U.S. Data Center Boom Fuels Local Growth and Higher Power Prices

Rising electricity demand, increasing water usage, and expensive subsidies for companies like Google and Amazon are often linked with the DC boom. Due to efficiency gains, DCs in the US have fulfilled surging demand without proportional increases in electricity usage. However, these centers generate high environmental costs through water and power consumption.

Systematic empirical evidence is lacking on the impact of DC entry on public finances, local incomes, and jobs. This evidence is crucial in economics and management. While few studies have investigated third-party DC entry strategies, little is known about their broader economic effects.

While DCs in Virginia, the largest DC hub in the US, generate highly productive jobs with robust regional spillovers, they directly employ a small share of the workforce. Meanwhile, DC expansion has sparked local resistance in many areas, with communities in California, Iowa, and Virginia mobilizing against new DC projects, citing resource strain, land-use concerns, and noise.

Thus, studying the AI-driven DCs' economic role, including their impacts on economic growth, geography, regional development, and labor markets, has become critical owing to its increasing importance in shaping the economic future.

The Research Methodology

For the first time, researchers Daniel Yue and Yiyang Zeng of the Scheller College of Business at Georgia Institute of Technology conducted a comprehensive empirical analysis of how DC entry affects local economies. In particular, they studied how DC activation affects building permits, unemployment rates, median household income, establishments, wages, and employment using comprehensive facility-level information from Aterio, combined with county-level outcomes from sources such as the Quarterly Census of Employment and Wages (QCEW).

The Callaway and Sant’Anna difference-in-differences estimator framework was employed in this study to avoid the negative weighting issues plaguing conventional two-way fixed effects designs and to properly account for staggered treatment timing.

The researchers combined U.S. DCs facility-level information with county-level electricity prices and economic activity measures. Aterio, a commercial data provider, compiles the most comprehensive facility-level U.S. DC coverage by aggregating data from satellite imagery, permit records, industry sources, and utility filings.

The dataset reports estimated capacity in megawatts (MW); utility connections; operator name and facility type; current operational status (under construction/canceled /announced/active) and activation date; and precise street-level location (longitude, latitude, county Federal Information Processing Standards code) for every facility.

This granularity allowed researchers to identify DC location, their timing of entry, and scale. They defined treatment based on the first DC activation in a county to obtain a clean date of adoption for the staggered difference-in-differences design. Additionally, they constructed several classification variables to evaluate their hypotheses, including operator type, metro status, and entry cohort.

Yue and Zeng explored whether the first DC facility was a “Big Tech”-operated facility (e.g., Oracle, Apple, Meta, Microsoft, Google, or Amazon) or operated by other operators, using the operator type variable. The entry cohort determined the timing of first DC activation, classified as post-2020, 2010–2019, or pre-2010. Metro status classified the county as non-metropolitan or metropolitan through statistical area delineations based on Office of Management and Budget (OMB) core definitions.

The researchers also constructed additional heterogeneity variables for supplementary analyses. County population measures included population density and total population from the Census Bureau’s Population Estimates Program, while rural-urban continuum codes classified counties based on adjacency to metro areas and metropolitan status on a 1–9 scale.

Key Findings of the Research

The researchers successfully used facility-level data on DC entry combined with county-level economic outcomes and applied modern difference-in-differences methods to study local impacts.

Results showed that DC activation increased the number of business establishments by 4.7%, wages by 5.0%, and employment by 3.5%, indicating statistically robust and meaningful local economic benefits despite limited direct employment.

Yet, these gains were distributed unevenly, with metropolitan counties benefiting significantly while non-metropolitan areas showed negligible effects. Big Tech hyperscale facilities had a greater impact than smaller third-party DCs.

Early first-entry timing counties and those receiving additional DC facilities within five years witnessed stronger gains compared to those with recent first-entry timing and isolated entry counties, although the study notes that weaker effects among more recent entrants may partly reflect shorter follow-up time and that the larger gains linked to additional entries do not necessarily prove that each extra facility independently caused those effects.

Findings also showed that DC entry increased electricity prices in regions where utility structures enable cleaner identification, indicating a hidden cost.

Thus, both facility operators and local context play a key role in determining whether DC investment translates into economic benefits and imposes infrastructure-related cost pressures.

In conclusion, this study's findings effectively inform current policy debates on DC expansion and its local economic consequences.

*Important notice: SSRN publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as definitive, used to guide development decisions, or treated as established information in the field of artificial intelligence research.