UN report says AI growth could consume water equal to needs of 1.3 billion people

A United Nations study warns that the rapid expansion of artificial intelligence is creating an unsustainable environmental footprint. By 2030, data centers that power AI models could use 945 terawatt‑hours of electricity annually, and the water required for cooling and power generation may equal the domestic water consumption of 1.3 billion people. The report also projects up to 2.5 million tonnes of toxic electronic waste each year, a burden that would fall disproportionately on lower‑income countries lacking safe recycling capacity. Daily AI model inference accounts for roughly 80‑90 % of total AI energy demand, making everyday usage the primary driver of emissions.

Other research offers a more optimistic view, suggesting AI applications in power, food and mobility could cut global greenhouse‑gas emissions by 3.2‑5.4 billion tonnes annually by 2035, potentially offsetting AI data‑centre emissions. However, a MIT paper cautions that without global carbon pricing, mandatory disclosure of full system‑wide environmental costs, and strong state intervention, AI could ultimately worsen climate change.