HAMILTON, Jun 9: A new United Nations report has warned that the rapid expansion of artificial intelligence could significantly increase global energy and water consumption over the next decade, raising concerns about its long-term environmental impact.
The report estimates that by 2030, AI-related operations could account for nearly 3 per cent of the world’s electricity use, while generating carbon emissions comparable to those of the United Kingdom. It also projects that cooling requirements for data centres may consume more water annually than the global population needs for drinking.
Contrary to expectations that technological improvements will curb resource consumption, the report highlights the “Jevons paradox” an economic principle suggesting that greater efficiency often leads to increased overall use. As AI systems become cheaper and more accessible, demand for new applications and wider adoption is expected to rise, potentially offsetting gains from improved efficiency.
The findings underline the growing environmental footprint of data centres, which already consume electricity on a scale comparable to Saudi Arabia. If current projections materialise, the resulting emissions would require billions of trees over a decade to absorb the additional carbon output.
The report also points to global disparities in AI infrastructure. While only a limited number of countries host advanced AI cloud facilities, most of the world’s computing capacity is concentrated in the United States and China. Developing nations, meanwhile, often shoulder environmental costs linked to mineral extraction, manufacturing and electronic waste without enjoying comparable technological benefits.
Researchers stressed that the environmental burden of AI depends not only on the volume of use but also on the types of tasks performed. Resource-intensive applications such as image and video generation generally require far more computing power than text-based functions.
To address these challenges, the report calls for stronger oversight across the entire AI lifecycle, from sourcing raw materials to equipment recycling and disposal. It recommends routine environmental disclosures, improved energy planning and greater international cooperation to ensure technological advancement does not come at the expense of sustainability.
The study concludes that environmental considerations must become a central part of AI development strategies, warning that efficiency improvements alone are unlikely to resolve the sector’s growing resource demands.