Environmental Cost of Artificial Intelligence: Carbon, Water, and Land Footprints

This report from the United Nations University Institute for Water, Environment and Health (UNU-INWEH) can help you to understand the growing scope and scale of the environmental costs of AI and data centers. The report quantifies the carbon, water, and land footprints of AI use around the world, noting that by 2030 global data centres powering artificial intelligence are projected to consume 945 terawatt-hours of electricity by 2030; use the equivalent of the basic annual domestic water needs of all 1.3 billion people in Sub-Saharan Africa; and have a land footprint that exceeds 14,500 square kilometers - approximately double the size of the Jakarta metropolitan area, which is home to more than 32 million people.

The report explains that AI’s environmental costs depend not only on how much electricity is used, but also on where that electricity is generated and which energy sources power it. The report also explains that AI’s footprint is shaped by both major infrastructure trends, including the rapid growth of data centers, and everyday use patterns, such as model choice, output length, modality, and the growing use of text, image, and video generation. The report calls for a responsible AI ecosystem built on six principles (transparency; efficiency by design; equity and environmental justice; lifecycle responsibility; global cooperation; and sustainable use), and provides practical recommendations for key stakeholders.