United Nations Report Warns of the Hidden Water Cost of Artificial Intelligence
Artificial intelligence is transforming how people learn, work, and communicate. However, behind every chatbot, image generator, and recommendation system is a growing network of data centers that requires enormous amounts of electricity and water.
According to a report from the United Nations University Institute for Water, Environment and Health, discussions about AI’s environmental impact often focus too heavily on carbon emissions. While carbon pollution remains important, experts say it is only part of the story.
Every time an AI system is trained or used, it consumes electricity. Producing that electricity often requires water, and data centers also use large amounts of water to keep their servers cool. As AI technology expands, those demands are expected to rise dramatically.
Researchers estimate that by 2030, AI-related data centers will consume 945 terawatt-hours of electricity worldwide each year. Their annual water use could reach 9.3 trillion liters. To put that figure into perspective, it is roughly equal to the basic yearly water needs of all 1.3 billion people living in sub-Saharan Africa.
Experts warn that focusing on a single environmental measure can hide important trade-offs. For example, substituting coal with bioenergy may lower carbon emissions by about 70%, but it can raise the water footprint by more than 30 times. A technology that appears environmentally friendly in one area may create new challenges in another.
The effects can also be felt at the local level. Some large data centers use up to 5 million gallons of water per day for cooling. In drought-prone regions, that demand can place additional pressure on already limited water supplies. Communities in Mexico’s Queretaro and Uruguay have raised concerns about new data center projects during periods of water scarcity.
The issue extends beyond water. As of 2025, only 32 countries host specialized AI data centers, while about 90% of global computing capacity is concentrated in the United States and China. This imbalance has sparked concerns about environmental justice, as countries with little access to AI may still bear the costs of mineral extraction and electronic waste.
U.N. scientists say creating a more responsible AI future will require greater transparency, better planning, and a broader understanding of the technology’s environmental footprint. They urge governments to incorporate data centers into national water and energy strategies. They also call on technology companies to reveal their full land and water consumption, not just their carbon emissions.
Sean Jung R&D Division Director teen/1782706765/1613367592
1. How does artificial intelligence contribute to rising global water consumption rates?
2. Why is focusing solely on carbon emissions problematic for environmental policy?
3. What regions face the most significant pressure from data center demand?
4. Which countries currently hold the majority of global AI computing capacity?
1. Should tech firms prioritize water transparency over mere carbon emission reports?
2. Does artificial intelligence development justify the depletion of local water resources?
3. How can nations balance economic AI growth with essential water security?
4. Is it fair for nations without AI access to bear costs?
Artificial intelligence is transforming how people learn, work, and communicate. However, behind every chatbot, image generator, and recommendation system is a growing network of data centers that requires enormous amounts of electricity and water.
Experts warn that focusing on a single environmental measure can hide important trade-offs. For example, substituting coal with bioenergy may lower carbon emissions by about 70%, but it can raise the water footprint by more than 30 times. A technology that appears environmentally friendly in one area may create new challenges in another.
