The rapid expansion of artificial intelligence infrastructure could place unprecedented strain on global water resources, with water consumption linked to AI-powered data centres projected to reach levels equivalent to the annual basic needs of 1.3 billion people by 2030, according to a recent report by the United Nations University Institute for Water, Environment and Health (UNU-INWEH).
The report highlights a largely overlooked environmental consequence of the AI revolution. While public debate has largely focused on the carbon emissions associated with artificial intelligence, researchers say the industry’s growing demand for freshwater poses an equally serious challenge. As AI systems become more sophisticated and widely used, the network of data centres that powers them is expanding rapidly, driving up both electricity and water consumption.
According to the study, AI data centres require enormous volumes of water to keep servers operating at safe temperatures. Advanced AI models run on powerful computer processors that generate significant heat during operation. To prevent equipment failure and maintain efficiency, many facilities rely on water-based cooling systems, including evaporative cooling towers. Beyond cooling, water is also indirectly consumed through electricity generation and the production of materials needed to construct data centre infrastructure.
Researchers estimate that by the end of the decade, the water footprint of global data centres could reach approximately 9.3 trillion litres annually. At the same time, electricity consumption by these facilities is expected to surge dramatically, reflecting the growing demand for AI-driven services across industries and households. The report cautions that efforts to reduce carbon emissions alone may not address the broader environmental burden created by the sector. Some energy sources considered climate-friendly could still have substantial impacts on water resources.
Evidence of the problem is already emerging in several parts of the world. In Northern Virginia, one of the world’s largest data centre hubs, hundreds of facilities consumed billions of gallons of water in recent years, with usage rising sharply as new centres came online. Similar concerns have surfaced in regions facing recurring droughts, including parts of Mexico and Uruguay, where residents and environmental groups have questioned whether scarce water resources should be allocated to industrial projects during periods of water stress.
The UN report warns that the burden of AI’s resource consumption is often unevenly distributed. While technology companies and consumers may reap the benefits of AI-driven innovation, communities located near data centres frequently bear the environmental costs. Researchers argue that vulnerable regions facing water shortages could be disproportionately affected if current trends continue.
To address these concerns, the report calls on governments, regulators, investors and technology firms to adopt stronger environmental safeguards. Recommendations include mandatory disclosure of water usage, stricter assessment of proposed data centre projects, and greater consideration of local water availability before approving new facilities. The researchers also urge the industry to develop more efficient cooling technologies and integrate sustainability measures into future AI infrastructure planning.
The findings add to a growing global debate over the environmental footprint of artificial intelligence, suggesting that the race to build more powerful AI systems must be balanced against the need to protect increasingly scarce natural resources. Without timely intervention, the report warns, AI’s hidden thirst could intensify water stress in vulnerable regions and create new challenges for communities already struggling to secure reliable access to clean water.
