When a large project is proposed in a community, the conversation about tax incentives is rarely the first one and is almost never the easiest one. Residents hear that a developer will pay less in some category of tax than another taxpayer would, and the question that follows is reasonable: if this project is so valuable, why does it need a tax break, and who is paying for it?
The answer is neither “incentives are always good” nor “incentives are always bad.” It is closer to: the structure of the incentive matters more than the existence of the incentive. The communities that have come out of large-development decisions with something durable to show for it have been the ones that asked structural questions early, demanded specific terms in writing, and held the developer to what was agreed. The communities that have ended up with the worst outcomes have been the ones that took numbers in press releases as commitments and only learned the difference later.
This post is a primer. It walks through what a tax incentive is, why it shows up in large industrial development, what separates a well-structured incentive from a poorly structured one, and two specific cases (one positive, one cautionary) that show what each looks like in practice.
What a tax incentive is
A tax incentive is a public-policy decision to reduce a defined category of tax liability for a defined kind of taxpayer for a defined period, in exchange for some level of public benefit (employment, investment, infrastructure, or all three). It is not a subsidy in the everyday sense of the word. The community is not writing a check. The community is forgoing some category of revenue that, in many cases, would not have existed without the project being there.
Several specific tools fall under the broad heading of “incentive”:
- A sales and use tax exemption removes the sales tax that would otherwise apply to the equipment a project buys. Virginia’s data center sales and use tax exemption is the cleanest example: it has been in place since 2010, sunsets in 2035, and requires qualifying projects to invest at least $150 million and create at least 50 jobs to qualify, per the Virginia Department of Taxation.
- An ad valorem (property) tax abatement reduces or defers the local property tax owed on real or personal property associated with a new investment, typically for a defined period (5 to 30 years, depending on the state and the program). Texas Chapter 312 was the most-used version of this for industrial projects until its 2022 sunset; Oklahoma’s five-year ad valorem exemption for qualifying manufacturing and data center investments is documented by the Oklahoma Policy Institute.
- A payment in lieu of taxes (PILOT) is a negotiated alternative to standard property taxation in which the project pays a fixed schedule of payments to local taxing districts.
- A tax increment financing (TIF) district captures the increase in property tax revenue generated by new development in a defined area and uses that revenue stream to fund infrastructure or other capital investment for the project.
- A performance grant or refund program rebates a portion of state-level tax revenue (income tax, withholding, or sales tax) tied to documented job creation or investment milestones.
Most large industrial projects use a combination of two or three of these tools, layered, with different sunsets and different administrators (state, county, school district, special-purpose district). That structural complexity is part of the reason these conversations are hard to have in a community meeting.
What makes an incentive work, and what makes one fail
A well-structured incentive answers four questions:
- What is being exempted, and for how long? The tighter the category and the shorter the window, the cleaner the deal. A five-year ad valorem exemption on equipment is a more bounded commitment than a thirty-year PILOT covering all real and personal property.
- What does the public get in return, and is it enforceable? A jobs threshold without a clawback is an aspiration, not a commitment. A clawback without a verification mechanism is a paragraph in a contract no one is going to enforce. A real public-benefit threshold is one where the agreement specifies how performance is measured, who measures it, what happens if the project misses, and how the community recovers what it gave up.
- Who pays for the project’s infrastructure? Substations, transmission upgrades, water and sewer improvements, road improvements: in a well-structured deal, these are funded by the developer. In a poorly structured deal, the public funds them through TIF or general obligation debt, and the public bears the risk if the project fails to materialize at scale.
- Is the net fiscal picture transparent and verifiable? A community evaluating an incentive should be able to see the foregone revenue figure, the projected new revenue figure, the time horizon, and the assumptions behind both. If those numbers are not in writing, they are not real.
A positive case: Mayes County, Oklahoma
In 2007, a hyperscale data center campus was announced for MidAmerica Industrial Park, just outside Pryor in Mayes County. The project came in under Oklahoma’s standard five-year ad valorem exemption, with the state’s Ad Valorem Reimbursement Fund partially offsetting the local revenue loss during the exemption window.
The structural mechanic that made this case work for the community was Oklahoma’s State Aid Formula. When local property tax revenue per student rises above a defined threshold, the school district stops receiving state equalization aid and is treated as locally self-funded. Under that mechanic, the host school district keeps every additional dollar of local property tax revenue rather than seeing it offset by reduced state aid.
In a 2025 NewsOn6 interview, the Pryor Public Schools superintendent described the trajectory: the district’s net assessed value grew from $80 million in 2007 to approximately $1 billion today. The district has been off the formula since the data center campus came online, which means the local tax gain has been retained in full and is funding teacher compensation, facilities, and capital investment at the local level.
At the state level, Mayes County alone accounts for roughly 42 percent of Oklahoma’s $93 million Ad Valorem Reimbursement Fund disbursement, per the Oklahoma Tax Commission’s 2025 Ad Valorem Annual Report. That is a concentration figure worth being honest about. The Mayes case is favorable because the structure was favorable, not because it scales identically to every community.
A scale comparison: Loudoun County, Virginia
For a sense of what a long-running, well-structured incentive program looks like at scale, Loudoun County’s data center cluster is the cleanest national example.
According to Loudoun County’s official FAQ, data centers occupy approximately four percent of the county’s commercial parcels but generate 38 percent of the county’s general fund revenue and roughly half of all property tax collections. Total property tax revenue from data centers is projected at approximately $1.3 billion in the FY2026 budget cycle. The revenue from this category has allowed Loudoun to reduce the residential real property tax rate every year for the past decade, eliminate the $25 vehicle license fee, and propose a vehicle personal property tax rate reduction beginning in tax year 2026. The current Loudoun residential rate is the lowest in northern Virginia.
Statewide, data centers generated approximately $40 billion in total economic impact and supported more than 112,000 jobs in Virginia, per the NVTC / Mangum Economics 2025 report.
The honest framing is that none of this is automatic. Loudoun’s outcome reflects two decades of incentive-program stability, an enforceable structure, and county-level investment in the infrastructure (power, water, fiber) that allowed the cluster to grow. It is a result, not a default.
A cautionary case: Foxconn in Mount Pleasant, Wisconsin
The cleanest example of an incentive structure that produced a poor outcome for a community is the Foxconn deal in Mount Pleasant.
In 2017, Wisconsin announced what was then the largest economic development incentive deal in U.S. history: up to $3 billion in state subsidies (with additional local TIF spending) for a Foxconn manufacturing campus that was projected to deliver $10 billion in investment and 13,000 jobs. As of mid-2024, the Wisconsin Economic Development Corporation reported that 768 jobs had been created at the site. The deal was renegotiated in 2021 to a maximum of $80 million in state subsidies tied to a 1,454-job target.
The state-level subsidy structure in this case held up reasonably well, because Wisconsin’s incentives were performance-based and were ultimately scaled to actual delivery. The state did not write a check it could not recover. The structural failure was at the local level. Mount Pleasant and Racine County had already issued more than $700 million in TIF-backed debt for site preparation, water and wastewater infrastructure, and road improvements before the project was scaled down. As Strong Towns documented, that local debt is being repaid by local taxpayers regardless of what gets built on the site.
The structural lessons:
The state-level deal was structured around delivery and was ultimately bounded by what the company built. That part worked.
The local-level deal was structured around the project’s promised scale. The infrastructure was built for a 22-million-square-foot campus. The local debt that funded it does not scale down when the project does. That part did not work.
The takeaway is not that incentives are bad. It is that incentives that put the public on the hook for capital costs in advance of delivery carry risks that must be priced in.
What this means for a community evaluating a proposal
The four questions above are the practical version of the takeaway. A community looking at an incentive proposal should ask: what specifically is being exempted, what is the public-benefit threshold and how is it enforced, who pays for the infrastructure the project requires, and are the foregone revenue and projected new revenue figures both in writing.
When the answers are clean, the deal can produce a Mayes or a Loudoun outcome over the operational life of a project. When the answers are not clean, the deal can produce a Mount Pleasant outcome regardless of the company’s intentions.
The structure is the question. The answer follows from there.
Sources cited above: Virginia Department of Taxation, Data Center Retail Sales and Use Tax Exemption; Oklahoma Policy Institute, Ad Valorem Manufacturing Exemption; Tulsa World, “Google investment in Pryor reaches $3 billion”; NewsOn6, “Pryor Public Schools talks impact of Google’s data center”; Oklahoma Tax Commission, 2025 Ad Valorem Annual Report; Loudoun County Data Centers official page; Loudoun County FAQ on resident impact; NVTC / Mangum Economics, “The Impact of Data Centers on Virginia’s State and Local Economies,” 6th Biennial Report; Wisconsin Public Radio, “Foxconn qualifies for second round of state subsidies”; Strong Towns, “Wisconsin Foxconn Deal Cost Taxpayers Millions”; Good Jobs First Subsidy Tracker, Foxconn.