The binding pricing event for a 200 MW AI campus is no longer the PJM cluster study, no longer the state public service commission rate case, no longer the hyperscaler procurement committee. It is the regularly scheduled Tuesday evening meeting of a five- to nine-member elected county board in a jurisdiction that, in most cases, has fewer than 100,000 residents. The decision being made in that room will move 150 to 400 million dollars of underwriting up or down by more than the entire prior twelve months of engineering, transmission, and tenant work combined.
The events are documented and named. Prince William County, Virginia, voted 5 to 2 in December 2024 to advance the PW Digital Gateway rezoning over more than 4,000 hours of public comment and a record-length hearing sequence. Pima County, Arizona, and the City of Tucson rejected the Project Blue campus in 2025 after the project sponsor declined to disclose end-user identity. Memphis-Shelby County reversed two air permits issued to xAI's South Memphis facility in 2025 after community organizing produced documented turbine-related emissions exceedances. Washington County, Oregon, and the City of Hillsboro froze new data center allocations through the Tualatin Valley Water District in 2024. Loudoun County, Virginia, revised its data center zoning ordinance in 2025 to add transmission-line setback requirements, noise studies, and water-use disclosure. Fairfax County passed a parallel zoning text amendment in early 2026. Six pricing events. Each one moved more capital than the regulatory ruling that preceded it.
The Five Votes That Priced The Pipeline
The Prince William County 5-to-2 vote on December 12, 2024, approved the comprehensive plan amendment for the PW Digital Gateway, a 2,100-acre rezoning bordering Manassas National Battlefield Park. The decision came after 27 hours of public comment across two hearings, a procedural marathon now treated inside the industry as the longest single zoning hearing in modern Virginia history. The vote unlocked an estimated 23 million square feet of buildable data center floor area and stabilized roughly 4.5 GW of forward power demand inside the Northern Virginia load pocket. The two dissenting supervisors represented districts containing the contested parcels.
The Tucson and Pima County Project Blue rejection in 2025 was the inverse event. The project sponsor, operating under an NDA structure that named neither the hyperscaler tenant nor the end-use, sought entitlements for a campus of approximately 290 acres adjacent to the Pima County wastewater system. The city council voted unanimously against the proposed development agreement after the sponsor declined to disclose tenant identity, projected water draw, or projected economic terms inside the public hearing record. The rejection eliminated a development that would have consumed 1 to 2 million gallons per day of reclaimed water at full buildout. No subsequent application has been filed at the same parcel.
The Memphis-Shelby County air permit reversals against xAI's Colossus facility came in two stages in 2025. The Shelby County Health Department issued the original permits in mid-2024 for a temporary configuration involving approximately 35 mobile gas turbines installed at the South Memphis former Electrolux site. Community organizing led by the Southern Environmental Law Center and Memphis Community Against the Pollution documented in-field emissions exceedances using third-party monitoring. The Health Department reopened the permits, imposed additional conditions, and the County Commission held parallel oversight hearings. The episode established that an air permit issued under state and local authority remains revisable post-issuance when site monitoring data conflicts with application representations.
The Hillsboro-Washington County allocation freeze in 2024 was a water-side pricing event. The Tualatin Valley Water District, governed by an elected five-member board, imposed a moratorium on new large-volume industrial allocations exceeding 1.0 MGD pending completion of the next basin-wide supply study. The moratorium captured at least four announced or pending data center applications in the Hillsboro corridor, each of which had already advanced through Oregon Department of Environmental Quality preliminary review. The freeze remains in effect pending the 2026 supply update.
The Loudoun County zoning revision in 2025 and the Fairfax County zoning text amendment in early 2026 are the two most-cited ordinance revisions in the Northern Virginia corridor. Both add noise study requirements, transmission-line setback minimums of 200 to 300 feet from residential parcels, mandatory water-use disclosure inside the site plan, and substation siting review separate from the data center building review. Both ordinances apply retroactively to applications not yet under construction. The Loudoun revision moved approximately 4 to 6 GW of pipeline capacity into reconfiguration. The Fairfax amendment moved another 1 to 2 GW.
Why The Pricing Event Moved To The County
The pricing authority migrated to the county because three conditions converged. None of them is reversible inside the planning horizon.
1. Federal and state preemption ran out. Under the federal Telecommunications Act, the Federal Power Act, and the Natural Gas Act, certain categories of energy and communications infrastructure enjoy federal preemption against local zoning. Data center campuses, despite their power consumption, are not preempted. They are by-right or special-use industrial structures under state-delegated zoning authority, which is exercised at the county or municipal level in every US jurisdiction. The same is true of water withdrawal permits below state-threshold sizes, of cooling tower air permits in non-attainment areas, and of substation siting outside the FERC-jurisdictional transmission backbone.
2. The scale crossed the visibility threshold. A 50 MW campus on 30 acres in a county zoned for light industrial draws a different civic response than a 1,500 MW campus on 800 acres across three precincts. The 2022 to 2025 inflection in campus size, driven by training cluster economics and the move from CPU-dense to GPU-dense floor plate ratios, pushed the median permitted project past the visibility threshold in roughly 40 of the top 60 US data center counties. Above the threshold, the project becomes a county-board agenda item by default rather than a planning-staff administrative approval.
3. The procedural calendar compressed. Hyperscaler tenants in 2024 to 2026 are signing build-to-suit leases with delivery dates inside 28 to 36 months from LOI. The historic municipal entitlement timeline of 18 to 36 months no longer fits inside the lease. The developer must front-load community engagement into the pre-LOI window, which means the county-board calendar now sits upstream of the procurement decision rather than downstream of it. The supervisor's meeting moved from approval ceremony to negotiation venue.
The aggregate effect is that a board of supervisors of five to nine members in a county of 40,000 to 400,000 residents now holds the dispositive vote on whether 100 to 800 million dollars of capital deploys inside that county's boundaries. The vote is unappealable inside the project's commercial timeline. A court challenge to the zoning decision typically resolves in 18 to 30 months. The lease delivery date does not move.
The Institutional Response
The capital response to a pricing event that lives inside a county chamber has begun forming along three axes. The first axis is firm specialization. The second is in-house function buildout. The third is third-party verification.
The firm specialization axis is the emergence of community engagement firms that operate at the campus-development scale and that price their work as a capital-markets input rather than a public-relations cost. Cleyera, founded in 2024 and headquartered in Washington, builds civic-license diligence packages for hyperscaler and developer clients across Virginia, Texas, and Arizona. ClearPath Strategies and the related ClearPath Action policy arm operate in the same adjacent space, layering polling and constituent analysis onto specific county entitlement campaigns. Civic Capacity, founded in 2025 by alumni of the 2024 Prince William hearing cycle, retails county-by-county engagement playbooks at five- to seven-figure annual retainers per active project. These firms are not contract lobbyists. The function they sell is the same function an underwriter buys from a third-party engineering firm: a defensible, repeatable diligence output that can be referenced in the closing binder.
The in-house function buildout is the emergence of the civic-asset chief reputational officer or community capital officer role inside the largest five to ten data center developers and the top three hyperscalers. The role title varies. The function is consistent. The CRO maintains a directory of county-board listening cadences, attends supervisor district meetings before any LOI is signed, manages cumulative-impact disclosure documentation across multiple campuses in a single jurisdiction, and owns the internal go/no-go recommendation on each civic-license question. Compensation is set in the 350,000 to 600,000 dollar range with carry exposure or restricted equity participation in the campuses the function touches.
The third-party verification axis is the emergence of benefit verification firms. A board of supervisors approving a campus rezoning increasingly conditions the approval on quantified community benefits: a per-acre payment-in-lieu-of-taxes schedule, a job-creation floor with hourly-wage minimums, a water-use offset commitment, a transportation impact mitigation deposit, a transmission undergrounding contribution. The benefits are documented inside the development agreement. The verification that the benefits arrive on schedule and at the committed magnitude is now contracted out to specialist firms operating in the same space as muni-bond continuing disclosure agents. Two to four such firms are now retained on the largest Virginia and Texas campuses on three- to five-year initial terms.
The Civic Due Diligence Stack
The diligence package that an underwriter or hyperscaler procurement committee now expects to see before approving a campus investment has acquired a defined shape. The package is layered the way a Phase I environmental site assessment is layered. The components have begun to converge.
1. The county-board composition map. A two- to four-page brief naming each supervisor, their district, their term-expiration calendar, their declared positions on data center development, their largest campaign donors of record, their voting history on the three most analogous prior land-use decisions in the same jurisdiction, and the swing-vote arithmetic against the specific application.
2. The cumulative-impact disclosure. A jurisdiction-level inventory of every announced, permitted, under-construction, and operating campus inside the same county or adjacent counties, with aggregate MW of load, aggregate acres of footprint, aggregate water draw at full buildout, aggregate diesel backup generation, and aggregate transmission-line miles. The disclosure is increasingly required by ordinance and is being requested by sophisticated supervisors even where ordinance does not require it.
3. The community-license polling baseline. A statistically defensible poll of likely voters inside each affected supervisor district, with cross-tabs by precinct, length of residency, homeowner-vs-renter, and prior data center proximity. Polls are commissioned at 15,000 to 50,000 dollars per district at the campaign-grade tier and are conducted at 60-day intervals across the entitlement window.
4. The benefits schedule with verification protocol. The quantified community benefits committed in the development agreement, with a stipulated verification cadence (semi-annual or annual), a stipulated verifier of record, and a stipulated cure mechanism if a benefit is missed. The cure mechanism is increasingly an escrow draw rather than a remedial-action plan.
5. The cumulative-noise study and the transmission-line setback compliance memo. Both items now standard in Virginia, Oregon, and select Texas jurisdictions. The noise study is conducted by an independent acoustic engineer with a state-recognized credential. The setback memo references the specific ordinance section and the specific surveyor's plat of record.
6. The tenant-disclosure decision. Whether the end-user hyperscaler will be named inside the public hearing record, and if not, what the substitute disclosure language is. Tucson made tenant disclosure a binding test in 2025. Other jurisdictions are following.
7. The water-use disclosure. The full-buildout makeup water demand stated in MGD with seasonal variation, the source-water identification, the alternative-source backup, and the cooling-system design specification. Increasingly the disclosure is referenced inside the development agreement as a not-to-exceed commitment with monitoring requirements.
How A Rating Emerges
The structural progression from ad hoc civic engagement to a standardized institutional rating moves through four phases. The first three phases are visible in the current market. The fourth is forming.
Phase one is firm-by-firm proprietary scoring. Each large developer maintains an internal county-risk grid, scored on a five- to seven-point scale, updated quarterly, and treated as confidential commercial information. The grids are not consistent across firms. A county scored at the highest risk tier by one developer is sometimes scored at the lowest risk tier by another, reflecting differences in track record, prior community-benefit deployment, and supervisor relationship depth. Phase one is the current state in roughly 90 percent of active markets.
Phase two is third-party diligence consortia. Two or more developers, frequently competing for adjacent sites inside the same county, jointly retain a common engagement firm and a common verification firm. The shared firm produces a common county-license diligence output that is referenced inside each developer's underwriting. The structure mirrors the title-insurance loss-pool model. Phase two has begun appearing in Northern Virginia, Central Ohio, and selected Texas counties.
Phase three is rated benefits issuance. A community benefits commitment with an escrow-backed cure mechanism begins to look, structurally, like a municipal continuing disclosure obligation. The rating agencies have not yet moved into this space. The municipal advisors and bond counsel have. The first rated community-benefits agreements, structured as wrap-around obligations to project-finance term loans, are appearing in 2025 to 2026 transactions in Virginia and are expected in 2027 to 2028 in Ohio and Texas.
Phase four is a standalone community-license rating. The rating is assigned to a jurisdiction rather than to a transaction. The inputs are public-record vote history of the relevant elected body across the last five to seven years, ordinance stability, the existence and enforcement of a benefits-verification protocol, the cumulative-impact disclosure standard, and the documented track record of campaigns that converted from announcement to operating buildout without ordinance reversal. The rating becomes a covenant input. A campus inside an investment-grade community-license jurisdiction prices 50 to 150 basis points tighter than the same campus inside a sub-investment-grade jurisdiction. By 2028, the spread is documented inside underwriting models at the top four campus lenders.
What Becomes True By 2028
By 2028, every campus above 200 MW underwritten by an institutional investor will reference a third-party civic-license diligence output inside its closing package. The output will be produced by one of six to ten specialist firms operating across the Virginia, Ohio, Texas, Arizona, and Oregon corridors. The diligence will be a covenant deliverable, not a discretionary attachment.
By 2028, at least 12 to 18 US counties will have published cumulative-impact ordinances that cap aggregate data center MW, acres, or water draw at the jurisdiction level. The caps will be reviewed on a 24- to 36-month cycle, indexed against utility-supply forecasts and watershed studies. The map of compute siting will be redrawn around the cap geometry.
By 2028, the in-house community capital officer role will exist at every one of the top 20 US campus developers and at the top six US hyperscalers. The role will report into either the chief development officer or the chief financial officer, not into the corporate communications function. Compensation will sit inside the 450,000 to 750,000 dollar band with carried-interest exposure on the campuses the function touches.
By 2028, the first jurisdiction-level community-license rating will be issued, either by a new specialist firm or by an existing municipal-rating analyst inside one of the established agencies. The rating will be referenced inside campus term loan documentation in the same paragraph as the offtake rating and the host-utility rating. A downgrade event at the jurisdiction level will trigger a covenant test the way a downgrade at the offtake counterparty triggers a covenant test today.
By 2028, the gap between the highest- and lowest-rated US data center counties will be wider, more legible, and more priced than at any prior point in the asset class. Counties that have built durable civic-license infrastructure will absorb the marginal next decade of campus deployment. Counties that have not will become permanent inventory of stranded transmission, stranded substation, and stranded entitlement applications. The pricing authority will not return to the procurement committee. It will stay where it moved. The five-member board of supervisors in a county of forty thousand residents is now the venue at which the asset class clears.