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

Key Points:

• The UNU-INWEH report highlights the significant environmental footprints of AI, emphasizing carbon, water, and land impacts associated with the energy used to power AI systems, including data centers and infrastructure.
• It reveals that AI's environmental costs vary based on electricity sources and generation locations, noting that low-carbon energy does not necessarily mean low water or land use, underscoring the complexity of AI's resource demands.
• The report identifies both large-scale infrastructure growth and everyday AI usage patterns—such as model choice and content generation—as key factors influencing AI's environmental footprint.
• It frames AI's environmental impact as a governance and justice issue, with environmental burdens often concentrated in specific communities, calling for transparency, equity, lifecycle responsibility, and global cooperation in AI development.
• By quantifying AI’s carbon, water, and land footprints, the report aims to inform integrated planning across energy, climate, water, and land sectors to promote sustainable AI innovation that avoids shifting environmental costs to vulnerable populations.