Preventing Hyperscale Harm:
Analyzing the Potential Impact of Opportunity Zones 2.0 in Rural Communities Experiencing Accelerating AI Data Center Construction
August 2026
Simon Wang, Economic Mobility Specialist
Bruce C. Mitchell, PhD, Principal Researcher
Jad Edlebi, GIS Data Engineer
Interactive Map Application
The Hyperscale Data Center interactive map application shows rural low-income communities eligible to become Qualified Opportunity Zones under the new OZ program.
Executive Summary
Data centers are the backbone of the digital economy. Since the rise of the internet, the United States has become the data center capital of the world, housing nearly
40% of the world’s data centers to date. Before recent breakthroughs in large language models (LLMs) shifted the landscape of artificial intelligence (AI) investment, data centers operated relatively innocuously, often housing institutional data and servicing basic cloud computing needs.
However, the demands of newer AI models have required historic amounts of investment in larger, more resource-intensive data centers. In general, more data centers of all kinds are needed to support the wider integration of AI into the digital economy. While hyperscale data centers are increasingly viewed as critical infrastructure for the digital economy, their rapid expansion places considerable demands on local energy grids, water supplies, land use and public services, generating growing concerns about their effects on host communities.
The Growth and Acceleration of AI Data Center Construction
By the end of this year, an estimated $500 to $700 billion will have been put toward data center development, with a stamp of approval from the federal government to move forward as quickly as possible. In an increasingly stagnant economy suffering
from diminished consumer spending as well as slowing job growth, data centers
and capital expenditure commitments by the so-called “Magnificent 7” Big Tech conglomerates (Apple, Microsoft, Amazon, Google, Meta, Tesla and NVIDIA) have become the engine of the economy. Some estimates suggest spending on data centers contributed as much as 92% of gross domestic product (GDP) growth during the first half of 2025.
While the financial markets continue to show confidence in AI, people on the ground have shown growing opposition to the construction of data centers and apprehension about the broader economic and societal impacts of AI. Poll after poll demonstrates Americans strongly oppose data center development in their communities. The most recent Gallup survey from March 2026 found that seven in 10 Americans oppose constructing AI data centers in their local areas, including 48% who strongly opposed it. Although some question the breadth of data center opposition, the consensus from most polling suggests that Americans across the political spectrum oppose data center development.
The billions of dollars in AI data center investment are placing a physical toll on communities, especially in rural areas. As data center proposals show up across the country, local residents are learning in real time just how powerful OpenAI, Anthropic and the rest of the Big Tech companies really are as well as what the costs of AI are to their land, resources, health and ways of life.
Structural Factors Pushing AI Data Centers to Rural Communities
As AI models have changed, so too has the physical footprint of the data centers
powering them. Recent estimates suggest that the overwhelming majority of new data center projects in the US are planned to be constructed in rural areas, even though most of those currently operating are located in cities. The shift in the geography of data center projects is largely a byproduct of alignment between the needs of the latest AI technology and the advantages of developing newer, larger AI data centers in rural areas of the country.
However, all data centers are not created equal. As frontier AI labs have doubled
down on models that require enormous amounts of data to train and operate, the size and scale of data centers have grown astronomically. These so-called hyperscale data centers and mega-campuses (henceforth referred to as AI data centers) are built specifically for AI and have completely altered the landscape of data center construction across the board, expanding quickly into communities that had largely been left behind in the digital economy until recently.
The move to hyperscale data centers is a marked shift from the wave of data centers built during the rise of the internet and cloud computing. This previous cohort of data centers was built primarily to optimize latency (the time it takes for data to pass from one point on a network to another), which meant building near urban communities with high concentrations of end users or suburban areas with access to high-speed fiber optic cables (e.g., “Data Center Alley” in Northern Virginia).
Figure 1, shown below, shows the dense concentration of data centers found along power lines in Northern Virginia. Access to government users and close proximity to the submarine transatlantic cables that run along the ocean floor made Northern Virginia an ideal location for data center construction.
Figure 1. “Data Center Alley” outside Washington D.C. in Ashburn, VA. Large circles are hyperscale data centers. Orange lines are high voltage power transmission lines.
These data centers were much smaller, housed fewer servers and, by extension, required less resources and introduced fewer adverse effects to the surrounding environment. In fact, most data centers operated relatively innocuously, sometimes housed in downtown buildings nearly indistinguishable from office parks. However, in recent decades, the rise of cloud services like Google Drive and Microsoft SharePoint in addition to growing interest in big data increased interest in larger data centers.
Throughout the 2010s, hyperscale data centers began to emerge around the world to meet this demand for storing and processing data. Although demand for data centers was already rising steadily, the emergence of generative AI has led to a dramatic expansion. A recent McKinsey analysis estimates around 70% of the growing demand is for data centers built specifically to support AI.
To build AI data centers, vast amounts of natural resources (e.g. land, energy, water) are needed in quantities that are magnitudes greater than the data centers primarily used for other purposes. The availability and price of these resources, especially land, in rural areas have made them particularly attractive to companies involved in the construction, management and use of AI data centers. As a result, rural communities are increasingly finding AI data center developers at their doorstep.
There are five key resource and infrastructure requirements that are required to develop an AI data center: land, energy, water, electricity grid and fiber optic network infrastructure. Rural areas vary in the cost and availability of these inputs, but generally they offer more favorable environments for AI data center construction and operation.
Land
Hyperscale data centers have historically been defined as taking up at least 10,000 square feet for the actual facility. However, more recently, AI data center developers have been buying thousands of acres of land for facilities spanning millions of square feet. Some of the cheapest available raw land is found in large, rural states with low population density and high land supply.
Energy
The equipment housed in a hyperscale data center requires vast amounts of energy to operate. Estimates now suggest that data centers could make up roughly 1% of global power demand and 11% of total domestic electricity consumption by 2030 – a share projected to grow substantially as AI workloads expand.
Water
The heat generated by hyperscale facilities place substantial demands on local water used in cooling systems, with some large facilities consuming millions of gallons yearly. While newer direct liquid cooling, air cooling technologies and closed-loop systems are beginning to reduce this burden, these improvements in water use depend on system design and local climate conditions.
Electricity Grid
AI data centers place growing stress on an already aging nationwide electrical grid, often requiring major improvements to infrastructure in order to gain access to electricity at the scale and frequency needed to support AI development.
Fiber Optic Network Infrastructure
Data centers must be located within a reasonable distance of the fiber optic backbone that spans the country because latency and bandwidth constraints make rural remote siting viable only where it can be accessed or extended economically. Unlike previous eras of data centers focused on latency, distance to internet infrastructure is a secondary factor to the electricity grid when it comes to AI data centers.
Tradeoffs of Rural AI Centers
Demographic Access and Exclusion
As rural communities face growing interest from AI data center developers, disinvestment and demographic shifts away from rural areas have shaped how communities and elected officials have responded. Rural areas of the country have been losing population density for decades, putting local governments under significant financial pressure as their tax bases have declined. Even in communities that recognize the risks and negative effects of AI data centers, many have been overlooked by other, perhaps more desirable, economic
development projects. As a result, elected officials are left without an alternative.
Based on the lack of interest from other developers, many states and local governments have embraced AI data centers as the key to restoring tax revenue in rural areas. Some even introduced tax incentives and abatements to compete with other regions and attract data center developers to their communities. In return for subsidizing development, local officials are often promised jobs, tax revenue and other community development initiatives.
However, the costs of an AI data center are not just financial. AI data centers also introduce a multitude of externalities that are not accounted for on the balance sheet, such as straining energy supply and electricity grids, contaminating the local water supply and polluting the air. Ultimately, the net impact of data centers is difficult to forecast, especially without transparency standards, updated zoning codes and environmental impact assessments.
The demographic shifts and resulting financial challenges facing rural communities are yet another contributor to the acceleration of AI data center development. In recent years, grassroots opposition has mounted as people from across the political spectrum are coalescing to reject data center builds.
Stuck between fervent local opposition, pressure from developers to move quickly and the tradeoffs of allowing construction to move forward, rural policymakers have been put in an impossible position. Either circumvent constituents and accept the costs of constructing an AI data center or risk jeopardizing the financial security of their community. All of this has turned AI data centers into one of the most polarizing economic development projects in the country.
Federal Tax Policy May Further Incentivize Rural AI Data Centers
Recent changes to a federal community development initiative called the Opportunity Zones (OZs) program could provide additional economic incentives for AI companies to develop hyperscale data centers in rural areas of the country. The OZ program was first created through the Tax Cuts and Jobs Act of 2017 as a place-based economic development initiative designed to channel private capital into low-income communities.
By offering investors preferential tax treatment on capital gains, policymakers hoped to stimulate business activity, job creation and long-term community development in economically distressed census tracts.
By 2026, it is estimated that investors had directed roughly $100 billion nationwide into economic development projects through the OZ program. However, the effectiveness of that money in uplifting disadvantaged communities has been widely debated. Although some argue the OZ program has attracted capital to communities that had long struggled to secure investments, further analysis reveals the hollow nature of those contributions. Estimates suggest a substantial share of the funds invested through the first OZ program ended up flowing into projects that would have happened even without the tax incentive.
Some analyses have found that many of the projects delivered minimal direct benefits to residents of the low-income communities it was designed to serve. Rather than helping the persistently distressed communities that the OZ program was purportedly created to serve, the investments that did materialize were disproportionately put toward constructing market-rate real estate in the most well-off enclaves of the zone.
By allowing investors to lift and defer taxes on long-term investments, OZs reward economic development projects that are likely to increase in value and generate capital gains over time, such as market-rate housing or AI data centers. As a result, investments made through the OZ program often ended up going to areas with the least need rather than being put toward job-creating businesses in distressed communities.
Despite these concerns, Congress renewed and expanded the program in 2025, making OZs a permanent feature of the federal economic development landscape and creating a new facet of the program called the Qualified Rural Opportunity Fund to focus specifically on driving investment in rural communities. In addition to questions about whether this action will offer tangible benefits to rural communities and not just wealthy investors, the race to develop AI data centers across rural America introduces another set of concerns about whether the projects that rural OZs may attract will translate to meaningful impact.
AI data centers are exactly the kind of capital-intensive, long-duration, asset-heavy investment that OZ-style incentives are structured to reward. Rural areas designated as OZs that offer comparatively cheap land costs, access to affordable energy and proximity to fiber optic infrastructure present an attractive investment profile for AI companies seeking to maximize the tax advantages of long-term asset appreciation.
While this situation might be beneficial for investors, constructing an AI data center imposes many costs on residents without many clear direct benefits beyond some immediate relief to the local tax base and perhaps a handful of long-term jobs. Just as market-rate housing in wealthier metropolitan neighborhoods became the go-to project for OZ investors, AI data centers could repeat this pattern in rural OZs, albeit at a much greater scale of investment and with far greater negative consequences.
Identifying Rural Communities Impacted by Data Center Development in Opportunity Zones
To anticipate which rural communities could become targets for AI data center development, we used ArcGIS to create a map of the United States at the census tract level. Next, we constructed an AI data center siting index using the characteristics identified. The AI data center siting index was constructed using the locations of existing data centers to identify the factors that best predict where future data centers are likely to be located.
These factors include the density of the electrical power grid, their proximity to internet exchange points and the high-speed fiber-optic backbone. Locational data came from the FracTracker Alliance, who publishes a publicly available dataset of more than 1,600 data centers. The map below (Figure 2) shows in darker colors areas with a greater risk of development. Next, the results of the index were overlaid on a map of all low-income census tracts identified in the 2024 Census data. Finally, we plotted the location of the data centers using the data from FracTracker Alliance.
Hyperscale Data Center Interactive Map
Overall, 14% of all data centers were located within OZs. However, the concentration was notably higher among approved, permitted and under-construction facilities, with 17.3% being located in OZs. Given the share of all Census tracts designated as OZs, this rate is 46% higher than expected. Operating facilities were also overrepresented, while cancelled projects exhibited the highest concentration, with nearly one in five located in an OZ.
State and Local Policy Recommendations
OZs should be used to promote real economic development with real benefits for communities rather than jeopardizing community health, safety and autonomy in return for modest economic benefits. Moreover, rural governments should not feel forced to accept projects that their constituents are against. To avoid repeating the mistakes of the first round OZ program, there are two complementary courses of action that state and local governments impacted by rural OZ designations could take to preserve community autonomy and long-term interests.
Incentivize Community-Aligned AI Data Center Alternatives
First and foremost, governors must listen to community groups (vs. industry) and nominate LMI census tracts that have high potential for OZ investment to benefit their communities. Then, local and state governments representing rural communities should pair the OZ designation with additional strategies for attracting community-aligned projects.
That begins with rigorous planning that empowers residents to build a vision for their community’s future. Local governments should ensure they have strong, ambitious plans and goals co-created alongside community members. The resulting plans and objectives offer a roadmap for policymakers and community members that represents their shared priorities. In fact, many communities already engage in this process.
These may include but are not limited to implementing the following:
- A water usage plan,
- An energy usage plan,
- A heat mitigation plan,
- An economic development plan,
- Clear environmental standards (such as air pollution and emissions
reduction guidelines), - Concrete environmental justice goals,
- Public health goals, and
- A city equity plan.
Using these plans, local officials should first identify the kinds of economic development projects that match residents’ priorities. Then, they should encourage the creation of those projects by stacking additional incentives onto census tracts designated as rural OZs. Doing so will help to direct Qualified Rural Opportunity Fund investments into projects that will more positively impact the community, cause fewer negative effects and align better with residents’ needs.
At the local level, zoning ordinances offer policy levers for both preventing AI data center construction and further incentivizing other projects. For example, municipalities can change the zoning code to restrict data centers from being developed in OZ-designated census tracts. In Atlanta, Georgia, the city council amended their zoning code to prohibit data centers in the Beltline Overlay District, an initiative to link green space, trails, transit and economic development. At the same time, local governments can make it easier to build projects that better align with community priorities. They could, for example, reduce permitting requirements or shorten review periods for affordable housing, local businesses or other preferred developments.
At the state level, lawmakers can pass state‑specific OZ tax credits or subsidies for other kinds of economic development projects that have stronger evidence of delivering long-term community benefits. Stacking tax credits for certain kinds of projects adds another layer of incentives for investing in projects that advance the community’s goals (e.g. state-specific Low-Income Housing Tax Credit programs).
States can also pave the way for residents to participate as investors in their own communities. Offering state tax incentives for state resident investments in local businesses and incentivizing community investment funds allow ordinary investors to participate in OZs and guide OZ investments to locally led projects. Passing state-level Community Reinvestment Acts can also increase the amount of investment from financial institutions that is available to communities in these areas. Overall, this place-based, people-led approach helps to align economic incentives with what the community wants and needs, encouraging more democratic development.
Condition Data Center Permits on Alignment and Commitment to the Community
More broadly, state and local governments should set policies that establish a baseline set of guardrails that ensure data center proposals are introduced transparently and with the support of their communities. Given the power imbalances between Big Tech companies and local governments, AI data centers offer little possibility for parity in shared agreements.
Not every community will have what it takes to organize a successful negotiation with AI companies and developers. Under voluntary frameworks, every shared agreement requires strong organizing campaigns and local leadership to negotiate with a Big Tech company. If the company breaches their agreement, then local governments are also unlikely to have the resources needed to withstand long legal fights.
AI data centers should fit into the existing community planning goals mentioned earlier. Local governments can achieve this by establishing a conditional use permitting process for data centers (e.g. Limerick Township, Pennsylvania). Conditional use permits require developers to apply to the zoning board for permission to use the land for specific purposes. Developers must meet specific conditions set by the local government before they receive approval, with their permits being revoked if conditions are not met throughout the lifespan of the property. Local governments can set rules for what kinds of projects require approval and what is required to be approved for a permit. Approval for AI data centers should be conditioned on adherence to transparency measures as well as alignment with community goals.
The approval process for data center projects should reject applications that fail to comply with these expectations. The burden should be on the data center developer to prove how the project will fit within a community’s priorities, with clear and transparent timelines, plans and accountability measures. Failure to effectively demonstrate this should result in rejection. Moreover, successful applications should be accompanied by a requirement that the developer complete a legally enforceable community benefits agreement (CBA) with residents.
One of the key goals of a CBA is to secure real commitments for community members from data center developers before construction proceeds. They have long been used in a variety of large-scale economic development projects, including professional sports stadiums, clean energy projects and bank investments. In recent years, several localities have signed CBAs with data center developers. In Lancaster, Pennsylvania, a CBA secured a water use cap, emissions controls and a $20 million community fund. In St. Louis, Missouri, the city negotiated binding commitments on noise, sustainability and water use, retaining the right to seek legal recourse if provisions go unmet.
Growing resident concerns about data centers are prompting a range of state and local policy responses, from moratoriums to broader debates over how proposals should be evaluated through permitting and zoning processes. CBAs are increasingly seen as an important piece of that puzzle (e.g., the NAACP CBA framework and the Federation of American Scientists’ CBA framework). By tying the permitting process to developers’ commitment to community priorities by, for example, conditioning approval on completion of a CBA, local governments can ensure that data center projects benefit the community and have buy-in from residents. When paired with broader safeguards, CBAs are one mechanism policymakers and communities may consider for increasing community participation in decisions about whether and how data centers are built.
Conclusion
Advances in AI technology and the expansion of big data in the modern digital economy have vastly increased the demand for large, hyperscale data centers over the past decade. In recent years, AI companies in particular have expanded investment, increasing the pace and scale of their push to construct AI data centers. The scale and intensity of resources needed to build and operate these AI data centers has shifted development to rural areas of the country, which have often been overlooked by other economic development projects.
However, current evidence suggests that the long-term economic impact of AI data centers is limited. Although they can provide much-needed support for the local tax base, AI data centers create very few jobs beyond the construction phase and impose a host of negative consequences on the surrounding environment with significant effects on public health and community life. As grassroots opposition grows, local officials are increasingly squeezed between representing the views of their constituents and protecting the long-term financial security of their communities.
The creation of Rural Opportunity Funds further complicates the picture by providing additional federal tax incentives for large, capital-intensive projects like AI data centers. That said, local and state governments representing rural communities that are designated as OZs can take actions that incentivize Rural Opportunity Fund investments to be used for more beneficial economic development projects than AI data centers.
Moreover, they can introduce formal processes and mechanisms for ensuring that AI data center projects are approved transparently with community input, align with broader community planning goals and remain accountable via community benefits over the entire course of the project. Communities that are successful at aligning economic development policy with planning goals co-created by residents and policymakers can ensure Rural Opportunity Funds are leveraged for meaningful, long-term community impact rather than as an accelerant for the AI data center buildout.
TABLE OF CONTENTS
Subscribe
Get NCRC news and
alerts by email.
