The Same Pattern, Different Resource: West Virginia's Data Center Boom and the Coal Playbook¶

There is a pattern in West Virginia's economic history that has held for a hundred and fifty years, and it is in the process of repeating. The actors and the resource have changed; the structure has not.

The coal industry was built on four legs: outside capital, marginalized local voice, natural resource extraction, and an equity-of-opportunity gap that left native West Virginians on the wrong side of the wealth their land produced. House Bill 2014 — signed by Governor Patrick Morrisey in late April 2025 and effective July 11, 2025 (WV Legislature, HB 2014 Enrolled; Observer WV) — set those same four legs under West Virginia's emerging artificial intelligence and data center industry. The resource being extracted has changed from anthracite to electricity, water, land, and grid capacity. The pattern has not.

This post is Big Nerd Idea's reading of that pattern. It is an editorial position, not a research paper — but the structural claims below are sourced, and the speculative ones are labeled [HYPOTHESIS] per our standing Epistemic Honesty directive.

The four parallels, plainly stated¶

1. Outside capital, dominant¶

Absentee ownership of West Virginia land is not a metaphor. It is the documented baseline of the state's economy.

By 1810, as much as 93 percent of present-day West Virginia was held by absentee owners (Appalachian Voices, "Owning the Mountains").
A 1974 study found two-thirds of private land in the state was owned or controlled by absentee landlords, and that in nearly half of West Virginia counties, at least half the land was owned by out-of-state corporate interests (Appalachian Voices).
Pocahontas Land Corporation — a Norfolk Southern subsidiary, corporate descendant of the Flat-top Coal Land Association formed by the president of the Norfolk & Western Railway (e-WV: Pocahontas Land Corporation) — has controlled over a million acres of natural-resource property across Alabama, Illinois, Kentucky, Tennessee, Virginia, and West Virginia. In Wyoming County, WV alone it owns roughly 25 percent of all privately-owned land, assessed at an average of less than $350 per acre in 2016 (Appalachian Voices).

The new actors are different in name only. Combined 2026 capital expenditure for the four largest hyperscalers — Microsoft, Google, Amazon, and Meta — is projected at $600–700 billion, with Microsoft's FY2026 capex guidance alone at $80 billion, roughly double FY2023 (Next Waves Insight). None of those companies is headquartered in West Virginia. Their interest in the state is, as the Daily Yonder put it, the data center rush in Appalachia — coal country re-cast as energy country, with West Virginia hollows and former strip mines suddenly attractive for the same reason they were attractive in 1880: cheap land, dispersed political opposition, and access to a natural resource that the coast badly needs.

2. Marginalized local voice — preempted by statute¶

This is the parallel where the historical record and the current statute book line up most directly.

HB 2014 created two new entity types in West Virginia law: Certified High Impact Data Centers and Certified Microgrid Districts. The bill is explicit: these districts and facilities

"may not be subject to county or municipal zoning, horticultural, noise, viewshed, lighting, development, or land use ordinances, restrictions, limitations, or approvals."

The bill further prohibits counties and municipalities from enacting or enforcing ordinances, regulations, or rules that would "prohibit or hinder" these districts (WV Gazette-Mail; Intelligencer / News and Sentinel). The associated microgrids, when more than 70 percent of generation is consumed by data centers in the district, are also exempted from the state Public Service Commission's jurisdiction over rates, certificates of convenience, service conditions, complaints, and net metering / interconnection standards (Observer WV).

Read that as a sentence: the legislature has prohibited counties, towns, and the rate-setting commission from exercising the authority that protects local communities from industrial siting decisions made over their heads. That is not an incidental feature of the bill. It is the bill.

The historical analogue is not subtle. Coal severance taxes were captured weakly. Mine safety enforcement was deferred to federal authority because state mechanisms were too captured to act. Mountaintop removal proceeded under permits that, in practical effect, the people living downstream could not block. The HB 2014 structure modernizes that pattern: not a refusal to regulate, but an explicit statutory carve-out from the local instruments that would otherwise let communities push back.

3. Natural resource extraction — water, electricity, grid capacity¶

The resource has changed. The extractive mechanic has not.

A medium-sized data center consumes around 110 million gallons of water annually for cooling; large facilities can consume up to 5 million gallons per day (Consumer Reports; Climate Solutions Law).
Northern Virginia's data center cluster — the model the WV statute is partially designed to recruit business from — consumed nearly 2 billion gallons of water in 2023, a 63% increase from 2019 (Consumer Reports).
Large around-the-clock data center electricity loads are projected to drive higher greenhouse gas emissions, local air pollution, and higher energy bills for residents, businesses, and other industries — unless energy planning specifically supports new clean generation (World Resources Institute).
HB 2014 explicitly authorizes microgrid powering "from both fossil fuels and renewable energy generation" (Observer WV). In a state whose existing baseload is largely coal-fired, "fossil fuels" is not an abstraction — it is the lit fuse of a feedback loop in which AI infrastructure built in West Virginia becomes a new reason to keep burning the same resource the state was supposed to be transitioning away from. [HYPOTHESIS] — the long-term carbon path of WV-sited AI compute is unknown until specific projects file specific plans, but the structural incentive is plain.

4. Equity-of-opportunity gap for native West Virginians¶

Coal-era West Virginia produced enormous wealth. Most of it left the state in the form of mineral royalties, transport revenue, processing margin, and refined-product markup. Locally, what stayed was the workforce — and the workforce's exposure to the worst externalities of the extraction.

The AI data center economy is structurally even more lopsided in this regard. Coal at least required hundreds of thousands of local workers to extract. A 250-megawatt data center, once built, typically employs [HYPOTHESIS] 50–150 operations staff — and the high-paid roles among those (site reliability engineering, hyperscale platform engineering, ML infrastructure) are filled from a national labor market that overwhelmingly recruits from outside the region. The construction-phase jobs are real, time-limited, and largely awarded to specialty contractors who are not local hires either.

This is the parallel Big Nerd Idea takes most personally. The state's young people are being asked, once again, to participate in an economy as consumers of its externalities and as a tax-base supplier — not as authors of its design. A "tech boom" that requires you to leave home to participate in is not a boom for you.

The counterargument, steelmanned¶

The strongest version of the case for HB 2014 is not bad faith. It runs like this:

West Virginia has been losing population and tax base for two decades. Some county governments would genuinely prefer the data center revenue to nothing.
Construction-phase jobs are real, and they pay. So is the local supplier spend during the buildout.
Other states are aggressively competing for the same projects. If West Virginia does not preempt local zoning, the projects go to Virginia, North Carolina, or Tennessee, and West Virginia gets neither the upside nor any meaningful say in the regional siting outcome.
Some data center operators do invest in local workforce programs, renewable energy procurement, and community grants. The good actors deserve credit, and the structure should be designed to attract them rather than the bad actors.

We take these seriously. We do not think they dissolve the structural parallel.

The case against the structure is not "no data centers." It is "not on these terms." Stripping local zoning, viewshed, noise, lighting, and PSC rate authority from communities is not a neutral act of competitiveness. It is a transfer of decision-making power from West Virginians to entities that do not live here, do not pay state income tax here, and whose primary obligation is to a shareholder base on the coasts. That same transfer is exactly what made the coal pattern so durable and so harmful — and what makes "this time it'll be different" implausible without a different structure to support it.

It is worth noting: community pushback is not theoretical. Between March and June 2025, organized community opposition led to roughly $98 billion in data center projects being blocked or delayed nationally (WRI). The NAACP has stated that "no community should be forced to sacrifice clean air, clean water, or safe homes so that corporations and billionaires can build energy-hungry facilities." The local voice that HB 2014 preempts is, in the rest of the country, where the actual oversight is happening.

What Big Nerd Idea thinks should be different¶

BNI is small. We do not have the leverage to rewrite state statute. But our project portfolio has direct, concrete bearing on three of these four parallels, and we are deliberate about it:

On extraction: Our Sustainability & Carbon Awareness directive makes carbon a first-class design axis in every project we sponsor. The real-world Toaster Chef Trike is solar-first. Toaster Chef IT's whole thesis is refurbish-don't-manufacture. wholefolk's thesis is shorten the supply chain. None of these alone disproves the data center carbon story, but each of them is a small bet on the opposite economy.
On equity: The Toaster Chef IT curriculum exists specifically because the AI economy's workforce barrier-to-entry is the gap. A donated, rebuilt laptop and an 8-module open-source curriculum is one of the few interventions that actually moves a West Virginian from "consumer of the externalities" toward "participant in the design."
On voice: MPowerUP's P2P, off-grid-capable architecture is a deliberate [HYPOTHESIS] that vulnerable populations are best served by systems that do not require them to be visible to a central operator to function. That is a small but real counter-pattern to the data center model in which user data, identity, and behavior are extracted to coastal servers as a precondition of receiving service.

These are tools, not a movement. We do not pretend otherwise. But our position — and the reason this post exists in a "Major Issues" series — is that the open-ecosystem strategy is more than a technical preference. It is the same fight the state's prior generation had with coal, fought with different tools, on a much shorter timeline.

The Power Generation and Consumption Act is two years old as of this writing. The first certified districts have not yet broken ground. The pattern is not yet locked in. That makes the next eighteen months unusually important.

Sources¶

West Virginia HB 2014 / Certified Microgrid Program¶

Out-of-state / absentee ownership of WV land¶

Data center water & electricity demand¶

Hyperscaler capex and Appalachia siting¶

Validation status¶

Claim type	Marker
HB 2014 statutory language quoted	Documented — primary source
Historical absentee ownership percentages	Documented — secondary sources, widely cited
Hyperscaler capex projections	`[HYPOTHESIS]` — forward-looking, source-attributed
WV AI data center carbon trajectory	`[HYPOTHESIS]` — depends on specific projects not yet sited
Operations-staff headcount per 250 MW data center	`[HYPOTHESIS]` — industry range, project-specific in practice
Editorial position ("not on these terms")	Big Nerd Idea editorial position, not a research claim

This is an opinion piece anchored to documented evidence, not a peer-reviewed analysis. Corrections welcome via the Content Manager or GitHub issues.