What Do Communities Actually Get From a Data Center?
A question worth asking, and a few models worth knowing
A data center is coming to your county. Thousands of acres of land. Hundreds of millions, sometimes billions, of dollars in capital investment. The announcement comes with a press release about economic growth, high-paying jobs, and a tech-forward future.
Then the facility opens. And the jobs don’t quite materialize the way the pitch deck implied.
This dynamic has sparked a growing conversation about whether communities hosting data centers should negotiate a direct share of the value they generate. A few economists have raised the idea. Some city councils have asked the same question. The public finance literature has relevant things to say here, and the honest answer is more complicated than either side usually admits.
The Subsidy Problem Nobody Talks About
Before debating whether communities should get more, it helps to establish what they actually get now.
The standard narrative is that data centers pay taxes, create jobs, and stimulate local economies. That is partially true. But in most states, data centers also receive substantial public subsidies, often before a single server goes online.
At least 37 states offer some form of tax exemption for data centers, most tied to sales taxes on equipment purchases. The scale of these programs has grown dramatically with the AI buildout.
The negotiation is rarely between a neutral community and a tax-paying company. It is often between a subsidizing community and a company that has already extracted the first concession before the ground breaks.
The Jobs Problem
The other pillar of the economic development pitch is employment. And here, data centers face a structural limitation that has nothing to do with bad faith. They are capital-intensive and labor-light by design.
The most automated hyperscale campuses operate with as few as 20 to 30 permanent staff per 100 megawatts of capacity.A facility drawing 500 megawatts might employ fewer permanent workers than a large grocery store.
Anthropic announced a $50 billion data center investment in late 2025. The projected permanent employment: roughly 800 jobs. That works out to about $62 million in capital investment per permanent position. Outside Reno, a 1.1-million-square-foot Vantage Data Centers facility was projected to create just 73 permanent jobs over the next decade, even as more than 4,000 temporary construction positions cycled through the site. Virginia’s statewide data center investments from 2020 to 2025 averaged roughly $54 million per permanent job created.
Construction employment is real and meaningful. But it ends. The community is left with a large, quiet building drawing enormous amounts of power and water, staffed by a handful of technicians, operating under a tax exemption that may not sunset for years.
This is why some economists and local officials have started asking a different question. If jobs are not the mechanism for value transfer, what is?
The Profit-Sharing Idea
The proposal circulating in recent months is some form of profit-sharing or revenue-sharing between data center operators and the communities that host them. The logic is intuitive. Communities provide land, infrastructure, electricity grid access, water, and often direct tax subsidies. The companies capture the value. Should communities receive a structured share?
The idea is easier to state than to implement. The first technical problem is the word “profit.”
Corporate profit at the facility level is highly malleable. Intercompany debt, depreciation schedules, and transfer pricing arrangements can all reduce reported profit substantially without reflecting economic reality. A community negotiating for a cut of “profit” may end up with very little, through entirely legal accounting.
Revenue-sharing is more tractable. It is harder to obscure, easier to audit, and has stronger precedent in extractive industry regulation. Which points toward the more instructive analogs.
Models That Already Exist
The idea of communities receiving a structured dividend from resource extraction is not new. It has existed in multiple forms, at scale, for decades.
The Alaska Permanent Fund is the most direct precedent. Alaska established the fund in 1976 after oil revenues began flowing from the North Slope. Rather than routing all proceeds through the general fund, where they could be spent, shifted, or politicized, the state constitutionally set aside a percentage of oil royalties into a permanent investment account. The earnings are distributed annually as a direct dividend to every qualifying Alaska resident.
As of June 30, 2024, the fund’s total value reached $80.5 billion. The 2024 dividend was $1,702 per resident. Every Alaskan, every child, every retiree, every working adult, receives a check each fall simply for being a resident of a state that decided to treat its natural resources as a public asset.
The fund’s early years are instructive. The dividend was front-loaded, paid out before the fund had fully compounded. As the fund grew, the two lines converged. The structure rewarded patience and shielded the mechanism from short-term political pressure to spend down the principal.
Oil severance taxes follow a related logic. States like Wyoming, North Dakota, and Montana levy taxes on the extraction of oil, gas, and coal, not on profits, but on the act of extraction itself. The resource leaves the ground and does not come back. The tax compensates the public for that permanent depletion.
Mineral royalty structures at the federal level apply the same framework to mining on public lands. Companies pay to extract. The public receives a share.
Data centers extract something too. They draw electricity at scales that strain regional grids, in some cases raising rates for residential customers. They consume water for cooling in regions already under stress. They occupy land and require infrastructure investments that communities bear. The extraction framing is imperfect, but it is not unreasonable.
The Complications Are Real
A fair treatment of this idea requires engaging with the strongest objections.
The first is the permitting leverage problem. Community benefit agreements in California have drawn criticism, including from researchers sympathetic to the underlying goal, for creating discretionary negotiations that become vehicles for rent extraction by well-organized local interests. A standardized, statutory revenue-share mechanism addresses this more cleanly than project-by-project negotiation, because it removes discretion from the table. But designing that mechanism is not simple, and the political economy of passing it through state legislatures that have spent years competing to attract data centers is genuinely difficult.
The second objection, raised by the industry, is competitive pressure. Data center companies report that tax exemptions factor heavily into location decisions, and most states that Virginia or Texas competes with have similar exemptions. Whether this is a strong empirical claim or a negotiating position is an open question. Companies ultimately go where grid capacity and land availability are, and those factors are not easily replicated by tax policy alone. But the argument is not frivolous.
The third complication is distributional. Data centers cluster in particular counties: Loudoun County in Virginia, Maricopa County in Arizona, Travis County in Texas. A statewide revenue-sharing mechanism would spread benefits broadly. A local one might amplify geographic inequality.
An Open Question
None of this resolves neatly. The public finance literature on community benefit structures is real and growing, but the application to data centers is new enough that the evidence base is thin.
What is not thin is the subsidy record. Communities have spent years offering tax concessions to attract facilities that employ relatively few people and draw heavily on shared infrastructure. The conversation about whether a different structure is possible, one with precedent in how this country has managed oil fields, fisheries, and mineral deposits, seems worth having.
The tools exist. The models exist. Whether communities will demand the same deal from data centers that they have long given to oil wells and copper mines is a political question, not an analytical one.