The Permitting Arbitrage

The tension between state economic authority and local land-use sovereignty has a long history in American public law. Data center permitting is emerging as its latest iteration.

American municipalities derive their authority entirely from the state, which can grant, limit, or revoke it at will. This principle, established in Hunter v. Pittsburgh in 1907 and elaborated by legal scholars including Richard Briffault, means that local zoning authority is a political achievement rather than a constitutional right. Communities can accumulate significant control over what gets built within their borders, but that control rests on state tolerance. When state economic priorities shift, the tolerance can be withdrawn.

Richard Schragger’s City Power traces the withdrawal across policy domains: minimum wage ordinances, environmental regulations, firearms laws. In each case states reasserted authority when local preferences conflicted with state economic strategy. The stated rationale was usually some version of efficiency and regulatory uniformity. The practical effect was to remove local communities from decisions that affected them most directly. In each case the preemption was contested, sometimes reversed, and always circling the same unresolved question: whose interests does the land serve, and at what scale should that be decided?

Data center permitting is the latest version of that fight, and the historical record is clear enough about how it tends to resolve. Preemption stabilizes when states can sustain a credible public-benefit argument, and erodes when communities come to see it as the state delivering their land, water, and grid capacity to corporate actors without adequate return. Which category data centers occupy is not yet settled, and the evidence is running in both directions. What that looks like in practice depends on which of three levers a state has pulled.

What is moving through state legislatures is more varied than the word “preemption” suggests. States are reaching for different levers, each with a different theory of how state authority should displace local control, and each producing a materially different risk profile for the developer and a different political sustainability profile for the state.

Preemption, simply put, is when a state uses its legal authority to override or constrain what local governments can do about a particular land use. But in practice, preemption is not a uniform phenomenon.

Standard practice is for a developer to file an application with a local planning commission, which can hold hearings, request studies, impose conditions, or simply extend the process. A determined opposition campaign does not need to win the final vote, but it does need to run the clock long enough that carrying costs exceed projected returns and the developer walks. Delay, in other words, is the opposition's primary weapon, and preemption exists to take it away.

States have taken three distinct approaches, each targeting a different part of the local approval process:

Substantive preemption is the most aggressive form. West Virginia’s HB 2014 eliminates county and municipal jurisdiction over certified high-impact data centers entirely. Local governments cannot enforce zoning, building, noise, viewshed, or land use rules against certified projects. Property tax revenue is redirected: 70 percent to a state income tax fund, 30 percent to the county, with school districts receiving nothing absent a separate levy. Senate President Randy Smith said publicly he expects “a lot of changes to HB 2014 over the next couple of years.” More than 930 residents submitted public comments opposing the implementing rules. West Virginia bet everything on a law its own Senate President says will change. For a developer underwriting a ten-year hold there, that is the lead.

Procedural preemption constrains how long local governments can take rather than removing their authority. The original version of Pennsylvania’s SB 939 proposed a shot clock of 30 days for a use permit and 120 days for conditional use approval, with applications deemed approved if the window passed without action. Those provisions were stripped in committee in February 2026 under municipal and environmental pressure. The shot-clock concept lives in other pending bills. As a mechanism, it is more durable than substantive preemption because it preserves the form of local authority while limiting its exercise. But, it is also weaker because a practiced opposition can accomplish a great deal within a 120-day window.

Incentive conditionality works through fiscal pressure rather than legal displacement. Jurisdictions that create permitting friction lose eligibility for state incentive programs. A jurisdiction committed to blocking a project will absorb that loss. The mechanism is weak as coercion but durable politically, because it does not require the state to claim that community voice is legally irrelevant, only that it carries fiscal consequences.

Georgia state Senator Jaha Howard introduced a bill in early 2026 to ban new data center construction for a year. His constituents were watching electricity bills rise to pay for grid infrastructure serving hyperscale facilities, while the promised jobs had not arrived. “We welcome investment from technology companies,” he said, “but our constituents need peace of mind on the short and long-term impacts of these projects on their communities.” The bill went nowhere. Six others like it, also, went nowhere. Lifelong Republican voters in suburban Atlanta districts are now telling Democratic candidates they are reconsidering their political alignment over data centers. Georgia is estimated to be losing $2.5 billion annually to data center sales tax exemptions. When the fiscal transfer is visible and the jobs are not, the arithmetic eventually turns electoral.

Political scientists call this venue shopping: losing parties shift to the arena where they have more leverage. Communities blocked at the planning commission push for moratorium statutes. Developers blocked locally push for state preemption. Neither side is standing down, which is why the country is bifurcating rather than finding equilibrium.

In Virginia, which contains more data center capacity than any other state, the incoming governor entered office committed to requiring data centers to fund their own energy infrastructure. Bills have been introduced to eliminate tax exemptions, impose excise taxes, enact siting restrictions, and halt new applications until interconnection queues clear. A state that spent years competing for data center investment is now legislating against it. The fortification wave has its own momentum, its own legal infrastructure, and its own electoral logic. This is not to be mistaken as the losing side of a fight. To be clear, this is simply a second front.

Capital markets reprice assets when risk profiles change, but the repricing lags when the risk is qualitative and recent. The structural divergence in permitting regimes has been empirically visible for only eighteen to twenty-four months. Land basis in preemption-stable states does not yet carry the premium that reliable, compressed timelines should command. In contested markets, basis does not yet reflect the risk-adjusted cost of the political environment sitting beneath it. That gap will close as cancellations and delays accumulate into a documented pattern rather than anecdotal evidence.

The mechanism is already visible at the edges. LP due diligence checklists are beginning to include permitting regime questions that did not appear two years ago. Co-investment term sheets in contested markets are starting to carry entitlement contingencies. A handful of institutional lenders have begun requiring explicit entitlement risk disclosure as a condition of construction financing. None of this is yet standard, but when it is, the repricing accelerates, and the window for acquiring preemption-stable ground at pre-repricing basis closes.

If your portfolio has exposure in either regime type, the repricing is coming for you in one direction or the other.

The trap is treating “preemption state” as a proxy for low entitlement risk. West Virginia is the clearest case. HB 2014 is the most aggressive preemption measure in the country and is generating the most organized political blowback: calls within the Republican caucus for revision, nearly a thousand public comments against the implementing rules, and the diversion of school district property tax revenues that is hardening into a durable local grievance. Preemption that defunds public schools in a state where one in five residents lacks reliable broadband carries a specific and measurable political liability. The Senate President’s own prediction of significant changes over the next couple of years is not a footnote. For a developer underwriting a ten-year hold in West Virginia, that prediction is not something you skim, it needs to be underwritten.

Although, the mistake is written in both directions. Take Pennsylvania for example, it is not a preemption state. It's a state where preemption was proposed, debated, stripped in committee, and replaced with a coordination framework. Treating it as one static category in a portfolio model is a category error, and the relevant investment variable is not preemption status at acquisition but regime trajectory across the holding period.

WilmerHale’s February 2026 analysis identified at least three distinct federal vectors that could constrain state data center legislation independent of state constitutional challenges: Telecommunications Act claims, Federal Power Act arguments over wholesale market interference, and Dormant Commerce Clause challenges targeting utility rate differentials. The National League of Cities has been active in supporting municipal challenges. More than 200 bills were introduced across all 50 states in 2025, with over 40 enacted. That volume almost guarantees the drafting inconsistencies courts find actionable.

The federal executive order of July 2025 explicitly did not preempt state permitting requirements, zoning laws, or energy regulations. Washington chose to work around the bifurcation rather than resolve it.

The jobs argument that historically anchors pro-development preemption is exceptionally weak for this asset class: nine in ten construction jobs disappear when the building opens. Preemption stabilizes when states can sustain a credible public-benefit argument. For data centers, that argument is failing. The incentive programs are now publicly quantified: $2.5 billion annually in Georgia, $1.6 billion in Virginia in a single fiscal year. Those numbers do not require an opposition movement to become politically dangerous; they only require a budget line and a legislator willing to read it aloud.

On June 16, the San Marcos city council voted 4–3 to rewrite its zoning code and ban data centers citywide, with no active proposals pending inside city limits. It used the one tool available that falls outside the statutes being cited against local governments: a zoning code change rather than a moratorium. Three hundred and fifty-two other Texas cities hold the same authority, and they are watching the legal challenge that is coming and deciding whether to act before it resolves or wait for the precedent. If you have sites in Texas, the legal challenge playing out in San Marcos is your underwriting problem.

West Virginia’s HB 2014 looked like durable preemption architecture in spring 2025 and is generating calls for revision less than a year later. Georgia’s data center tax program looked permanent and is now subject to sunset proposals with bipartisan support. Pennsylvania’s preemption provisions were stripped in committee. Virginia’s incoming governor campaigned against the existing deal. None of these developments was unforeseeable. Each was a recognizable move in a recurring pattern of state-local conflict over industrial land use, visible to anyone tracking the structural dynamics rather than the individual project outcomes.

In a country that has split into structurally distinct regulatory regimes for the same asset class, the relevant unit of risk analysis is no longer the parcel or even the market. It is the regime itself: its mechanism type, its legal vulnerability, the fiscal sustainability of its incentive architecture, and its trajectory across an investment horizon rather than its status at acquisition. The firms that will price this landscape correctly are the ones that understand permitting environments as political equilibria and have built the capacity to assess whether a given equilibrium is hardening or eroding. The losses accumulating in the contested half of the country are the cost of not having that capacity, and they compound late in development timelines where exit is most expensive and capital is least recoverable. The market has not yet built the capacity to track regime trajectory rather than regime status.