AI's Growing Thirst: A Future of Water and Energy Concerns
A United Nations report warns that AI's growing energy and water consumption could soon surpass human drinking needs, highlighting the environmental impact of data centers.
As artificial intelligence continues to advance, its energy and water demands are becoming increasingly significant, sparking concerns about its environmental impact. According to a recent United Nations report, AI's energy consumption could double by 2030, using up to 3% of the world's electricity and a substantial amount of water for cooling purposes—potentially exceeding the annual drinking water needs of the global population.
The report challenges a common belief that AI models will naturally become more efficient over time, thus reducing their environmental footprint. Instead, it suggests that AI's increasing efficiency could lead to greater overall consumption in line with the Jevons paradox, an economic principle observed in the 19th century that increased efficiency can result in higher overall use of a resource.
To mitigate these impacts, the report proposes a framework for responsible AI usage, emphasizing transparency, global cooperation, and sustainable practices. It warns that without such measures, AI's growing demands could exacerbate existing inequalities, particularly as most AI-specific infrastructure is concentrated in just 32 countries, primarily the US and China. The report calls for comprehensive governance across the AI lifecycle, from mineral sourcing to recycling, to ensure a balance between technological advancement and environmental stewardship.