What Does AI Say About the Economic Case for Data Centers?

by Levi A. Russell

Are data centers really reliably driving investment and long-term growth around the communities they inhabit?

May 2026

Industry studies paint a rosy picture in aggregate. According to a 2025 PwC analysis commissioned by the Data Center Coalition, direct employment in the U.S. data center industry grew over 50% from 2017 to 2023, compared to just 10% economy-wide, and the industry’s total employment contribution (direct, indirect, and induced) grew from 2.9 million to 4.7 million jobs. The industry’s total fiscal support to federal, state, and local governments roughly doubled over the same period, reaching $162.7 billion (Data Center Coalition / PwC, 2025). That is real money, and these figures are widely cited by data center developers in permit proceedings and state legislative hearings.

The problem is that national multipliers do not translate neatly into local prosperity. As Brookings Institution researchers noted in February 2026, both developers and communities have largely treated data centers like warehouses — routine projects that produce a surge in short-term construction jobs followed by a steady stream of tax revenue, without many direct contributions to or interactions with the local economy. Economic benefits of a typical large data center decline substantially after the construction phase (Brookings, February 2026).

This is the critical distinction for local policymakers. The construction boom is real but temporary. The permanent direct employment footprint of even a large hyperscale campus is surprisingly small — often measured in dozens to low hundreds of highly skilled workers, not thousands.

Most of the economic case rests on input-output models (IMPLAN, RIMS II) commissioned by industry groups or local boosters. As economist Michael J. Hicks observed in November 2025, published efforts to model the causal effects of data centers on local labor markets or fiscal conditions are conspicuously absent; analysis has mostly been limited to descriptive work or a series of industry-sponsored input-output analyses (Hicks, November 2025). Input-output models are well known to overstate multiplier effects because they do not account for crowding out, factor price changes, or counterfactual scenarios. The absence of rigorous quasi-experimental or causal inference work should give policymakers pause when evaluating developer-sponsored economic impact claims.

The costs borne by host communities are often diffuse, invisible in project-level analyses, and shifted onto residential ratepayers. In the PJM grid region serving 65 million people across 13 states, power supply costs jumped from $2.2 billion to $14.7 billion in a single year, with data centers accounting for nearly two-thirds of the increase. Residential electricity rates nationally rose about 32% between July 2020 and July 2025. Communities near proposed facilities also face noise, strained infrastructure, and the loss of farmland (Brookings, May 2026).

The World Resources Institute reported in February 2026 that over $60 billion worth of rate increases occurred countrywide in 2025 alone, with Americans paying nearly 10% more for electricity on average compared to 2024. Compounding matters, many developers enter non-disclosure agreements with local officials, limiting public access to information about project scale, resource needs, and potential impacts (World Resources Institute, February 2026). The asymmetry of information between well-resourced developers and local governments is itself a structural problem.

Local resistance is no longer a fringe phenomenon. The World Resources Institute found that more than $64 billion in data center projects were delayed or canceled between May 2024 and March 2025 due to organized local opposition (World Resources Institute, February 2026). That scale of project cancellation reflects genuine skepticism from communities weighing the promised tax revenues against rising utility bills, water consumption, noise, and agricultural land conversion.

Data centers do generate real tax revenue and can become among a county’s largest taxpayers. But the long-term local economic development narrative — high-wage spillovers, labor market upgrading, broad community prosperity — is largely unproven by rigorous causal research. The honest assessment is that data centers function more like capital-intensive facilities than engines of broad local growth: modest in permanent direct jobs, meaningful in tax contributions, and increasingly costly in energy and infrastructure demands that often fall disproportionately on residential ratepayers.

Whether that trade-off is a good deal depends heavily on the specific community’s fiscal situation, its existing tax base, and how effectively local officials negotiate the terms of development agreements. Communities considering data center siting would be well-served by requiring independent (non-developer-funded) economic impact analyses, transparent utility cost-allocation plans, and enforceable community benefit agreements before granting approvals.