Who Bears the Costs? Data Centers, Water Governance, and Compounding Local Inequities

• In Event: State Policy, Local Water Systems: Affordability, Conservation, and Infrastructure Finance

Abstract

Data centers are among the fastest-growing sources of industrial water demand in the United States, yet their governance implications for local water systems remain largely unexamined. Public discourse has focused on aggregate consumption figures, but the more consequential question is institutional: which communities bear the costs of rapid hyperscaler growth, and do they have the capacity to negotiate it? This paper argues that data center siting systematically co-locates with communities that face compounding infrastructure disadvantages in water governance and broadband connectivity, raising distinct concerns about affordability, regulatory capacity, and equity. Using a national inventory of 6,586 US data centers (active, under construction, or announced) from the Aterio dataset, I conduct two linked descriptive analyses. First, I examine co-occurrence between data center siting and FCC Broadband Data Collection (BDC) coverage at the county scale. Consistent with Warner and Bravo's (2024) critique of FCC self-reported availability data, I find that even under the BDC's upper-bound estimates, 53.6% of data centers in counties with both urban and rural reporting strata are located in counties where the wireline broadband gap between urban and rural residents exceeds 20 percentage points at the FCC's current 100/20 Mbps benchmark. These 213 counties host an estimated 190,876 MW of combined data center capacity. Second, drawing on EPA's Environmental Protection Community tool (EPIC) and authoritative community water system (CWS) boundary data, I document a tripartite pattern of water governance contexts in Texas, the second-largest data center-hosting state. Approximately 47% of Texas data centers connect to large or very large community water systems with health-violation rates (11%) well below the state baseline (32%). However, 9% connect to small or very small CWSs, systems whose economies of scale per Teodoro (2026) translate directly into higher per-customer costs and lower infrastructure quality, and 35% fall entirely outside mapped CWS service areas, disproportionately in the Texas Panhandle and Permian Basin. Critically, 75% of data centers in the small-CWS cluster are located in counties with urban-rural broadband gaps exceeding 20 percentage points, compared to 42% in the large-CWS cluster, suggesting that the water and broadband disadvantages are not independent. These findings carry direct implications for water affordability policy. When hyperscale demand lands on small or unmapped water systems, the institutional capacity to assess impacts, negotiate agreements, and protect existing ratepayers is most constrained precisely where background conditions of aging infrastructure, compliance burdens, and limited revenue are most challenging. I conclude by discussing the role of state policy variation in shaping whether localities can respond, and what collaborative governance frameworks might better distribute the fiscal and regulatory burdens of industrial water demand.

Author

• Laura Joaquina Morales, University of Pennsylvania
  Presenting Author