Energy, water use and pollution of AI and data centers rival most countries

A United Nations University report warns that global data centers already consume more electricity than all but 10 countries, emitting 208 million tons of CO2 annually—equivalent to Argentina—and predicts their energy use will double by 2030, driven by AI. The study highlights water consumption, pollution, and the need for transparency in an industry projected to account for 3% of global electricity demand by 2030, with AI-related energy use set to rise from 20% to 40%." "article": "A United Nations University report released Wednesday reveals that data centers now consume 448 trillion watt-hours of electricity annually, surpassing all but 10 countries globally. Their operations generate 208 million tons of CO2—comparable to Argentina’s emissions—and require 1.2 trillion gallons of water for cooling, according to the study. By 2030, data centers are expected to account for nearly 3% of the world’s electricity use, totaling 935 trillion watt-hours, ranking sixth-highest if classified as a country. CO2 emissions would nearly double to 440 million tons, though the report did not assess water use further. The report emphasizes AI’s role in driving growth, with AI-related energy consumption projected to rise from 20% to 40% of data centers’ total usage by 2030. Study co-author Kaveh Madani, a water scientist at the UN University Institute for Water, Environment and Health in Canada, described the scale as comparable to nations, stressing the physical infrastructure and energy demands behind AI operations. Experts like Fengqi You, a Cornell University energy engineering professor, noted the report’s significance in addressing AI’s environmental impacts holistically, including carbon, water, and land use. While industry leaders like the National Artificial Intelligence Association and the Data Center Coalition emphasize efficiency improvements and public benefits, critics argue the findings underscore the need for greater transparency and sustainable practices. The study highlights AI’s hidden environmental footprint, contrasting its perceived cleanliness with the substantial energy and resource demands of data centers. Madani warned that AI’s physical impacts—including hardware, energy, and infrastructure—often go unnoticed despite their real-world consequences.