read its text
ISSUE 01
THE DATA CENTER BOOM
An investigation into the infrastructure race powering the AI economy.
ISSUE 01
Why we built a rigorous look at the infrastructure behind AI.
Projected demand versus what utilities can actually deliver.
Inside the county that became the center of America's data center economy.
How a data center campus actually gets built — land, power, water, concrete.
Ratepayers, utilities, and the fight over who funds new transmission.
A conversation on interconnection queues and reliability risk.
Five questions every community should ask before a data center breaks ground.
A guiding line for how this issue approaches the boom.
AS IT TURNS OUT — ISSUE 01 (REVISION 01) · 2
Editor's Letter
A boom is reshaping land, power and local economies faster than regulation can track.
omewhere between a rezoning hearing and a utility rate case, the data center boom stopped being a technology story and became a land-use story, a power-grid story, and in many towns, the only economic development story that matters. This issue was built to follow that shift as carefully as we could, county by county and megawatt by megawatt.
We began with a simple observation: the infrastructure decisions enabling the AI economy are being made at planning commission meetings and interconnection queues, largely outside public view, long before most residents learn a facility is coming. Our reporting tries to put those decisions back in frame, with attention to who benefits, who pays, and who gets a say.
That meant favoring documented records, public filings and on-the-record interviews over sweeping claims. Where figures are estimates or illustrative, we say so. Where a number is reported by an official source rather than independently verified, we attribute it plainly. Accountability, we think, starts with being precise about what is actually known.
What follows is not an argument for or against the buildout. It is an attempt to describe it accurately, from the substation to the tax ledger, so that the communities living beside it, and the readers deciding what to think of it, have something sturdier than headlines to stand on.
THE EDITORS
Illustrative estimates charting the widening distance between projected data center electricity demand and utility grid capacity through 2030.
3x
Estimated share of total U.S. electricity demand attributable to data centers by 2030, up from roughly 3% today, per illustrative industry modeling.
Approximate multiple by which projected data center load growth is estimated to outpace planned regional transmission capacity additions, illustrative figure.
5–7 yrs
Typical estimated interconnection queue wait reported by utilities for new large-load projects seeking substation capacity, illustrative range.
Estimated share of utilities' pending large-load requests tied to data centers in high-growth regions, illustrative modeling figure.
Target year by which several regional grid operators estimate current capacity plans will fall short of forecast demand, absent new investment.
A residential street backs up to data center campuses in eastern Loudoun County, where housing and server farms now sit side by side.
DATA CENTER ALLEY
In a single Virginia county, the industry that powers the AI economy has become the largest taxpayer, the newest neighbor, and the heaviest draw on the regional grid.
BY THE AS IT TURNS OUT RESEARCH DESK revenue, money that has gone toward schools, roads and other public services without, officials say, raising residential tax rates at the same pace as neighboring jurisdictions.
oudoun County did not set out to become the world's densest concentration of data centers. It became one anyway, as fiber routes, available land and proximity to Washington, D.C. drew server farms to what was once dairy and horse country. County officials have reported that data centers now generate a substantial share of local tax
How a hyperscale data center campus actually gets built — from site selection through permitting, power interconnection, water cooling infrastructure, and construction execution. The physical anatomy of the buildout that reshapes regional land use and grid demand.
BY THE AS IT TURNS OUT RESEARCH DESK
The stated purpose is to help assign certain costs to the customers whose needs cause them. [L07/LE07]
That is evidence of a regulatory mechanism. It is not evidence that a particular residential bill has already gone up, gone down, or been fully insulated from every effect of new demand.
Cost allocation is a mechanism, not a promise. The central reader concern is often short: Will I pay for this? The responsible answer is longer. In Virginia, the State Corporation Commission describes a framework for large-load customers that includes a separate rate class, minimum transmission and distribution cost obligations, collateral requirements in specified situations, and direct-assignment policy work for facilities that would not otherwise be needed without a large customer.
In its rate-case materials, the Commission and its staff have also treated forecast assumptions and the alignment of projected investment and revenue as matters requiring scrutiny. That is precisely why a reader should be cautious about a single before-and-after claim.
The bottleneck isn't steel or silicon anymore — it's the multi-year wait for a signed interconnection agreement.
As data centers multiply, utilities and consumer advocates are at odds over who should cover the cost of the transmission lines and substations needed to serve them.
BY THE EDITORIAL DESK very new data center campus eventually needs more than land and a lease; it needs power delivered at scale, and often that means new substations and transmission upgrades. The question of who pays for that infrastructure has become one of the most contested issues in utility regulation, according to filings in several states with heavy data center growth.
Regulators in a number of jurisdictions are now considering new tariff structures specifically for large, high-density customers like data centers, according to public commission dockets. Observers caution that outcomes vary significantly by state, utility structure, and the specific terms of interconnection agreements, and that no single national standard currently governs the allocation of these costs.
E Utility companies generally argue that large customers requesting significant new load should bear the costs tied directly to their service, through special tariffs or contribution-in-aid-of-construction charges. Representatives of several utilities have said in regulatory proceedings that this approach protects other ratepayers from subsidizing infrastructure built primarily to serve a single large customer.
Consumer advocates counter that utilities sometimes spread the cost of grid upgrades across their entire customer base through general rate cases, meaning households and small businesses could end up paying a share of infrastructure built to serve data centers. Advocacy groups in several states have asked regulators to require clearer accounting of which upgrades are driven by large new loads.
REQUESTED
UNDER REVIEW
An operator is evaluating the request. The review is not an approval.
A developer or customer has sought a connection or described a planned need. The request is not operating demand.
OBSERVED ENERGIZED
The system or record identifies actual operation. This is the status closest to an outcome, though it may still not answer every water, cost, employment, or emissions question.
A status can be further along than a study approval and still not describe observed operation.
WELL, LOOK AT THAT!
A reported large-load “project” may be a planning-pipeline state rather than operating electricity demand.
What this does not tell you: A status can be further along than a study approval and still not describe observed operation. The system or record identifies actual operation. This is the status closest to an outcome, though it may still not answer every water, cost, employment, or emissions question.
Five questions residents, journalists and local officials can put to data center developers and utilities before the next approval vote.
AS IT TURNS OUT RESEARCH DESK
1. What is the facility's projected water use, and where will that water come from?
Ask for disclosure of expected daily water withdrawal for cooling, the source — municipal supply, groundwater, or reclaimed water — and whether usage figures will be reported publicly on a recurring basis, according to utility planners interviewed for this issue.
2. What are the exact terms of any tax abatement or incentive package?
Request the duration, dollar value and performance conditions attached to any local tax abatement, and whether the agreement includes clawback provisions if promised jobs or investment levels are not met, as recommended by municipal finance officials.
3. Who pays for the grid upgrades this facility requires?
Ask whether new transmission lines or substation capacity triggered by the project will be billed to the developer, spread across the general ratepayer base, or shared under a cost-allocation formula, and request that formula in writing from the utility.
Infrastructure decisions made quietly today will shape electricity costs and land use for a generation.
AS IT TURNS OUT — ISSUE 01
As It Turns Out is a quarterly investigation into the systems that quietly shape modern life — the land, power, and policy decisions made far from public view but close to every bill we pay.
Next issue: beneath the servers. We turn to the water powering the boom, tracing how hyperscale cooling systems are reshaping regional water supplies, utility contracts, and the communities that depend on both.