UN warns AI could consume 945 TWh and generate 2.5M tonnes e-waste by 2030

A new study from the United Nations University (UNU) paints a stark picture of artificial intelligence's growing environmental toll. The report warns that the resource demands of AI data centers are expanding rapidly, threatening water, land, and climate systems in ways that go far beyond the commonly cited greenhouse gas emissions. By 2030, global AI data centers could require up to 945 terawatt-hours of electricity annually—equivalent to the total electricity consumption of a country like Japan or Germany. Additionally, the rapid hardware refresh cycles driven by AI will generate as much as 2.5 million tonnes of electronic waste each year by the end of the decade, creating new challenges for recycling and hazardous waste management.

The UNU report emphasizes that AI's environmental impact is not isolated to energy use. Water consumption for cooling data centers, land use for new facilities and mining for rare earth minerals, and the carbon footprint of manufacturing chips all contribute to a complex ecological burden. The authors urge national governments and tech companies to treat AI infrastructure as a critical component of national resource planning, integrating data center development into energy grids, water management systems, and land-use policies. Without coordinated action, the report warns, AI's rapid deployment could exacerbate resource scarcity and environmental degradation. The study calls for standardized environmental reporting across the AI industry and for investments in more efficient hardware, renewable energy, and circular economy practices to mitigate long-term damage.