AI data centres risk creating global water and land crisis, UN warns

The rapid expansion of artificial intelligence is consuming planetary resources at an unsustainable rate, a United Nations (UN) study has warned.

The report, by the UN University Institute for Water, Environment and Health (UNU-INWEH), reveals that data centres now use more electricity than entire nations such as Saudi Arabia.

By 2030, the technology's global water footprint is projected to hit 9.3 trillion litres — enough to meet the drinking needs of the world's population for more than a year and a half.

Researchers warn that the public focus on carbon emissions has obscured the "profoundly physical" toll of the technology on water supplies, land use and toxic electronic waste.

Though often described as weightless and virtual, the reality of AI is profoundly physical," said Tshilidzi Marwala, rector of the United Nations University and a UN under-secretary-general.

Behind every prompt, image or video lies a growing infrastructure of energy systems, water withdrawals, land use, mineral extraction and electronic waste."

The UN report calculated that data centres consumed 448 terawatt-hours (TWh) of electricity in 2025, which would rank them as the world's 11th-largest electricity consumer if they were a country.

By 2030, this figure is projected to rise to 945TWh, representing nearly 3% of global electricity consumption.

The study found that day-to-day use of AI models — known as inference — accounts for 80% to 90% of their total energy consumption, far outstripping the power needed to train them.

The report also highlights how the resources required vary dramatically by task.

While a single text prompt requires very little energy, generating an AI image uses enough electricity to power a 10-watt LED bulb for 17 minutes and consumes about two tablespoons (29ml) of water.

A high-complexity AI video requires enough power to run the same light bulb for 42 hours, consuming 4.1 litres of water — almost a two-day supply of drinking water for one person.

The expansion is also creating a deep "digital divide", the UN warns. As of 2025, 90% of global AI computing capacity was concentrated in just two countries: the US and China.

Meanwhile, the environmental costs of mineral extraction and disposing of up to 2.5 million tonnes of toxic e-waste each year by 2030 are disproportionately borne by poorer nations.

Representatives from the technology sector said they were actively working to address the environmental footprint of their systems.

Josh Levi, president of the US-based Data Center Coalition, said the industry remains committed to working with policymakers and local communities.

We are committed to ensuring that as data centres grow, they do so responsibly, transparently, and in ways that reflect the best available practices," Levi said in a statement.

Other industry figures pointed to the potential benefits of the technology, arguing that AI could help manage resource grids more efficiently and accelerate climate science.

The UN researchers have called for a "responsible AI ecosystem" that enforces mandatory environmental disclosures, greater community consultation, and stricter government standards.