The rapid expansion of artificial intelligence (AI) could come at a significant environmental cost, with a new United Nations report warning that the rising energy demands of AI systems may place additional pressure on global resources.
According to the report, AI-related energy consumption could double by 2030, potentially accounting for around 3 per cent of the world’s total electricity use. The resulting carbon emissions could reach levels comparable to the United Kingdom’s current emissions, while water consumption for cooling data centres could exceed the annual drinking water requirements of the global population.
The report highlights that AI’s growth could follow the economic principle known as the “Jevons Paradox”, wherein improvements in efficiency lead to increased overall consumption rather than a reduction in resource use.
Named after economist William Stanley Jevons, the paradox was first observed in 19th-century England, when greater efficiency in coal use resulted in higher overall coal consumption as costs fell and demand expanded.
The UN report suggests that as AI models become cheaper and more accessible, their widespread adoption could spur new applications and significantly increase usage, potentially offsetting the benefits of improved efficiency.
Data centres already consume vast amounts of energy. Last year, their electricity consumption was estimated to be equivalent to that of Saudi Arabia, one of the world’s largest electricity consumers. If AI-driven energy demand doubles by 2030, the resulting carbon footprint could require nearly 6.7 billion trees grown over a decade to offset the emissions.
The report further estimates that data centres could require approximately 9.3 trillion litres of water and land covering an area nearly ten times the size of Mexico City to support future AI infrastructure.
Beyond environmental concerns, the report points to growing global inequalities in AI infrastructure. Only 32 countries currently host AI-specific cloud infrastructure, with nearly 90 per cent of that capacity concentrated in the United States and China.
It warns that countries lacking AI infrastructure may face a widening digital divide while simultaneously bearing environmental costs associated with mineral extraction, supply chains and electronic waste.
The report identifies two key factors shaping AI’s environmental footprint: the scale of AI deployment and the nature of the tasks being performed. Text generation, coding, image creation and video processing each require varying levels of computing power and energy.
It also notes that the choice of AI models plays a crucial role, as different systems carry varying energy demands and environmental impacts.
Calling for responsible AI development, the UN report emphasises the need for comprehensive value-chain governance, covering everything from mineral sourcing and infrastructure development to recycling and safe disposal.
The report urges governments and industries to strike a balance between AI innovation and environmental responsibility, ensuring that technological progress does not come at the expense of the planet.