Ideas worth negotiating
If the deal happens, how could Pikeville get the most out of it?
Most of this site examines whether the proposed data center's impacts are acceptable. This page asks a different question: if the city decides to proceed, what would it take to turn a one-time deal into lasting value? Nothing here is hypothetical invention: every idea is something another community has actually done, scaled to Pikeville's real numbers.
How to read this page
These are not recommendations, and this site is not affiliated with any party to the negotiation. Each idea is presented the same way: what another place did (sourced), what the equivalent arithmetic looks like here (shown), and what would have to be in the Development Agreement for it to happen. Whether any of it is right for Pikeville is the community’s call. The window for asking is now: the city has committed to publishing the complete agreement for public review before any vote.2
Idea one: don't waste the heat
Precedented elsewhere
A 25–30 MW data center is also a 25–30 MW heater running around the clock, and the AI-era cooling designs produce hot water (roughly 50–60°C from direct-to-chip systems), warm enough to be genuinely useful next door.134 The precedents are small, unglamorous, and real: Notre Dame’s server racks heat South Bend’s municipal greenhouse and save the city about $70,000 a year,131 a 28 kW micro data center cut an English town pool’s gas bill 62%,132 and Amazon’s Seattle towers are warmed by a neighboring building’s waste heat.130 A regional policy group, ReImagine Appalachia, has argued for exactly this use on reclaimed mine land.136
The Pikeville version writes itself. The data center would sit inside an industrial park the city spent a decade building, with other tenants and a completed Build-Ready pad next door.159 The city’s own review asks how a data center would affect the park’s ability to serve other industrial employers;2 heat reuse flips that question. Free or near-free process heat could make the remaining lots more attractive, not less: greenhouses, food processing, aquaculture, or simply heating a neighboring tenant’s building. The first candidate already has a budget line: the city’s FY2026-27 budget funds a $6 million, 50,000-square-foot spec building inside the same park, built to attract a future employer.5 If the data center and the spec building are designed together, the spec building could offer a tenant something almost no industrial park in America can: heat as a utility, nearly free.
What would have to be in the agreement: heat-recovery plumbing designed in from day one. It is cheap at construction and prohibitively expensive to retrofit, and no U.S. operator will volunteer it unasked, because almost no U.S. town has asked.133
Idea two: bank the windfall, don't just spend it
Precedented at state scale
The pattern everywhere resource money has appeared suddenly (oil, gas, and now data centers) is that places which invested the windfall kept benefiting after the boom, and places that simply absorbed it into the budget did not. The precedent ladder, from biggest to closest:
- Alaska put a share of its oil revenue into a permanent fund starting in 1976. The fund is now roughly $85 billion143 and paid every eligible Alaskan a $1,000 dividend in 2025.144
- New Mexico amended its constitution in 2022 to draw an extra 1.25% a year from its oil-and-gas-fed Land Grant Permanent Fund (about $150 million annually, 60% of it for early childhood education), which is how it became the first state to offer universal free child care.145
- North Dakota created its Legacy Fund by ballot measure in 2010; defined earnings now flow to the state on a percent-of-value rule, including a $686.9 million transfer to the general fund in the 2023–25 biennium.146
- West Virginia (the data center version): a 2025 law routes property taxes from large data centers by formula: 50% to a Personal Income Tax Reduction Fund, 30% to the host county, 10% shared among all counties, and 5% each to grid and economic-development funds.139 The host-county share exists because counties pushed back and won a bigger slice,140 and critics argue the model diverts money from local schools and public safety to fund tax cuts, a trade-off any version of this idea has to face honestly.141
- Loudoun County, Virginia (the pay-as-you-go alternative): no endowment, but data center revenue (38% of its general fund, $100M+ of new revenue a year) let it cut the homeowner property tax rate every year for a decade, to the lowest in Northern Virginia.142
Pikeville’s actual numbers. The city’s general fund collected about $21.4 million in FY2024: $12.2 million of it from the 2% occupational license fee, and just $1.07 million from property taxes.137 The FY2026-27 budget runs $29.2 million.5 The proposed project’s initial phase exceeds $250 million of capital investment, much of it computing equipment the city itself notes is “subject to standard Kentucky personal property tax assessments,” and the city has said it will not cut local taxes as an incentive.2
Kentucky’s official 2025 rate tables make the tax math concrete. Computer equipment is taxed at the state’s 45-cent rate plus full local rates: 44.7 cents across Pike County’s districts, 14.6 cents for the City of Pikeville, and 74.7 to 82.8 cents for the school district, roughly $1.80 per $100 of value, every year.148 On $150 million of servers, that is about $2.7 million a year in total property taxes while the equipment holds its value, and the split is the real story: the school district collects the most, around $1.1 million a year; the county’s districts about $670,000; the state about $675,000; and the city itself only about $220,000, against its current $945,000 property-tax line.1485 Server values depreciate, but data center hardware is replaced on a constant cycle, which keeps assessments replenished.
The city’s sleeper revenue stream is the electric franchise. Under a franchise agreement on file with state regulators, Kentucky Power pays the city 3% of all revenues from electric service inside city limits.149 That line is budgeted at $730,000 for FY2026-27, which implies roughly $24 million of electricity sold in the entire city today.5 A 30 MW data center running around the clock would buy nearly that much power by itself (about $16-21 million a year at typical industrial prices), adding roughly $470,000 to $630,000 a year to the franchise line; at the 75-100 MW build-out, $1.6-2.1 million a year. Two caveats: the facility’s service must actually be billed inside city limits, and the current franchise term runs ten years from April 2020, so renewal on the same terms is an assumption, not a guarantee.149 Combined with the equipment taxes and whatever land terms are negotiated (a lease rather than a sale, host-community payments), a city share of $1-2 million a year at the initial phase is arithmetic, not optimism; the Development Agreement decides whether it happens.
What disciplined banking could build. If the city set aside a fixed amount every year for 20 years, earning a 5% average return, then switched to drawing 4.5% of the fund per year (roughly the rule big endowments live on), the perpetual annual income looks like this:
Math: 20 annual contributions compounding at 5% (future-value factor 33.07), then a 4.5% annual draw. These are illustrations of compounding, not predictions: actual deal revenue is unknown, returns vary, and inflation erodes fixed draws. For scale: the top row equals roughly a quarter of today’s entire general fund,137 arriving every year whether or not the data center still exists.
Two design realities. First, Kentucky law limits what cities may do with public funds: essentially government obligations, insured or collateralized deposits, and highly rated paper, under a written investment policy.147 An endowment that owns stocks would need an independent trust or foundation structure, or a change in state law. Solvable, but it must be designed, not assumed. Second, the West Virginia critique applies at any scale: a dollar banked is a dollar not spent on today’s needs.141 Whether to favor the next budget or the next generation is a values question only the community can answer; the precedents just prove both answers are available.
Idea three: structure the deal so the leverage lasts
Precedented elsewhere
The quietest lesson from other towns is that how the deal is papered matters more than its headline numbers. Three structures with track records:
- Lease, don’t sell. The city still owns the land; the MOU transfers nothing.2 A long-term ground lease produces recurring revenue a sale can’t, and the land reverts if the project dies: no abandoned shell the city has to buy back.
- Tie commitments to the land, not the company. Data centers change hands. The city’s review already flags this: commitments should survive any future sale or restructuring, with recourse at every stage.2
- Milestones with consequences. Memphis got a groundbreaking ceremony for an $80M water-recycling plant; then the project was paused.111 Mesa, Arizona wrote tiered water caps directly into its agreement.109 Pledges need dates, measurements, and defined consequences, including for the jobs numbers, which the city says it intends to make enforceable rather than aspirational.2
The honest caveat
Every scenario on this page depends on terms that do not exist yet. The MOU commits the city to nothing but a period of exclusive negotiation, and the developer has no signed end user.2 The numbers above are arithmetic on cited inputs, not forecasts. What the precedents establish is narrower but useful: communities that decided what they wanted before signing got it written in; the ones that decided afterward mostly didn’t.