The UN University research behind the report estimates that AI's associated water footprint could, by 2030, equal the basic annual domestic water needs of 1.3 billion people in sub-Saharan Africa, while its land footprint could exceed 14,500 sq km. It also challenges the way AI's environmental cost is commonly understood: while attention has focused on the energy required to train large models, day-to-day usage or inference accounts for roughly 80–90% of total AI energy demand.

The scale of everyday use is staggering. One widely used AI service is estimated to process around 2.5 billion prompts a day. Energy demand also varies dramatically by task: an AI-generated image can require around 1,450 times the energy of basic text classification, while video generation is considerably more resource-intensive. One widely used AI service is estimated to process around 2.5 billion prompts a day.