The rapid advancement of artificial intelligence (AI) is powering global infrastructure through data centers, but this progress comes with significant environmental costs. Projections indicate that data centers could consume 945 terawatt-hours of electricity annually by 2030, a figure nearly three times the combined electricity usage of Pakistan, Bangladesh, and Nigeria, countries with a total population exceeding 650 million.
However, this electricity consumption represents only a fraction of the issue. Each unit of electricity used by data centers contributes not only to a carbon footprint but also generates a substantial “water footprint” for cooling and power generation, alongside a “land footprint” linked to energy production and supply chains.
Rethinking Sustainability Metrics for AI
New research from the United Nations University (UNU) suggests that AI-related water consumption could match the annual basic domestic needs of 1.3 billion people by the end of the decade. The land footprint is estimated to exceed 14,500 square kilometers, roughly twice the size of the Jakarta metropolitan area.
The report highlights a critical gap in measuring AI’s environmental impact. While focus is often placed on greenhouse gas emissions from training large models, this approach overlooks other environmental costs. Solutions perceived as “green” in one aspect might exacerbate pressures elsewhere, particularly in regions grappling with resource scarcity. For instance, a shift to certain renewable energy sources might reduce carbon emissions but significantly increase water consumption and land use.
Daily AI Use: The Underestimated Driver
While public discourse largely centers on the energy required for training advanced AI models, the study reveals that daily usage accounts for approximately 80% to 90% of the total energy demand. A popular AI service is estimated to process around 2.5 billion requests daily, consuming hundreds of gigawatt-hours of electricity annually. Energy usage varies greatly by task; generating a single AI image can require thousands of times more energy than simple text classification, with video production demanding even more resources. Efficiency improvements alone are insufficient to offset this escalating demand. The report points to the “rebound effect,” where lower costs and enhanced performance ultimately lead to increased usage, driving up overall resource consumption.
Local Burdens, Global Benefits
The environmental impacts of AI infrastructure are not evenly distributed. While the benefits of the technology are global, its costs are often concentrated in specific regions. In some countries, data centers constitute a significant portion of national electricity consumption, straining energy systems. In others, expanding facilities heavily utilize water resources, sometimes under drought conditions.
Concurrently, the report draws attention to a growing electronic waste problem. AI infrastructure is expected to generate up to 2.5 million tons of e-waste annually by 2030. A substantial portion of this burden will fall on lower-income countries with limited capacity for safe disposal. The production of critical minerals required for AI hardware also raises concerns about environmental degradation and social inequalities in extraction regions.
Expanding Digital and Environmental Divides
The proliferation of AI infrastructure is also creating new inequalities in access and impact. According to the report, over 90% of AI-specific computing capacity is concentrated in just two countries, the United States and China. Meanwhile, more than 150 nations lack significant indigenous AI infrastructure. This imbalance not only limits economic opportunities but also raises questions of environmental justice, as some countries bear the environmental costs without reaping the benefits of AI-driven growth.
Towards Responsible AI
Despite these stark findings, UNU researchers emphasize that the report is not an argument against AI. Instead, it calls for urgent action to ensure the technology develops within planetary boundaries. The study presents a framework for a “responsible AI ecosystem” based on principles of 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 are encouraged to design systems that minimize resource consumption. Users also have a role to play by opting for lower-impact applications whenever possible. Ultimately, the report argues that the future of AI hinges not only on technological innovation but also on the governance decisions made today.
