A new report from the United Nations University Institute for Water, Environment and Health (UNU-INWEH) exposes the vast environmental footprint of AI infrastructure worldwide [1, 2]. The report highlights that data centres consumed an estimated 448 terawatt-hours (TWh) of electricity last year, putting their energy use on par with the entire country of France and ranking them as the 11th largest electricity consumers globally [1].
In addition to electricity, data centres used around 4.5 trillion litres of water—enough to fill 1.8 million Olympic-sized swimming pools—and occupied 6,900 square kilometers of land, nearly 4.5 times the size of Greater London [1]. The report warns these demands will intensify: by 2030, electricity consumption is expected to exceed 945 TWh, equivalent to the 6th largest electricity-consuming country in the world, with carbon emissions surpassing those of the United Kingdom [2]. Water use could rise to 9.32 trillion litres, enough to supply the entire annual basic water needs of sub-Saharan Africa, while land use will expand to an area 18 times the size of New York City [1, 2].
The environmental impact also extends to electronic waste. The AI sector may generate up to 2.5 million tonnes of e-waste annually by 2030—comparable to discarding the weight of 250 Eiffel Towers each year [1]. Currently, 90% of the world’s AI-specialised cloud infrastructure is hosted by the United States and China, who control access, pricing, and data governance [1].
The global AI market is projected to grow rapidly, from $189 billion in 2023 to $4.8 trillion by 2033 [2]. The popular AI tool ChatGPT alone processes about 2.5 billion commands per day, with annual electricity consumption roughly enough to power nearly 3 million people [2]. Generating short AI videos uses far more energy than creating AI images, with one video consuming the electricity equivalent of hundreds of images [2].
Amid these trends, UNU-INWEH director Kaveh Madani called for increased transparency on the environmental impact of AI. Madani said, "目前看到的可能只是冰山一角，業者必須提高透明度," meaning the visible effects may be just the tip of the iceberg [2]. The report urges AI companies to disclose their environmental footprints and recommends governments impose standardized reporting requirements on the industry [2].
The report consolidates data from recent years, noting the scale of AI’s resource use in 2025 and projecting sharp rises by 2030 [1, 2]. It sets the stage for regulatory efforts to address AI’s ecological costs as the technology expands rapidly. The UNU-INWEH is expected to follow up with industry stakeholders and policymakers to promote sustainable AI development.
