A University of Texas Austin report projects data centers could account for up to 9% of Texas's total water consumption by 2040, driven by expanding AI compute infrastructure. The finding represents one of the more precise resource-use estimates to emerge from academic research on AI's physical footprint.
The report does not specify current baseline consumption figures in the available summary, but the 9% ceiling would place data centers among Texas's largest single-sector water users in a state already managing chronic drought pressure.
Separately, residents near Lake Tahoe have reported power disruptions linked to grid capacity being redirected toward data center loads — a distinct but related signal that compute infrastructure is beginning to compete directly with residential utilities for constrained resources.
Academic projections of this type increasingly feed into state-level regulatory frameworks governing data center siting, permitting, and utility agreements. Texas, which has attracted substantial hyperscaler and AI infrastructure investment partly due to deregulated energy markets, may face mounting pressure to impose water-use reporting requirements or consumption caps.
Operators planning Texas deployments should factor water availability and potential future regulatory constraints into long-term infrastructure planning, particularly for facilities relying on evaporative cooling systems.