Majority of US AI Datacenters Being Built in Drought-Stricken Regions, Creating Water Crisis
Key Takeaways
- ▸Two-thirds of 809 planned US AI datacenters are located in drought-affected regions, with existing datacenters showing similar patterns
- ▸AI datacenter water demand will quadruple from 17 billion to 73 billion gallons annually by 2028, straining drought-stressed water supplies
- ▸Major tech companies prioritize arid locations for cost and operational benefits, but this strategy intensifies competition for scarce water resources
Summary
A Guardian investigation reveals that approximately two-thirds of the 809 planned AI datacenters in the US are being constructed in regions experiencing severe drought conditions, exacerbating water scarcity across already parched areas. Large AI datacenters consume up to 5 million gallons of water daily for cooling systems, with water demand from datacenters expected to reach 73 billion gallons annually by 2028—more than quadruple the 17 billion gallons used in 2023. Companies including Google, Meta, Microsoft, and Amazon are prioritizing drought-prone areas due to lower land costs, tax incentives, and reduced equipment corrosion in arid climates.
The expansion occurs amid record drought conditions affecting over 60% of the contiguous US, the largest spring drought extent in modern records. Researchers estimate that a single 100-word AI prompt consumes approximately 500ml of water through datacenter cooling. Major projects include a complex twice the size of Manhattan recently approved in Utah (deep in drought since summer 2025) and Amazon datacenters in drought-affected Walla Walla County, Washington. Water resources experts warn that without intervention, competing demands for diminishing water supplies will create an inevitable resource crisis, particularly in Texas, where datacenters could account for 9% of the state's total water use by 2040.
- Water experts warn that without policy intervention, the boom in AI infrastructure will trigger unavoidable water allocation conflicts in drought-prone states
Editorial Opinion
The AI industry's race for computational dominance is being built on a catastrophic blind spot: water scarcity in America's driest regions. While tech companies justify these locations through economics and engineering logic, the cumulative impact is indefensible. As the climate crisis intensifies droughts and datacenter water consumption explodes, policymakers must demand that AI infrastructure expansion include genuine water sustainability requirements—not just tax breaks. The time for reactive regulation is over.
