UN Alarm: By 2030, Artificial Intelligence Will Consume Water Equivalent to the Needs of 1.3 Billion People

The expansion of Artificial Intelligence could have an increasingly significant impact on global environmental resources.

According to a United Nations report titled "Environmental Cost of AI Energy Consumption: Carbon, Water, and Soil Footprint," by 2030, the sector could consume an amount of water equivalent to that needed for 1.3 billion people, a number comparable to the entire population of Sub-Saharan Africa.

The study also highlights that the energy demand of AI could reach 945 terawatt-hours of electricity. This amounts to about three times that used collectively by the approximately 650 million inhabitants of Pakistan, Bangladesh, and Nigeria.

According to the report, the environmental impact of Artificial Intelligence must be assessed considering the entire technological supply chain, from the construction of data centers to the daily use of systems and the management of electronic waste generated by the sector.

The document does not intend to question the value of technology. As explained by Kaveh Madani, head of the research group that developed the report, the study does not represent an indictment against Artificial Intelligence. On the contrary, it is described as an invitation to promote responsible use of technology and to proactively address its unintended effects, so as to make it sustainable and fair in the long term.

AI and Water Consumption: Other Insights from the Report

To support its conclusions, the report analyzes some concrete cases. In Ireland, one of the European countries hosting numerous data centers, these facilities consumed 21% of the national electricity in 2023, surpassing the consumption attributed to the urban population.

Another example concerns Uruguay, where projects for the construction of a data center characterized by high water consumption overlapped with the severe drought of 2023 that drained Montevideo's freshwater reserves, rendering the water distributed through the public network non-potable.

The study also emphasizes that the impact of AI is not only related to carbon dioxide emissions. Water consumption, necessary for cooling processors, and land use represent relevant factors. For this reason, the report concludes that it is not possible to correctly assess the environmental effects of Artificial Intelligence by observing a single parameter. The United Nations therefore considers a global political guidance for the sector to be necessary.