Property Rights, Not Zoning, Should Settle the Data Center Debate

The Mises Institute published a provocation in May 2026 that cuts straight to the core of a growing land-use conflict: who decides where data centers get built? The article, titled "A Rothbardian Case Against Bad Data Center Policy," rejects both the pro-growth faction that wants blanket permission for data center construction and the NIMBY coalition that wants blanket prohibition. Instead, it argues for something older and simpler: property rights. The framing matters because the AI boom has turned data center siting into a trillion-dollar political question, and neither side of the conventional debate is asking the right one.

The Scale of the Boom

The numbers behind the data center rush are staggering. Global data center capital expenditure is projected to exceed $350 billion in 2026, driven almost entirely by the compute demands of large language models and AI inference workloads. Microsoft, Google, Amazon, and Meta have each announced tens of billions in new data center spending. Microsoft alone committed over $80 billion for fiscal year 2025 data center construction. Amazon Web Services disclosed plans for more than $100 billion in capital expenditures over the coming years, with data centers consuming the bulk.

The physical footprint of this spending is enormous. Northern Virginia's Loudoun County, long known as "Data Center Alley," hosts more than 300 data centers and accounts for roughly 70% of global internet traffic routed through the region. But the boom has spread far beyond Virginia. Central Ohio, central Indiana, rural Georgia, and parts of Texas and Arizona have all become targets for hyperscale campuses that can span hundreds of acres and consume hundreds of megawatts of electricity.

A single hyperscale data center can draw 100 megawatts or more, roughly the power consumption of 80,000 homes. Some proposed campuses would require their own dedicated power plants. In 2025, utilities in multiple states warned that projected data center demand could strain grid capacity within three to five years. Dominion Energy in Virginia, AEP in Ohio, and Georgia Power have all flagged capacity constraints tied directly to data center growth.

The Two Bad Policies

The conventional debate over data center siting has settled into two camps, and neither is satisfactory.

The first camp favors aggressive state and local incentives to attract data center investment. Virginia, for example, has offered sales tax exemptions on data center equipment since 2009. The exemption has cost the state an estimated $1.4 billion in foregone tax revenue over its lifetime. Other states have followed with similar packages: property tax abatements, expedited permitting, dedicated utility infrastructure funded by ratepayers, and even direct cash grants. The logic is straightforward: data centers bring jobs, tax base, and technological prestige. But the incentives create a bidding war among jurisdictions, socializing costs while privatizing benefits. When a utility builds a new substation to serve a data center and spreads the cost across all ratepayers, the residents of a rural county end up subsidizing a tech giant's compute infrastructure. This is not a market outcome. It is industrial policy dressed in free-market language.

The second camp wants to block data center construction through zoning restrictions, moratoriums, and environmental review requirements. Loudoun County residents have organized against new projects, citing noise pollution from cooling systems, increased truck traffic, strain on water supplies, and the aesthetic blight of windowless concrete boxes. In 2023, Prince William County, Virginia, imposed a temporary moratorium on new data center applications in certain areas. Similar pushback has emerged in Oregon, where residents near The Dalles have objected to Google's water consumption, and in Ireland, where EirGrid warned that data centers could consume 30% of the nation's electricity by 2028, prompting a government moratorium on new connections in the Dublin area.

The moratorium approach is just as flawed as the subsidy approach, but from the opposite direction. Rather than using price signals and property rights to resolve competing land uses, it uses political fiat. A county board of supervisors votes to prohibit a land use, overriding the property rights of landowners who might willingly sell or lease their land for data center development. The restriction benefits some residents, usually those already established in the area, at the expense of both the landowner and the potential builder. Neither party gets a voice. The decision is made by majority vote, not by voluntary exchange.

The Rothbardian Alternative

Murray Rothbard's framework offers a different path. In "Man, Economy, and State" and later in "The Ethics of Liberty," Rothbard argued that property rights, strictly defined and rigorously enforced, are sufficient to resolve conflicts between competing land uses. The principle is straightforward: you can do whatever you want on your property, as long as you do not physically invade or damage someone else's property.

This framework handles the data center question elegantly. A landowner who wants to sell 200 acres to Microsoft for a hyperscale campus has every right to do so. But if the data center's cooling systems produce noise that crosses the property line at harmful levels, that constitutes a physical invasion, actionable as a tort or a nuisance claim. If the data center's water consumption draws down a shared aquifer, that is a property rights dispute over a common resource, resolvable through prior appropriation rules or negotiated easements. If the construction traffic damages a neighbor's road, that is a straightforward trespass claim.

The key insight is that none of these disputes requires a zoning board, a moratorium, or a subsidy. They require courts, clearly defined property boundaries, and the common law of nuisance and trespass. The government's role, if any, is to adjudicate disputes when they arise, not to preemptively decide which land uses are permissible.

This is not a fringe position. It is, in many ways, a return to the pre-zoning legal regime that prevailed in most of the United States before the 1920s. The first comprehensive zoning ordinance in the country was enacted in New York City in 1916. Before that, property disputes were handled through private negotiation, restrictive covenants, and the courts. Zoning was sold as a rational improvement over the messy common-law process. A century later, the results are mixed at best: zoning has produced housing shortages, industrial deserts, and now, a political fight over data centers that a property rights framework would handle without legislative drama.

The Government Distortion

The Rothbardian critique cuts deeper than land use. The entire data center boom is shaped by government distortions that a free market would not produce.

Start with electricity pricing. In most of the United States, electricity rates are set by public utility commissions, not by supply and demand. Industrial customers, including data centers, often negotiate below-market rates that are cross-subsidized by residential ratepayers. If electricity were priced at its true marginal cost, including the cost of new generation capacity needed to serve data center load, the economics of data center siting would look very different. Some locations would become prohibitively expensive. Others, particularly those near abundant cheap energy, would become more attractive. The market would sort this out efficiently. The regulatory process does not.

Then consider water. Many data centers use evaporative cooling, consuming millions of gallons per year. In the western United States, water rights are governed by prior appropriation doctrines that predate most environmental regulation. But in practice, water allocation is heavily politicized. Agricultural users with political clout defend their allocations. Municipal users demand priority. Data centers, as newcomers, face hostility. A pure property rights regime would allow water to be bought and sold at market prices, directing it to its highest-value use. The current regime allocates water through politics, which means that the loudest voice wins, not the highest bidder.

Tax incentives are the most obvious distortion. When Virginia exempts data center equipment from sales tax, it is picking winners. The revenue lost to the exemption must be made up elsewhere, either through higher taxes on other businesses and residents or through reduced public services. The data center operator captures the benefit. Everyone else pays. This is the opposite of a free market. It is mercantilism with a silicon veneer.

Bitcoin Mining and the Parallel Case

The data center debate has a direct parallel in the Bitcoin mining industry, and the lessons run both ways.

Bitcoin miners face identical siting conflicts. They consume large amounts of electricity, generate noise and heat, and often locate in rural areas where cheap power is available. They have faced moratoriums in New York state, regulatory hostility in multiple jurisdictions, and community opposition similar to what AI data centers now confront. New York's moratorium on proof-of-work mining, signed into law in November 2022, was the most aggressive intervention, effectively banning new mining operations that use fossil fuel power plants.

The Rothbardian response to Bitcoin mining restrictions is the same as the response to data center restrictions: property owners should be free to mine Bitcoin on their land, subject to the same nuisance and trespass constraints that apply to any other land use. If your mining operation's noise crosses my property line, I have a claim. If it does not, you are free to mine.

But the deeper connection is about monetary sovereignty. Data centers are the physical infrastructure of the digital economy. The companies building them are primarily serving fiat-denominated advertising, cloud computing, and AI businesses. The incentives, subsidies, and political battles surrounding their construction are symptoms of an economy in which government sets the terms of resource allocation.

Bitcoin offers an alternative. A monetary system based on proof of work does not need subsidies to function. It does not need tax exemptions. It does not need zoning variances or utility commission approvals. Miners locate where energy is cheapest because the protocol demands it. The price signal is clean and unambiguous. There is no room for political favoritism. In this narrow but important sense, Bitcoin mining is what data center construction would look like if the government got out of the way.

The Property Rights Gap

The reason property rights alone have not settled the data center question is not that the framework is flawed. It is that property rights in the relevant resources, land, water, airspace, electromagnetic spectrum, are poorly defined, unevenly enforced, and frequently overridden by political action.

Consider noise. In most jurisdictions, noise ordinances set arbitrary decibel limits rather than recognizing a property owner's right to quiet enjoyment as a function of pre-existing conditions. A data center that moves into a quiet rural area and generates 65 decibels at the property line may comply with the local ordinance but still materially degrade a neighbor's property. Under a strict Rothbardian framework, the neighbor would have a claim. Under current law, the neighbor may have no recourse.

Consider water rights. In much of the western United States, water rights are tied to "beneficial use" requirements that were designed for agriculture, not for industrial cooling. A farmer who sells his water rights to a data center operator may find the sale challenged by a state water board that does not consider data center cooling a "beneficial use." The property right exists on paper but is effectively conditional on government approval.

These gaps are not arguments against property rights. They are arguments for better-defined, more rigorously enforced property rights. The Rothbardian position is not that the current system works. It is that the current system fails precisely because it substituted political discretion for property rights, and the solution is to reverse the substitution.

What to Watch

Three developments will determine whether the data center siting debate moves toward property rights or deeper into political management.

First, watch the state-level subsidy competition. If states continue to escalate tax incentives, the misallocation of resources will worsen. The first state to eliminate data center subsidies entirely and rely on market pricing will be worth studying. Indiana and Ohio, which have attracted major investments despite offering less generous packages than Virginia, may be early indicators.

Second, watch water litigation in the western states. As data center water consumption rises, property rights claims over shared aquifers will reach the courts. The outcomes will reveal whether judges are willing to enforce prior appropriation principles against politically connected agricultural interests, or whether water allocation remains a political question.

Third, watch Bitcoin mining's regulatory trajectory as a leading indicator. Mining operations have already tested every argument that data center operators will eventually face: noise, energy consumption, community character, environmental impact. The legal precedents set in mining cases, particularly in Texas, where miners have integrated into the ERCOT grid as flexible load resources, will shape the data center debate. If mining can coexist with grid operators through voluntary market arrangements, the case for political management of data center siting weakens further.

The Rothbardian answer to the data center question is the same answer libertarians give to most resource allocation questions: define the property rights, enforce them through courts, and let the market sort the rest. The reason this answer feels radical in 2026 is not that it is untested. It is that a century of zoning, utility regulation, and tax incentives has made people forget it was ever the default.

---

Source: Mises Institute