#ai #datacenters #waterrisk #infrastructure #sustainability | Lorenzo Rosa | 92 commentsWe just finished mapping water risk for all 9,500 data c...
Published by Water Network Research, Official research team of The Water Network
#ai #datacenters #waterrisk #infrastructure #sustainability | Lorenzo Rosa |
92 comments
We just finished mapping water risk for all 9,500 data centers on Earth. The results should worry anyone planning AI infrastructure.
Over the past months, our team has been building the first global, facility-level assessment of data center exposure to water stress — combining hydrological modeling with the location of every currently identified data center worldwide. Paper is in progress, but the early findings are striking enough to share now.
What we're finding:
→ 1 in 4 data centers today already sit in locations where water demand exceeds local supply for at least part of the year
→ That rises to 1 in 3 under future hydrological conditions — this isn't a distant risk, it's a planning horizon problem
→ The US hosts nearly 4 in 10 of the world's data centers, and 1 in 5 of those are already in water-stressed regions
→ 4 in 10 data centers globally sit in areas where they're drawing from the same water as cities' municipal supply — meaning growth here is a direct governance and permitting issue, not just an engineering one
Why this matters for business decisions:
Water risk is hyper-local in a way energy risk isn't — you can import power, you can't import a river. That makes site selection, water rights, and community relationships a bigger part of AI infrastructure strategy than most capacity-planning models currently account for. The companies that get ahead of this — through siting choices, alternative cooling, or transparent water accounting — will face far fewer stranded-asset and permitting risks than those that don't.
Curious to hear how operators and investors in this space are already thinking about water exposure.
92 comments
We just finished mapping water risk for all 9,500 data centers on Earth. The results should worry anyone planning AI infrastructure.
Over the past months, our team has been building the first global, facility-level assessment of data center exposure to water stress — combining hydrological modeling with the location of every currently identified data center worldwide. Paper is in progress, but the early findings are striking enough to share now.
What we're finding:
→ 1 in 4 data centers today already sit in locations where water demand exceeds local supply for at least part of the year
→ That rises to 1 in 3 under future hydrological conditions — this isn't a distant risk, it's a planning horizon problem
→ The US hosts nearly 4 in 10 of the world's data centers, and 1 in 5 of those are already in water-stressed regions
→ 4 in 10 data centers globally sit in areas where they're drawing from the same water as cities' municipal supply — meaning growth here is a direct governance and permitting issue, not just an engineering one
Why this matters for business decisions:
Water risk is hyper-local in a way energy risk isn't — you can import power, you can't import a river. That makes site selection, water rights, and community relationships a bigger part of AI infrastructure strategy than most capacity-planning models currently account for. The companies that get ahead of this — through siting choices, alternative cooling, or transparent water accounting — will face far fewer stranded-asset and permitting risks than those that don't.
Curious to hear how operators and investors in this space are already thinking about water exposure.
Tags
- Water Risk Tools
- Water Supply
- Water Risk
- artificial intelligence
- Water sensitive design and planning
- Water Accounting
- United States