The National Bureau of Economic Research dropped Working Paper 35194 this May, and it's required reading for anyone tracking how AI infrastructure is rewiring the American economy at ground level. The paper—titled 'Data Centers and Local Economies in the Age of AI'—doesn't just catalog where these facilities are popping up. It quantifies what happens to the counties that host them.

Methodology: Getting Precise on Location Effects

The researchers went deep on causal identification, assembling a facility-level panel with precise coordinates and annualized revenue data for global data centers. They mapped these to U.S. counties and cross-referenced County Business Patterns, county-level IRS income figures, house prices, and electricity prices. To handle the endogenous siting problem—why do data centers end up where they do?—they deployed two shift-share instrumental variables leveraging pre-existing proximity to InterTubes long-haul fiber nodes and 1980 county shares of U.S. urban college population as shares, with Chinese and rest-of-world data center revenue growth as shifts.

The Employment Multiplier Effect

The IV estimates show measurable positive effects across multiple dimensions: total employment, data-processing employment specifically, construction employment, number of establishments, house prices, and electricity prices all moved upward at various horizons after data center growth. Beyond the direct job creation, researchers found positive spillovers on tax returns filed, adjusted gross income, and wages. Annual payroll showed a response too, though it was less robust than other metrics.

The Housing Market Wildcard

Here's where things get interesting for communities considering these investments: house prices rose in counties receiving data centers. That's great news if you're a property owner, but it's a double-edged sword for renters and younger residents trying to stay in their hometowns. The study suggests that as capital investment floods into an area, it creates winners among existing asset holders while potentially pricing out some of the workers the facilities employ.

Electricity Markets Feel the Strain

The paper's findings on electricity prices are particularly noteworthy. Data centers are voracious consumers of power, and the research confirms they're putting upward pressure on local grid costs. For counties already struggling with energy infrastructure or trying to attract manufacturing, this is a real consideration. The 'clean' image of tech jobs comes with some not-so-clean implications for utility bills.

Key Takeaways

  • Data centers create measurable employment gains across multiple sectors, including construction and data-processing roles
  • House prices rise in host counties, benefiting property owners but potentially squeezing renters
  • Electricity prices increase as demand from facilities strains local grids
  • Tax returns, adjusted gross income, and wages all show positive responses to data center growth

The Bottom Line

This NBER paper should be ammunition for community advocates negotiating with tech companies. The benefits are real—jobs, tax base, economic activity—but so are the costs. Counties need to be asking hard questions about affordable housing requirements, utility infrastructure funding, and what happens when the AI bubble eventually deflates. The researchers have given us the numbers. Now it's up to local policymakers to decide whether the tradeoff works for their constituents.