AI’s environmental costs threaten water, land and climate
AI’s environmental costs threaten water, land and climate
AI s environmental costs threaten water – The environmental toll of artificial intelligence extends beyond its contribution to greenhouse gas emissions, as its ecological footprint is growing rapidly and could overwhelm natural resources.
Data centres, which serve as the backbone of AI technology, are projected to use 945 terawatt-hours of electricity annually by 2030—nearly three times the combined annual energy consumption of Pakistan, Bangladesh, and Nigeria. These facilities power the algorithms that drive AI, but their operational demands are raising concerns about sustainability.
Beyond carbon emissions, the energy demands of AI systems create additional environmental burdens, including significant water and land usage. Every unit of electricity consumed by data centres is linked to a “water footprint” for cooling and energy production, as well as a “land footprint” tied to power generation and supply chains.
Global Impact of AI’s Resource Consumption
According to a new report by the United Nations University (UNU), AI’s water consumption could match the daily needs of 1.3 billion people by the end of the decade, while its land use may surpass 14,500 square kilometres—roughly equivalent to twice the size of Jakarta.
Energy use varies dramatically based on the task. A single AI image can demand over a thousand times the energy of simple text classification, and video generation requires even greater resources. The study estimates one popular AI service processes around 2.5 billion prompts daily, consuming hundreds of gigawatt-hours of electricity yearly.
Rebound Effect and Uneven Distribution
Efficiency gains alone may not counteract these rising demands. The report identifies the rebound effect, where reduced costs and improved performance lead to increased usage, ultimately raising total resource consumption.
The environmental impacts of AI infrastructure are unevenly distributed. While the technology’s benefits are global, its costs often fall disproportionately on specific regions. In some countries, data centres already account for a large portion of national electricity use, straining energy systems. Others face pressure on water supplies, especially during droughts, as new facilities expand.
The study also warns of an escalating e-waste crisis, projecting up to 2.5 million tonnes of electronic waste annually by 2030 from AI infrastructure. This burden is expected to disproportionately affect lower-income nations with limited waste management capabilities.
Global Inequalities in AI Development
Mineral extraction for AI hardware adds another layer of environmental concern, driving degradation and social inequities in mining regions. The expansion of AI infrastructure is also deepening global inequalities in access and influence. Over 90% of AI computing power is concentrated in the U.S. and China, while 150 countries lack substantial domestic capabilities.
This imbalance not only restricts economic opportunities but also raises concerns about environmental justice, as some nations shoulder the ecological costs without reaping the benefits of AI-driven progress.
Despite these challenges, UNU researchers argue that the report does not advocate against AI. Instead, it calls for immediate steps to align AI development with planetary boundaries. The study outlines a framework for a “responsible AI ecosystem,” emphasizing transparency, efficiency by design, equity, lifecycle responsibility, global cooperation, and sustainable use.
Authorities are urged to incorporate AI infrastructure into energy, water, and land-use planning. Companies are encouraged to design systems that minimize resource consumption, while users can contribute by opting for lower-impact applications.