The Data Center Boom Is Happening in My County. Here's What That Actually Means.

From My Front Porch

I live in Locust Grove, Virginia. Twenty miles from my house, a 2,600-acre development called Wilderness Crossing is being planned. Of those 2,600 acres, 750 are slated for data centers and distribution warehouses. The full project would draw an estimated 1.5 gigawatts of electricity, roughly the entire output of the North Anna nuclear plant. The proposed site sits at the gateway to the Wilderness Battlefield, where Grant and Lee fought one of the most consequential battles of the Civil War, and contains at least 15 historical gold mines that used mercury for extraction in the 19th century.

This isn't a hypothetical. This is my neighborhood.

I've spent the last year watching the national conversation about data centers and AI play out from a vantage point most people don't have. I'm an Air Force veteran. I'm a former Cloud Engineer at Accenture Federal Services, where I spent years inside the federal infrastructure that some of these data centers exist to serve. I now run JDX Software, an SBA-certified Service-Disabled Veteran-Owned Small Business that does federal cloud work and small-business websites out of this same town.

So when people argue about data centers online, I'm hearing two completely opposite arguments and neither one matches what's actually happening.

One side says data centers are an existential AI catastrophe destroying communities. The other dismisses every concern as Luddite panic. Both are wrong, and the gap between them is where the real story lives.

A quick disclosure up front, because integrity matters more than appearing neutral. I have a financial stake in the federal IT industry continuing to grow. JDX Software is positioned to do this work. I am not a disinterested observer. I'm telling you that because the rest of this post is going to argue some things that look like they cut against my own interest, and I want you to know I noticed.

Here's what's actually going on.

First, Stop Calling Everything AI

The most common confusion in this entire debate is that "data center" now means "AI." It does not.

Data centers are decades-old infrastructure. Every email you send, every patient record your doctor pulls up, every streaming show, every banking transaction, every cloud-stored photo, this very blog post. All of it runs through data centers, and most of that workload has nothing to do with AI.

The numbers say the same thing. The US went from about 1,000 data centers in 2018 to more than 5,400 in 2025. Of that capacity, the International Energy Agency's 2024 estimate puts roughly 28% in enterprise facilities (companies running their own IT), 36% in colocation and service-provider facilities (multi-tenant), and 37% in hyperscale campuses (the giant Amazon, Microsoft, Google, Meta complexes). AI is genuinely the steepest growth driver right now, but even at the bleeding edge, it accounts for roughly 30% per year of new server demand growth, not 100%.

When someone tells you "AI is going to destroy the grid," they are conflating two different things. The grid problem is real. The framing is misleading. Even if every AI company shut down tomorrow, you would still have a data center industry, you would still have a grid that's been deteriorating for two decades under the cover of flat demand, and you would still have communities like mine asking who pays for the upgrades.

AI accelerated the conversation. It did not invent it.

The Local Pain Is Real

Let me be specific about what's actually happening in Virginia, because this is where the national debate gets concrete.

Virginia is the only state in the country where data centers already consume more than 20% of total electricity. The Electric Power Research Institute's 2026 projection has that climbing to between 39% and 57% by 2030. Loudoun County alone hosts more than 200 data centers. Northern Virginia has approved over 8,900 diesel-powered backup generators across the region.

These facts have consequences for ordinary people:

Average US electricity prices were flat at around 13 cents per kWh for a decade. By the end of 2025 they averaged 19 cents per kWh, about 27% higher than in 2019. In some Virginia areas near data centers, prices have climbed up to 267% over five years.
Consumer Reports documented residents in Manassas seeing electric bills jump from $100 to $281 in a single month.
In the PJM grid region, capacity market prices jumped from $30 to $270 per megawatt-day in December 2024 and now sit at $330. 63% of price increases in PJM in 2024 were driven by data centers.
A health impact study for a single data center campus in Loudoun County estimated that its permitted emissions could lead to up to $99 million in annual health-related damages across the region. Diesel backup generator emissions, including nitrogen oxides and fine particulate matter, are concentrated in environmental justice communities with lower household incomes.
Northern Virginia data centers collectively used nearly 2 billion gallons of water in 2023, a 63% jump from 2019.

This is the part of the debate that deserves anger. Real anger. People in my region are paying higher utility bills, breathing worse air, and watching their water tables get drawn down so that hyperscale operators can serve workloads owned by some of the most profitable companies in human history.

The Wilderness Crossing Problem

The Wilderness Crossing development is the case study, and I'm taking a position on it. I'm against it, at least at the proposed scale.

Wilderness Crossing would be the largest rezoning in Orange County's history. The site sits adjacent to the Wilderness Battlefield, designated by the National Park Service as a Class A historic preservation priority. The American Battlefield Trust and the Piedmont Environmental Council have both come out against the project. They are right to.

The reasons stack up:

1.5 GW of new electricity demand requires massive new transmission infrastructure crossing Virginia's landscape, costs that will eventually flow through to residential ratepayers across the state.
The 19th-century gold mining on the property used mercury for extraction. Soil disruption from data center construction risks dispersing that mercury into local air and water sources, including the Rapidan River, which the EPA has already deemed impaired.
The site is the literal gateway to a battlefield where over 28,000 American casualties occurred in May 1864. Once industrialized, the historic landscape cannot be restored.
Estimated water demand runs into the millions of gallons per day, in a region where local aquifers are already under strain.

I am not against data centers. I'm against this data center, in this place, at this scale. There is a real difference.

But Here's What Almost Nobody Says

Now the pivot, and this is the part I think gets missed in nearly every piece of coverage about Wilderness Crossing or any other data center fight.

The data center boom did not break the American power grid. It exposed how broken the grid already was.

US electricity demand was essentially flat from 2009 to 2023. Fourteen years of flat demand let utilities, regulators, and politicians coast on aging infrastructure. Every efficiency gain we booked masked the fact that we were not seriously investing in new generation, new transmission, or grid modernization. The assumption was that demand would not grow enough to matter.

Then the bills came due.

Bank of America's 2025 analysis found that 31% of US transmission equipment and 46% of distribution equipment is within five years of the end of its useful life or has already passed it. More than 70% of power transformers and transmission lines are at least 25 years old. The American Society of Civil Engineers downgraded the nation's energy infrastructure to a "D+" grade, calling it a $5 trillion problem.

The interconnection queue, the list of new power projects waiting to connect to the grid, sits at roughly 2,300 GW as of late 2024. That is nearly double the entire installed capacity of the US grid. Wait times for new transmission lines run four to eight years. In 2024, the country built only 322 miles of new high-voltage transmission. The Department of Energy estimates we need to build roughly 5,000 miles per year to keep up.

The strain was always coming. AI, electric vehicles, building electrification, manufacturing reshoring, every one of these trends was going to reverse two decades of flat demand on schedule. Data centers just made it arrive faster and more visibly than the others.

Here is the uncomfortable but accurate framing: for the first time in a generation, we have political and economic will to actually rebuild the grid. The pressure is real. The reform conversations are real. Bipartisan nuclear support, FERC reforms on transmission queues, state-level legislation on rate fairness, hyperscaler dollars flowing into clean generation. None of this was happening in 2019.

That is not an argument that data centers are good. It is an argument that the right response is to use this moment, not waste it.

What's Actually Being Built

Most coverage stops at "data centers are bad and the grid is overwhelmed." That's where my frustration with the discourse really kicks in, because there's a parallel universe of solutions getting built right now, and almost nobody is writing about them.

Nuclear Is Back, and Big Tech Is the Reason

This is the most under-covered story in the entire debate. The same hyperscalers driving the demand are now the most important private-sector force pushing nuclear power back into the US energy mix.

Microsoft signed a 20-year, $16 billion power purchase agreement to restart Three Mile Island Unit 1, rebranded as the Crane Clean Energy Center. Target online date: 2028. 835 MW of carbon-free baseload electricity dedicated to Microsoft data centers.
Amazon invested $500 million in X-energy in October 2024, backing 5 GW of future small modular reactor capacity. Separate SMR agreements with Energy Northwest in Washington (up to 960 MW) and Dominion Energy here in Virginia.
Google signed with Kairos Power for up to 500 MW of SMR capacity starting 2030.
Meta signed a 20-year agreement to buy 1.1 GW of nuclear from Constellation Energy's Clinton Clean Energy Center in Illinois.
Oracle announced plans for a gigawatt-scale data center powered by three SMRs.

Collectively, the major tech companies have signed over 10 GW of new nuclear capacity in the past year. The same four companies hold over 50 GW of renewable power purchase agreements. That is more clean electricity procurement than the entire generation capacity of Sweden.

This is not greenwashing. These are 20-year financial commitments backed by some of the most capital-rich balance sheets on Earth.

The Nordic Answer

The most credible scaled alternative to the standard hyperscale playbook is happening in Iceland, Norway, Sweden, and Finland.

Iceland's grid is essentially 100% geothermal and hydroelectric. Average temperatures allow free-air cooling year round. Power Usage Effectiveness numbers in the 1.05 to 1.15 range are typical, compared to a global average closer to 1.55. OpenAI's "Stargate Norway" plans to deploy 100,000 GPUs on hydroelectric power. Brookfield committed roughly $10 billion to a 750 MW AI campus in Strangnas, Sweden. TikTok is putting €1 billion into a Finnish facility. Meta has expanded its Odense, Denmark campus.

The most impressive single facility in the world is probably Lefdal Mine Datacenter in Norway: 1.3 million square feet inside a converted olivine mine, 60 meters below sea level, cooled by 8°C fjord water pumped through heat exchangers. PUE of 1.15 guaranteed, water usage effectiveness near zero, 100% hydro and wind powered.

The lesson is simple and not very glamorous. Build it where the climate and grid are already optimized for it.

What the Hype Gets Wrong

Equally important is what is not working.

Microsoft killed Project Natick, the famous underwater data center research project, in mid-2024. Almost nobody in the public conversation knows this. Noelle Walsh, head of Microsoft's Cloud Operations and Innovation division, told Data Center Dynamics directly: "I'm not building subsea data centers anywhere in the world." The fundamental problem was upgradability. Sealed underwater pressure vessels cannot be opened to swap out GPUs, and in an era when an H100 has roughly an 18-month useful life, you cannot deploy hardware you can't service for a decade.

Underwater data centers are functionally dead in the West. China is the only country running them at commercial scale (Beijing Highlander's deployment off Hainan), and even there the long-term economics are unproven.

Low-Earth-orbit data centers (Starcloud, Axiom Space, Lonestar) are real companies attracting real capital, but the physics is harder than it looks. Vacuum is a perfect insulator, which means radiative cooling in space is harder than air cooling on Earth, not easier. LEO data centers are a 15+ year bet for specific workloads, not a 2026 fix for Wilderness Crossing.

Better Cooling, Right Now

The most boring innovation is also the most immediately useful. Microsoft's zero-water cooling design becomes the default for new builds by 2027, saving over 125 million liters per facility per year. Closed-loop cooling cuts water consumption by 90%+ at the cost of slightly higher electricity use. Direct-to-chip and immersion cooling for AI workloads are scaling fast.

The point is not that any of this fixes the problem on its own. The point is that the choice is not "build it the old wasteful way or don't build it." There are real options, and we should be requiring the better ones.

The Policy Fix Is Real and Already Working

The rate-hike crisis is not a technology problem. It is a policy problem, which means it has policy solutions, which means we can actually do something about it.

Utilities have historically spread the cost of infrastructure investments across all customer classes. When a single hyperscale data center requires hundreds of millions of dollars in new transmission and generation to serve, spreading that cost across residential ratepayers is functionally a subsidy from your mother's electric bill to Microsoft's quarterly earnings. That is the actual problem. It is solvable.

States are already figuring this out:

Virginia created a new rate class for customers using 25 MW or more of electricity, pushing hyperscale costs onto hyperscale customers.
Oregon passed the POWER Act in June 2025, giving regulators tools to negotiate fairer deals with large-load customers.
Ohio approved a data center tariff in 2025 requiring data centers to pay their own infrastructure costs. AEP Ohio's data center pipeline immediately dropped from over 30 GW to 13 GW. Translation: most of that "demand" was speculative. When the speculators had to pay their own way, half of them went away.
FERC announced in April 2026 that it would act on reforms for integrating significant new electrical loads into transmission infrastructure.

If you live in Virginia and you are angry about your electric bill, your fight is not with data centers. Your fight is at the State Corporation Commission and with your legislators in Richmond. That is where the cost-allocation rules get written. That is where this gets fixed or gets worse. The fight on social media changes nothing.

What I Actually Think

Time to make my position explicit.

I'm pro-data-center. They are infrastructure, the same as roads and water mains. Killing them is not on the menu. Every Virginian who sends an email, files taxes online, or pulls up an electronic medical record uses one. Pretending otherwise is not serious.

I'm pro-federal-contracting and I want this work happening in Virginia. The federal cloud workloads I helped build during my time at Accenture have to live somewhere. Virginia has the talent, the cleared workforce, the proximity to federal customers, and the existing fiber infrastructure to be the place where they live. As an SBA-certified SDVOSB, JDX Software is part of that ecosystem.

I'm also pro-Wilderness Battlefield. The historic landscape matters. The Rapidan River matters. The 19th-century mercury contamination on that site matters. My neighbors who are watching their property values, their air quality, and their water tables get sacrificed to a project that benefits people who do not live here also matters.

These positions are not in tension if siting is done thoughtfully. They are in tension only if we treat data center expansion as a binary "yes everywhere" or "no anywhere." That binary is the actual enemy of good policy.

The smart fight is not "should we have data centers." It is "where, with what cooling, with what power, with what air quality protections, and who pays for the grid upgrades." Three things I want my state and county to do:

Pass cost-allocation reform on the Ohio model. Hyperscalers should pay for the grid infrastructure required to serve them. Period. No more invisible subsidies from residential ratepayers.
Require zero-water cooling for new builds in water-constrained regions. Microsoft is making it the default by 2027. Virginia can require it now.
Honor the existing historic preservation overlay around the Wilderness Battlefield, and reject Wilderness Crossing at its proposed scale. A scaled-back, properly-sited project is fine. The current proposal is not.

If we get those three things, I would not have written this post.

From My Front Porch, Again

I started this on my front porch. Let me end there.

The data center industry is going to keep growing whether anyone in Orange County wants it to or not. The American grid is going to get rebuilt over the next decade whether anyone in Orange County wants it to or not. The federal cloud workloads are going to land somewhere whether anyone in Orange County wants them to or not.

The only real question is whether the people who actually live here have a seat at the table when those things happen, or whether we get steamrolled because we let the conversation get polarized into "AI good" versus "data center bad" while the actual decisions got made by people who will never set foot near the Wilderness Battlefield.

If you live in Virginia, especially in the Fredericksburg or Northern Virginia corridor, this is a fight worth showing up for. Show up to your county planning commission meetings. Read the SCC dockets. Email your delegate and your state senator about cost-allocation reform. Support the American Battlefield Trust and the Piedmont Environmental Council. Demand zero-water cooling and proper siting on every new permit.

I'll be doing the same. From the front porch, and from this blog.

If you want to talk about federal IT, SDVOSB teaming, or building a website for your Virginia small business, you can find me at JDX Software. The federal cloud work lives on our Federal Capabilities page, and the small-business work in our region is on the Fredericksburg Web Developer page. For more on my background, see my author page.

Sources

Every claim in this post is sourced. Expand the list below to verify any of it for yourself.

Show all sources

Local and Wilderness Crossing

American Battlefield Trust, Wilderness Battlefield endangered sites listing
Piedmont Environmental Council, Wilderness Crossing project tracking
EPA impaired waters listings for the Rapidan River (no permanent URL, search EPA's How's My Waterway tool)

National data center growth and AI

Virginia and Northern Virginia specifics

Grid infrastructure

Nuclear and clean energy procurement

Nordic and alternative siting

Underwater data centers (status)

Heat reuse

Crusoe and Stargate

State policy reforms

Jackie Johnson-Dallas

Founder of JDX Software. Air Force veteran and former Accenture Federal Services cloud engineer.

Twitter GitHub Instagram