The constitutional question now reaching state supreme courts is whether a transmission line built primarily to serve a private data center satisfies the “public use” requirement. The answer, emerging case by case, will determine whether the post-Kelo eminent domain regime can restrain the AI buildout or whether it has become the legal vehicle for the kind of private-interest-driven takings the 2005 reforms were designed to prevent. More than 3,000 data centers are already operating in the United States, with roughly 1,500 more in development, according to a Pew Research Center analysis cited by Aaron Walayat, an assistant professor of law at the University of Dayton. The facilities consumed more than 4% of the nation’s total electricity in 2024. The legal question matters because the answer will determine how fast the AI grid can be built and at whose cost.

The legal hinge

A relationship map of the dispute shows a bipartite structure held together by a single load-bearing node: the “public use” constitutional standard. One cluster is the infrastructure-and-energy chain. AI data centers drive electricity demand. That demand strains existing grid capacity. Power companies respond by building new transmission corridors. When landowners refuse to sell easements, utilities invoke eminent domain. The other cluster is the property-rights-and-legal chain. Eminent domain triggers the public use test. State supreme courts interpret that test. Post-Kelo reform laws constrain its scope. Public opinion shapes political pressure. Federal AI policy sets the macro context.

The public use standard is the only node connecting the two clusters. A judicial interpretation that tightens the standard — requiring that the primary purpose of a transmission line, not merely a spillover benefit, serve in-state customers — would sever the hinge and block the eminent domain pathway. The opposite ruling, affirming that any grid-reliability improvement qualifies, would give utilities a green light for AI-scale transmission. State-to-state variation in constitutional interpretation is the variable that determines bridge strength.

The state supreme courts that have already ruled illustrate how bridge strength varies. South Dakota and Vermont affirmed seizures where transmission lines provided grid reliability to in-state customers, finding that at least some local benefit satisfies the public use requirement. The Mississippi Supreme Court in 1984 blocked a condemnation where the line crossed into Louisiana without benefiting any Mississippi residents. That precedent provides the doctrinal seed landowners can invoke when the in-state-benefit claim is thin.

The reform that produced this paradox

The U.S. Supreme Court’s 2005 decision in Kelo v. City of New London — which allowed a Connecticut city to seize homes for private development around a Pfizer facility — triggered an immediate backlash. Forty-five states enacted eminent domain reform laws designed to prevent private-to-private takings. The Michigan Supreme Court in County of Wayne v. Hathcock (2004), and the supreme courts of Ohio and Oklahoma, prohibited the taking of private property to hand it to another private party purely for economic development.

But the reform movement preserved a utility exception that was structurally designed to keep the lights on, not to authorize AI-scale private compute facilities. The exception survived the Kelo backlash because the reforms targeted economic-development takings by municipalities, not condemnation by regulated utilities for grid infrastructure. It has never been stress-tested against the demand profile of a hyperscale AI facility.

The result is a legal paradox at the heart of the dispute. The reform laws intended to protect landowners from private-interest-driven seizure contain the very exception power companies now invoke to seize land for private compute facilities. Walayat notes that as data centers increase energy demand and stress current infrastructure, seizing land to improve grid reliability will likely qualify as public use, especially where the intention is to secure reliability for in-state customers. He also notes that arguments over whether new lines actually serve in-state customers “may give landowners grounds for a challenge.” That second observation is the doctrinal trap: courts ask whether the line benefits the public, but they do not ask who the line primarily serves, and they do not ask whether the public-use justification survives when the line’s anchor tenant is a private hyperscaler whose compute output is distributed globally.

The mechanism for landowner challenge is the scale mismatch between the legal standard and the physical system. Grid benefits operate systemically: a new transmission line that increases regional capacity benefits every customer on the connected network. But courts evaluate public use parcel-by-parcel, asking whether the specific taking before them serves a public purpose. A utility can demonstrate that the regional grid improves; a landowner can ask which specific in-state customers are served by a line whose anchor tenant is a private data center. The answer is not always straightforward, and that ambiguity is the procedural surface on which challenges proceed.

Who holds power, who doesn’t

A stakeholder map of the dispute surfaces a clear asymmetry. Power companies and data-center operators rank as definitive stakeholders — high power, high legitimacy, high urgency — and they form an aligned coalition with the federal executive’s pro-AI policy posture. The Trump administration has promoted AI advancement as critical to economic and national security, providing political cover for utility expansion without directly holding eminent-domain authority. State courts hold adjudicative authority. State legislatures hold the power to amend eminent-domain statutes but have taken no new AI-specific action.

Affected landowners rank as dependent stakeholders: high legitimacy through constitutional property rights and high urgency because land is being taken now, but low individual power against a utility with state backing. Their primary recourse is litigation, which is expensive, case-specific, and pits a one-time participant against a repeat player with in-house legal teams. Utility ratepayers — particularly lower-income and fixed-income households — have no procedural voice in eminent-domain proceedings. Local communities absorb higher utility bills, pollution, noise, and the loss of green space.

The map also exposes a structural paradox embedded in the legal standard. The public use test asks which in-state customers benefit from a transmission line; those customers are the doctrinal justification for the taking. But those same customers may face rate increases if the cost of new transmission corridors is socialized across the rate base. The test names the harmed as the beneficiaries, creating a doctrinal trap that makes challenge structurally difficult. The paradox persists regardless of which legal future materializes — whether courts affirm or restrict the eminent domain pathway, in-state ratepayers remain the doctrinal justification while bearing cost externalities no proceeding measures.

Polling shows seven in ten Americans oppose AI data center construction in their communities, citing higher utility bills, pollution, noise, and the loss of green space. But opposition measured in polls is not opposition organized into legal strategy, and the institutional channels that once enabled distributional-impact analysis for siting decisions have been stripped. Between January and February 2025, EJScreen was removed from the EPA’s website, the Climate and Economic Justice Screening Tool was taken offline, the Justice40 Initiative was terminated via rescission of Executive Order 14008, and Executive Order 12898 — the foundational federal environmental-justice directive — was revoked by Executive Order 14148. The Department of Transportation also rescinded guidance encouraging states to consider climate and environmental-justice impacts of infrastructure projects. Communities that might once have invoked federal screening tools to require distributional-impact analysis no longer have that scaffolding. The absence of formal channels for opposition is a structural condition produced by dated policy reversals, not a descriptive gap.

Parties structurally absent from the conversation include future ratepayers, who will inherit the cost structure of transmission infrastructure built today but have no representation in current proceedings; agricultural lessees and supply-chain businesses whose working landscapes are fragmented by easements but who have no standing in condemnation actions; out-of-state residents who bear reliability and cost externalities but are excluded from the in-state-customer test; workers in industries tied to undeveloped land; and tribal nations whose ancestral lands intersect transmission corridors.

Four futures for the buildout

Scenario analysis, projecting forward to 2035, identifies two critical uncertainties: whether courts clarify the public use standard or leave it fragmented, and whether community opposition consolidates into organized resistance or remains poll-measured. Their interaction produces four structurally distinct futures.

Transmission Express — courts affirm that grid reliability satisfies public use, and opposition stays diffuse. Utilities face minimal friction. Deployment accelerates. Aggressive capital deployment into greenfield corridors pays off only under these conditions; absent legal clarity, sunk capital in contested corridors becomes stranded.

Siege Buildout — courts affirm the standard, but opposition mobilizes. Utilities hold legal authority but face years-long delays per project — the analysis projects 18-to-36-month delay cycles — and project IRR drops from roughly 6% under permissive assumptions to 3–4% once delay costs are priced in. Reputational damage compounds.

Grid Gridlock — courts remain fragmented, and opposition consolidates. Takings are affirmed in some jurisdictions and blocked in others. Investment concentrates in states with favorable case law and weak opposition; high-demand jurisdictions stall. Operators who wait for national clarity lose position to those who pick permissive states and build local coalitions early.

Institutional Fog — courts stay fragmented, opposition stays passive. Utilities proceed project by project in an uneven, adaptive buildout: if one county says no, the utility crosses to the next. Slow but functional.

Strategies robust across all four futures include co-locating data centers near existing high-capacity transmission to reduce condemnation exposure; investing in behind-the-meter generation and battery storage as hedges against transmission-availability risk; diversifying across states with different legal regimes; building institutional relationships with state energy offices to demonstrate in-state benefit through jobs, tax revenue, and grid stabilization; and maintaining optionality through modular transmission designs that can be rerouted.

Wild cards

Two events sit outside the matrix because they would resolve the legal axis from above rather than through state-level dynamics.

A federal national-security reclassification of data center power delivery as critical infrastructure — via executive order or legislation invoking the Defense Production Act — could bypass state-level public use analysis entirely. The trigger would be a Department of Energy or Department of Defense report explicitly classifying data center power availability as a national security concern, or a visible AI-capabilities gap with a geopolitical rival becoming a galvanizing event. The national-security framing provides a doctrinal vector courts have historically treated with deference.

The opposite wild card — a U.S. Supreme Court ruling restricting public use for private data center infrastructure — would bar seizures for AI facilities unless the state establishes a public-ownership model for the line. The doctrinal seed is the 1984 Mississippi ruling. The indicator: a circuit split on the public use question, followed by a certiorari grant. If that ruling materializes, the buildout shifts entirely to voluntary easement negotiation or behind-the-meter generation, and community resistance becomes legally secondary because the taking authority itself has been withdrawn.

What to watch

Four developments will determine which scenario materializes.

Will courts require that transmission lines primarily serve in-state customers, or will spillover grid reliability be enough? The answer determines whether the post-Kelo utility exception stretches to cover AI-scale transmission — whether the reform-paradox continues to enable seizures or whether the courts close the loophole.

Will the seven-in-ten opposition translate from polls into institutional capacity: coordinated legal teams, county moratoria, ballot initiatives? Opposition currently has no formal channel into takings proceedings; building one would change the dynamics of every future condemnation case.

Will state legislatures act before courts force the question, and in which direction — codifying grid interconnection as public use, or adding procedural hurdles that slow or block seizures? Forty-five state reform statutes preserved a utility exception that now enables the kind of private-beneficiary taking the reforms aimed to prevent; legislatures have not yet addressed that paradox in the AI era.

And will the cost of building data-center-serving transmission fall on the ratepayers the legal standard names as beneficiaries, or will that paradox surface in a proceeding that forces the question into the open? The doctrinal mislabel — naming the harmed as the public-use beneficiaries — is the mechanism that makes challenge structurally difficult. Until a proceeding asks whether the in-state-customer benefit exceeds the ratepayer cost, the paradox operates silently.

The legal pathway and the engineering pathway operate on separate tracks. Courts decide who bears the negotiation cost — landowners, ratepayers, or utilities — but they do not eliminate the need for new transmission corridors. The infrastructure gap persists regardless of how the legal questions are resolved. What the courts decide now is not whether AI data centers get built, but how much of the cost of building them is borne by the people the legal standard names as the public they are being built to serve.

This analysis draws on reporting cited in the source article and three analytical lenses — relationship mapping, scenario planning, and stakeholder mapping. The facts are drawn from source reporting; the structural framing is the author’s. No claim is made about the intent of any party.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Relationship Mapping
Extracts the network of ties among people, institutions, and entities.
Scenario Planning
Builds a small set of distinct, plausible futures to plan against.
Stakeholder Mapping
Charts the parties to a situation — their interests, power, and alignments.