Note: Single-source report; awaiting corroboration.
Data centers supporting AI processes could consume up to 945 terawatt-hours of electricity annually by 2030, nearly tripling the combined annual electricity use of Pakistan, Bangladesh, and Nigeria, which together have over 650 million residents, according to a UN University study.
Beyond carbon emissions, every unit of electricity used by these centers entails substantial water use for cooling and energy production, as well as land use associated with power generation and supply chains. The report notes that AI-related water consumption could meet the basic annual domestic needs of 1.3 billion people by decade's end, while its land footprint might cover more than 14,500 square kilometers—about twice the size of the Jakarta metropolitan region.
The study highlights a significant gap in assessing AI's environmental impacts, pointing out that focusing mainly on greenhouse gas emissions linked to model training neglects other important factors. Some renewable energy transitions that reduce carbon emissions may, according to the report, increase water and land resource demands, posing challenges especially in resource-scarce regions.
Most of the energy demand comes from daily AI usage, accounting for approximately 80 to 90 percent of total energy consumption. For example, one widely used AI service processes around 2.5 billion prompts daily, consuming hundreds of gigawatt-hours of electricity annually. Energy needs vary significantly by task, with generating a single AI image using over a thousand times the energy of simple text classification, and video generation requiring even more. The report cautions that efficiency gains alone may not reduce overall consumption due to a rebound effect—enhanced performance drives greater use and thus higher total resource demand.
The environmental burdens of AI infrastructure are not evenly distributed worldwide. While AI's benefits are global, its resource costs are often concentrated in specific areas, where data centers already represent a sizable portion of national electricity use, increasing pressure on local energy systems.