UN calculates nation-sized environmental footprints for AI and data centers - ABC News
A United Nations University report reveals that the environmental footprint of data centers, heavily influenced by AI, now rivals that of large nations. It projects a doubling of their water and energy use and pollution by 2030. Experts are calling for public awareness, greater industry transparency, and increased efficiency to mitigate these impacts.
A United Nations University report details the significant environmental footprint of data centers, projecting their water and energy consumption, and associated pollution, to double by 2030, largely driven by the growth of artificial intelligence. In 2023, global data centers used 448 trillion watt-hours of electricity, surpassing all but ten countries, and generated 208 million tons of carbon dioxide. By 2030, this energy use is expected to rise to 935 trillion watt-hours, potentially making data centers the sixth-highest power consumer globally if they were a country, producing nearly 440 million tons of CO2.
Kaveh Madani, study co-author and director of the United Nations University Institute for Water, Environment and Health, emphasized the "enormous" scale of demand, comparable to nations. Fengqi You, a Cornell University energy engineering professor, highlighted the report's value in comprehensively framing carbon, water, land, and environmental justice impacts for an often-secretive industry. Jean Su, director of the Energy Justice Program at the Center for Biological Diversity, described it as the first U.N. report to shed light on AI's environmental harms.
While Caleb Max, president of the National Artificial Intelligence Association, defended the industry, citing AI's societal benefits and increasing efficiency, and Josh Levi, president of the Data Center Coalition, affirmed the industry's commitment to responsible growth, Madani countered that AI, despite appearing clean, relies on substantial physical infrastructure and energy, causing unseen environmental impacts. The report also suggests that reducing AI query length by 30% could cut energy use by 25%.
A key challenge identified by authors Miriam Aczel and Madani is the lack of transparency from companies regarding data center consumption and location, which hinders effective management of their environmental impacts.