Artificial intelligence companies such as OpenAI, Google DeepMind, Microsoft AI, Anthropic, etc. AI Billions of gallons of water are consumed each year. AI AI water consumption is expected to be enough to meet the annual basic water needs of 1.3 billion people by 2030, according to a new report from the United Nations University Institute for Water, Environment and Health (UNU-INWEH). People tend to focus only on the carbon emissions of artificial intelligence, but artificial intelligence also has a huge impact on water resources, the report said.
Data centers that power artificial intelligence require large amounts of water to cool servers and support their operations. Lead author Miriam Aczel said in a UNU-INWEH press release quoted in the Times report that the researchers were surprised to learn that some seemingly environmentally friendly energy options may actually have a worse impact on water resources from a carbon perspective. Azel warned that focusing solely on carbon emissions could create the wrong impression that renewable energy makes AI infrastructure fully sustainable.
Why artificial intelligence needs so much water
AI uses large amounts of water primarily to keep the system cool. Powerful servers run artificial intelligence models, and these servers generate a lot of heat. If machines overheat, they can slow down, get damaged, or stop working. To prevent this, data centers use water-based cooling systems such as evaporative cooling towers.
These systems use water to absorb heat and keep equipment at a safe temperature. Data centers also require water indirectly because it is used to generate electricity and construction materials. Within a data center, servers use electricity to perform calculations and store data. As the Environmental Law Institute explains, some of this electricity is converted into heat, which must be removed to avoid overheating and system failure.
AI’s hidden water problem
The environmental impact of artificial intelligence is already being felt around the world, not just in future projections, the report said. Researchers warn that this amount of water could be dangerous in a region already suffering from drought and water shortages.
A separate report from the EESI environment and energy research institute said some communities are already feeling the impact of data centers on water supplies. in the north virginiaThe United States, often called the data center capital of the world, will have more than 300 data centers consuming nearly 2 billion gallons of water in 2023, a 63% increase from 2019. Loudoun County alone consumed approximately 900 million gallons of water that year, causing the local water department to rely heavily on drinking water quality supplies to meet demand.
Data centers use more water
In Querétaro, a proposed fast data center project has raised concerns about local water supplies during an ongoing drought. In Uruguay, plans to build water-intensive data centers sparked controversy during a severe drought in 2023. Drought has reduced freshwater reserves in Uruguay’s largest city, making tap water unsafe to drink.
Residents protested because they believed industrial projects were more important than people’s access to drinking water. To stop communities suffering from the development of AI, we need a more responsible AI system, the report says. It calls for governments, investors and financial institutions to develop stricter rules to reduce the environmental damage caused by artificial intelligence.
The researchers also said licensing for new data centers should include checks on water use and local community concerns. The report clearly warns that if governments don’t act soon, AI’s growing water demands could put even greater pressure on people and natural resources.
