The Community Benefits of Data Centers

An Analysis of the Jobs, Tax, Grid, and Community Returns of Modern Data Center Development

The economic, environmental, and community impacts of data centers have fueled some of the hottest election debates of the 2026 cycle. As LLM AI drives the largest wave of infrastructure investment in a generation, cities and counties across the country are being asked to decide whether to welcome the expansion of one of America’s fastest growing industries and a well organized opposition has answered for many of them, arguing that data centers take far more than they give. These groups argue that data centers consume enormous amounts of power, straining fragile local grids and raising community power bills all while bleeding dry local water supplies. The trade off, they claim, are a handful of permanent jobs and hollow tax relief promises.

We are 81 years removed from the birth of the first data center progenitor, and in that time the industry has grown from a U.S. Army project in Pennsylvania to make up almost 3% of the US GDP and use about 4.5% of total US power consumption. This perspective of history allows us to look objectively at how to manage the ongoing growth of one of our most important home grown industries. History allows us to look at real and working towns across the country transformed by their data center partnerships. There are always pros and cons to fostering the development of any industry, and the robust debate on the how’s and why’s is integral to the process, but is meaningless unless we are all debating from the full set of facts. History does tell us of deals made in bad faith, corporations and local governments not being good neighbors, and deals that evaporated after the ground is broken, but if we aren’t also talking about the deals that made boom town, we are doing everyone a disservice.

This Paper is Organized Around Four Central Findings

Data centers convert low value land into a transformational local tax base.

The Eagle Mountain parcel generated roughly $35 per year in property tax before development; it now sends millions annually to the city, Utah County, and the Alpine School District.¹ In Loudoun County, Virginia, the nation’s densest data center market, data centers occupy about 4% of commercial parcels but yield 38% of the county’s general fund, revenue that has let the county cut its residential property tax rate every year for a decade to the lowest in Northern Virginia.²

Under modern “pay-your-own-way” agreements, data centers fund their own grid and can strengthen reliability for everyone.

A new generation of utility tariffs and community agreements requires large loads to pay the full cost of the power infrastructure they use, protecting, not raising, existing customers’ bills, while flexible operation lets the grid absorb far more load than previously assumed.³

The benefits are broad, and enforceable.

Jobs, school funding, small-business growth, workforce pipelines, and water restoration all follow when benefits are secured through binding agreements with every taxing jurisdiction, negotiated from an up-front impact study, as they were in Eagle Mountain.

This is also the argument that resonates with voters.

National polling shows that jobs and pocketbook benefits, carried by local voices and backed by transparency, are the strongest case for data centers, and that requiring public disclosure of a project’s water, energy, and tax terms is nearly free politically, moving far more voters toward support than against.⁴

Why Accurate Information Matters Here

When a community rejects a properly structured data center on the basis of bad actor projects elsewhere, it has not exercised informed self-government and it has been misled about its own options in a time where Americans are seeing higher electrical bills, aging infrastructure, and world wide droughts. Projects are delayed or even cancelled, leaving investment to migrate to other jurisdictions, and the community forgoes the tax base, employment, infrastructure, and grid investment it might have secured. The purpose of this paper is to add to the hard fought benefits of data centers in a community to the record, so that communities can negotiate for the terms Eagle Mountain got, and reject the terms it refused.

How a Data Center Deal Actually Works

To understand why the extraction narrative misfires, it helps to understand what a modern data center agreement actually contains, and what separates a good deal from a bad one. The difference is not the technology, but rather it is the process, and it is entirely within a local government’s control.

The Impact Study Comes First

The single most important step happens before approval. Eagle Mountain’s former mayor, Tom Westmoreland, describes it as the foundation of everything that followed. Commission an independent impact study that identifies every challenge and negative externality a project would create, so that each one can be addressed in the negotiated agreement before residents ever raise it.

Done early, the study becomes leverage. Every impact a community can name (traffic, noise, water, power, public safety, road wear) becomes a term to be mitigated or paid for in the binding agreement, rather than a grievance litigated after the fact. It also gives the public an honest basis for comparison. Westmoreland’s work shows that when a data center’s impact study is compared side-by-side with that of a big-box retailer, the data center produces less crime, less traffic, less noise, and fewer emergency calls, and more high-paying jobs, with higher electricity use the one genuine exception. Yet the retail store draws none of the same scrutiny.

Binding Agreements, Not Press Promises

In Eagle Mountain, the data center entered legally binding agreements with the City, Utah County, the Alpine School District, and the local fire, water, and redevelopment districts, each specifying the taxes to be paid, the infrastructure to be funded, and the environmental protections to be upheld.⁵ That structure is the reason the benefits are real and durable. It is also the model that policy should require everywhere.

Benefit One: Jobs and a Transformational Tax Base

In national message testing, a jobs argument centered on blue-collar, can’t-be-outsourced trades (electricians, HVAC installers, pipefitters) outperformed every other pro-data-center message and was the only one to win majority support across every political subgroup.⁶

The Jobs Are Real, and They Build a Local Economy

Meta’s Eagle Mountain campus, a $1.5 billion investment spanning more than 4.5 million square feet across five operational buildings, with two more under construction, supports over 300 permanent operational positions (electricians, HVAC specialists, server and network technicians, security experts, and engineers) and more than 1,200 skilled-trade workers at peak construction.⁷ Just as important is what the campus catalyzed. When Westmoreland took office, Eagle Mountain had almost no commercial base and could not persuade any business to be the first to invest. After the data centers arrived, others followed and the city now counts more than 900 businesses.⁸

Honesty about the jobs strengthens the case rather than weakening it. Independent analysis, including Virginia’s nonpartisan legislative audit, finds that a single facility employs a modest permanent staff, and that much of the direct employment is concentrated in the construction phase.⁹ The economic returns come primarily from three other channels: high-wage construction employment sustained over years of continuous buildout, a supply chain of local vendors and contractors, and, above all, the tax base. Loudoun County’s economic development director puts the return plainly: for every dollar the county spends servicing a data center, it receives about twenty-six dollars back in revenue, “a figure that no other commercial industry can come close to providing.”¹⁰

From $35 a Year to Millions: The Tax Base Story

The tax transformation is the heart of the community-benefits case. In Eagle Mountain, land that produced $35 in annual property tax now generates millions for the city, county, and school district, even net of the incentives that brought the project.¹¹

Loudoun County shows what that revenue does at scale and over time. Data centers there deliver more than $100 million in new revenue every year and now fund 38% of the county’s general fund while occupying roughly 4% of its commercial land, money that supports public schools and county services and has let the Board of Supervisors cut the residential real property tax rate every year for a decade, to the lowest of any county in Northern Virginia.¹² Virginia’s independent legislative audit put the statewide contribution at roughly 74,000 jobs, $9.1 billion in GDP, and $5.5 billion in labor income annually.

The pattern repeats in communities far smaller than Loudoun. In DeKalb, Illinois, county property-tax records show Meta paid $48.5 million in property tax from 2021 through 2024 and was billed $31.1 million for 2025 alone, with about 61% flowing to the local school district, which used the revenue to help build and open a new elementary school.¹³ In Mecklenburg County, Virginia, fourteen years of tax revenue from a single Microsoft campus helped fund a $154 million renovation of the county’s high school and middle school.¹⁴

Two communities are now testing whether that revenue can be returned directly to residents. In Columbia County, Georgia, county leaders have proposed using revenue from a large Google-anchored development to eliminate homestead property taxes, a plan that depends on a November 2026 voter referendum and later state action, but which reflects the scale of revenue at stake.¹⁵ And in Richland Parish, Louisiana, sales-tax revenue tied to Meta’s data center construction funded roughly $50,000 “thirteenth-check” bonuses for local teachers in 2026, a benefit worth citing honestly as a construction-period surge rather than a permanent entitlement.¹⁶

The model reaches small rural communities as well, and not only the largest hyperscalers. Applied Digital, an AI-focused operator building its Polaris Forge campuses in rural North Dakota, projects that its Ellendale campus alone will contribute roughly $346 million to the state economy and about $850,000 a year in property taxes to a community of a few thousand residents, with more than 200 full-time positions at full build-out.¹⁷ These are projections from the company’s economic-impact study rather than realized results, but they show how a single AI campus can reshape the finances of a town that had almost no industrial tax base before it arrived.

Benefit Two: A Stronger, Better-Funded Power Grid

No criticism of data centers is more potent than the fear that they will overwhelm the grid and push the cost onto ordinary ratepayers. The answer is that the tools to prevent cost-shifting already exist and are being adopted by utility regulators across the country.

Pay Your Own Way: The Ratepayer-Protection Tariff

Eagle Mountain’s agreement establishes the principle directly. Meta worked with Rocky Mountain Power to plan for its energy needs and to pay the full cost of the power it uses, so that residents and other businesses are not made to subsidize it, and the developer built the electrical substation serving the campus rather than drawing down existing capacity.¹⁸

What Eagle Mountain negotiated privately, regulators are now codifying for everyone. In July 2025, Ohio’s utility commission approved a landmark tariff requiring new data centers above 25 megawatts to pay for at least 85% of the capacity they reserve for up to twelve years, regardless of how much they actually use, with exit fees and financial guarantees, so that “the costs of expanding the system are appropriately borne by those customers who are responsible for incurring them.”¹⁹ Georgia, Indiana, and Virginia have adopted comparable rules requiring large loads to cover the upstream generation and transmission they trigger and to commit for fifteen years or more.²⁰

The effect on residential bills is measurable. When Virginia’s State Corporation Commission ordered Dominion to shift transmission costs onto the large loads that drive them, it cut the projected monthly increase for a typical residential customer from $2.90 to $0.94.²¹ At the federal level, a 2026 Ratepayer Protection Pledge, signed by Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI, and later extended across utilities serving roughly 80% of American homes and businesses, commits AI companies to “build, bring, or buy” the power they need and to pay for the delivery infrastructure, so households are not left with the bill.²²

Flexible Load: More Capacity Than We Thought

The premise that new data centers require a vast, immediate buildout of new power plants is itself being over stated. A 2025 Duke University study found that the U.S. grid could absorb roughly 76 gigawatts of new load, and up to 126 gigawatts, simply by having large users curtail consumption during the small number of hours each year when the system is most stressed, on the order of 0.25% to 1% of annual operating time.²³ Because AI training loads can often be paused or shifted, data centers are unusually well-suited to provide exactly this flexibility. Texas has already made it a requirement with its 2025 grid law directing large loads to be curtailable during emergencies and lets the grid operator call on them as a dispatchable resource, turning data centers from a threat to reliability into a tool for it.²⁴

The Honest Counterargument and Why Structure Answers It

Credibility requires acknowledging the other side of the ledger. Independent researchers have documented that where utilities strike opaque, below-cost “special deals,” the difference can be shifted onto ordinary ratepayers, with one Carnegie Mellon analysis projects data-center growth could raise the average U.S. electricity bill by around 8% by 2030, and more in the densest markets.²⁵ That same research also says harm comes from weak or secret cost-allocation, not from data centers by the simple nature of their existence. Where costs are assigned transparently to the customers who cause them, the outcome flips. A National Bureau of Economic Research analysis found that average residential rates actually fell in areas with data-center growth over 2015 to 2024.²⁶

Benefit Three: New Power on the Grid

Some data centers are doing more than paying for the power they use with some of the largest operators financing the construction of new electricity generation that adds to regional supply. Eagle Mountain’s electricity is matched with 100% clean and renewable energy, and Meta has backed 1,719 megawatts of new renewable generation in Utah, an amount its own reporting puts at roughly a third of all solar under contract in the state.²⁷ That includes a 685-megawatt solar field in Utah County, the largest in the state, developed with Rocky Mountain Power and Excelsior Energy Capital and brought online in 2025, and a 122-megawatt project in Iron County.²⁸

This is not unique to Utah. Meta’s campuses have backed 1,120 megawatts of new clean energy in Georgia, 847 in Ohio, 765 in Illinois, and 320 in Nebraska.²⁹ Across the industry, the four largest technology buyers now account for roughly half of all corporate clean-energy procurement worldwide, and new co-location models are pairing data centers with dedicated new generation, from a $20 billion Google-Intersect Power renewable buildout to an $18 billion Amazon nuclear agreement restructured, at federal regulators’ insistence, to remain fully grid-connected and transparent.³⁰

Benefit Four: Schools, Small Business, and Water Stewardship

The final category of benefit is the one residents feel most directly, in their schools, their local businesses, and their watershed. Since 2018, Meta has contributed more than $5 million in direct community giving to Utah County schools and nonprofits through over 240 grants and sponsorships, funding STEM education and workforce programs including a Small Business Launchpad with the local chamber of commerce, support for the Christa McAuliffe Space Center, a hands-on water-conservation STEM program, a community stargazing park built with Utah Valley University, and free digital-skills training for local small businesses.³¹

That civic investment is a consistent feature of well-run campuses, not an Eagle Mountain exception. Meta reports $5.6 million in community giving around its Ohio campus, $4.5 million in Georgia, and $4.1 million in Nebraska, and has launched a $115 million program to train workers for the construction trades that build these facilities.³² Microsoft’s Datacenter Academy, now operating in Southern Virginia and San Antonio, funds community college certificates that place local students directly into skilled data center jobs, the workforce pipeline benefit that outlasts any single facility.³³

The pattern holds for smaller, AI-focused operators too. Applied Digital funds a standalone philanthropic arm, Applied Digital Cares, that makes grants to local nonprofits in the North Dakota communities around its Polaris Forge campuses.³⁴ Its inaugural round in the fall of 2025 awarded $81,399 to five recipients, including $59,499 for the Harwood Fire and Rescue department to buy vehicle-stabilization kits and battery-powered extrication tools, and $11,000 for a local youth robotics association; its spring 2026 cycle awarded $619,595 across 34 community-led projects.³⁵ For a rural fire department or a school robotics team, grants of that size are life changing, and they show that the community-benefit model is not confined to the largest technology companies.

Data Centers are a Net Water Restorer

On water, Eagle Mountain is the strongest possible rebuttal to the extraction narrative. The campus uses roughly 108 acre-feet of water per year, well below initial projections and about 80% more efficiently than the industry standard, with every operational building certified LEED Gold, it pays the full cost of its own water and wastewater service, in addition it reuses treated water to irrigate a city park, a project that saved tens of millions of gallons of potable water in 2024 alone.³⁶ Beyond minimizing use, Meta has committed to restore 200% of the water the campus consumes, funding watershed projects, with the Central Utah Water Conservancy District and Springville Irrigation Company, that return on the order of 900 million gallons per year to the Provo River watershed and protect the endangered June sucker in Utah Lake. A facility critics would cast as a drain on a water-stressed region is, on the record, a net contributor to it.

Newer AI campuses are being designed for near-zero water use from the outset. Applied Digital’s Polaris Forge facilities in North Dakota use a closed-loop, direct-to-chip cooling system that recirculates its coolant and draws on the cold regional climate for free cooling roughly 200 days a year, a design the company reports is engineered for near-zero freshwater use in cooling and a power-usage-effectiveness rating of about 1.18.³⁷ That is the same direction described in this paper’s companion on water, and it is becoming standard practice for new construction.

What the Public Actually Responds To

The benefits are not only real, they are the benefits that the public is looking for in new developments in their communities. The Rainey Center’s July 2026 national survey of 1,017 registered voters tested nine pro-data-center arguments head-to-head and mapped which messengers and benefits matter most to voters.³⁸

Three key findings were found at the core of the polling data. First, jobs win. The blue-collar jobs argument took 65% of all head-to-head matchups, the top result, and was the only message that was net positive with every subgroup, from Republicans to somewhat-liberal voters. Growth (60%), the truth about power and water (57%), and consumer prices (57%) followed. On the opposite side, every national-security and “beat China” frame finished at the bottom and was net negative with every group, including Republicans. To voters, the case for data centers is a pocketbook matter, not a geopolitical one.

Second, local voices carry it. The most effective messengers are unofficial and close to home. A local small business owner (plus 26), a family member (plus 25), a farmer or landowner (plus 25), while every institutional voice, from governors to company employees, is weak.

Third, concrete benefits and transparency both move voters toward support, because, unsurprisingly, voters want to know what they are signing up for by inviting in a new neighbor. The benefits residents respond to most are tangible and personal, reduced electricity costs (61% more likely to support), lower or eliminated property taxes (60%), guaranteed local jobs and infrastructure upgrades (57% each). And requiring a project to publicly disclose its water use, energy use, and any tax breaks makes 47% of voters more likely to support it and only 9% less likely.

Policy Implications: What Policymakers Should Do

The findings in this paper point to a clear agenda for local, state, and federal policymakers. The goal is not to wave projects through or to block them, but to ensure that every data center is structured the way the best ones already are.

Require Impact Studies and Binding Community-Benefit Agreements

Localities should require an independent impact study before approval and should condition approval on a binding agreement that specifies the taxes, jobs, infrastructure, and environmental commitments the project will deliver. This is the Eagle Mountain model, and it is what converts promises into enforceable obligations.

Adopt “Pay-Your-Own-Way” Ratepayer-Protection Tariffs

State utility commissions should adopt cost-causation tariffs, on the model already approved in Ohio, Georgia, Indiana, and Virginia, that require large loads to pay for the generation and transmission they trigger, with long-term commitments and financial assurances, so residential customers are protected. Regulators should favor transparent tariffs over confidential special contracts.

Require Disclosure

Permitting processes should require developers to disclose projected water use, energy use and sourcing, cooling technology, and any tax incentives received. Disclosure is not a barrier to development; polling shows it moves voters decisively toward support, and it lets regulators distinguish best-in-class projects from the rest.

Reward Flexibility and New Generation

Large-load interconnection should prioritize projects that can curtail during grid stress and that add new generation to the system (“build, bring, or buy”), turning data-center demand into an engine for grid investment and reliability rather than a strain on it.

Conclusion

The extraction argument against data centers rests on the projects that were negotiated badly, but the terms that make a data center a genuine community asset are already well understood, already in binding contracts, and already producing results in American towns across the country. Eagle Mountain turned a $35 parcel into a multimillion-dollar tax base, financed its own substation and roads, drew down no one else’s water or power, backed the largest solar project in its state, and restored more water than it consumes, because it did the impact study first and put every commitment in an enforceable agreement.

The relevant question for any community is not whether to allow a large facility. It is whether to require that facility to be structured the way Eagle Mountain’s was, to pay its own way, fund local infrastructure, protect ratepayers, and leave the community better off. That is a question of policy and negotiation, entirely within a community’s power to answer well. The honest answer, drawn from the record rather than the caricature, is that a well-structured data center is not a burden a community absorbs. It is an anchor investment a community secures.