June 2026 – Data centres, the backbone of artificial intelligence, are on track to consume 945 terawatt-hours of electricity annually by 2030. That figure is nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria — countries home to more than 650 million people. But electricity is only part of the story. Each unit of energy used by data centres carries a water footprint for cooling and energy production, and a land footprint tied to power generation and supply chains.
A new study from UN University (UNU) warns that AI-related water consumption could equal the basic annual domestic needs of 1.3 billion people by the end of the decade, while its land footprint may exceed 14,500 square kilometres, roughly twice the size of Jakarta’s metropolitan area. The report highlights a critical gap in how AI’s environmental impact is measured: greenhouse gas emissions are often prioritized, while water, land, and waste costs remain overlooked.
Daily use of AI, not just training large models, is the main driver of demand. The study finds that 80 to 90 percent of total energy consumption comes from everyday usage. One widely used AI service processes around 2.5 billion prompts per day, consuming hundreds of gigawatt-hours annually. Tasks vary dramatically in energy intensity: generating a single AI image can require more than a thousand times the energy of simple text classification, while video generation demands even greater resources. Efficiency improvements alone may not offset these rising demands due to the rebound effect, where lower costs drive higher usage.
The environmental burdens are unevenly distributed. While AI’s benefits are global, its costs are concentrated in specific regions. In some countries, data centres already account for a significant share of electricity consumption, straining energy systems. In others, facilities draw heavily on water supplies amid drought conditions. The report also warns of a looming e-waste crisis, with AI infrastructure projected to generate up to 2.5 million tonnes annually by 2030, much of it falling on lower-income countries with limited disposal capacity.
Critical minerals required for AI hardware raise further concerns about environmental degradation and social inequities in extraction regions. Meanwhile, more than 90 percent of AI-specialised computing capacity is concentrated in the United States and China, leaving over 150 nations without significant infrastructure. This imbalance risks widening both the digital and environmental divide, as some countries bear the costs without sharing in the benefits.
Despite the stark findings, UNU researchers stress that the report is not an argument against AI but a call for urgent action. They propose a framework for a responsible AI ecosystem, built on transparency, efficiency by design, equity, lifecycle responsibility, global cooperation, and sustainable use. Governments are urged to integrate AI infrastructure into energy, water, and land-use planning, while companies must design systems that minimize resource consumption. Users also play a role by choosing lower-impact applications where possible.
The future of AI will depend not only on technological innovation but also on governance choices made today. Ensuring that AI develops within planetary limits is essential to balance its global benefits with its local environmental costs.
