NSF AI grant $1.21M: tackling the "water" problem of AI data centers with water-resource digital twins (University of Alabama)
AI infrastructure (data centers) needs large amounts of water for cooling. The NSF awarded about $1.21M to develop three AI models that assess water availability and predict risks to guide siting — using physics-informed AI and digital twins to analyze water sustainability.
Grant overview (primary data)
- Award amount$1,214,308 / Est. total $1,016,594
- RecipientUniversity of Alabama Tuscaloosa（AL）
- ProgramEDUCATIONAL LINKAGES, GEO CI - GEO Cyberinfrastrctre
- Period2025-10-01 〜 2028-09-30
- FunderU.S. National Science Foundation (NSF) / NSF
Key points
- AI data centers need large amounts of cooling water — this project tackles that water-resource challenge head-on
- (1) large-scale AI for water-availability siting; (2) digital twins to reveal supply risks; (3) AI to predict wastewater/thermal-pollution impacts
- Built on multi-scale hydrologic modeling with physics-informed AI and hierarchical digital twins
- Frames AI's physical footprint — water as well as power — as a siting and policy question
- About $1.21M; University of Alabama (Tuscaloosa); 2025–2028
The NSF awarded about $1,214,308 to the University of Alabama (Tuscaloosa) project "CAIG: AI-Guided Water Availability Tracking and Twin Systems for Infrastructure Resilience" (NSF Award 2530564; program: GEO Cyberinfrastructure / Educational Linkages; October 2025 – September 2028).
Per the abstract, the project starts from the recognition that the rapid expansion of AI infrastructure across the U.S. presents challenges for water resource management — because AI infrastructure (data centers) requires a large water supply for cooling. It therefore develops three new AI models for water resources. First, a large-scale AI model that provides insight into water availability to help choose sites. Second, digital twins that reveal hazards which may disrupt local water supply. Third, an AI method that helps predict the impacts of wastewater and thermal pollution. The primary scientific goal is a multi-scale hydrologic modeling framework that integrates physics-informed AI and hierarchical digital-twin technologies to inform water management for AI infrastructure development.
Why it matters: discussion of the AI boom tends to center on compute and electricity, but the water consumed by data centers is an easily overlooked constraint. This project's distinctive angle is to bridge the demand side (AI infrastructure growth) and the supply side (water resources) using AI itself — physics-informed AI plus digital twins. Being able to visualize where siting is "water-feasible" means AI data-center planning can be optimized for water, not just power. It connects to the energy and regional data we cover and illustrates how AI's physical footprint (power, water, land) is becoming a siting and policy issue.
Why it matters
Research that concretely shows the constraints on AI-infrastructure growth extend beyond power to water. A framework that evaluates siting by water availability is instructive for data-center planning, sustainability, and regional water policy.
FAQ
Why is water relevant to AI?
What is a digital twin?
Sources (primary)
Source: NSF Award Search (U.S. National Science Foundation, public domain). Amounts are the obligated amount. For privacy, we do not handle principal investigator names.
- NSF Award (original, official)
- NSF Award ID: 2530564