A new report from the United Nations University says the rapid rise of artificial intelligence could significantly increase the environmental cost of data centres by 2030. The study warns that these facilities may consume twice as much power and water as they do now if current growth trends continue.
The report highlights a growing gap in public debate. AI is often discussed as software, but the researchers say it is also a physical system. It depends on data centres, electricity generation, cooling systems, transmission networks, chips, minerals, land and water. That means its expansion carries real-world costs far beyond the digital space.
According to the report, global data centres used 448 terawatt-hours of electricity in 2025. By 2030, that figure could rise to 945 terawatt-hours. The researchers say AI may account for around 40 percent of that total demand. They also project water use could reach 9.3 trillion litres by the end of the decade.
Kaveh Madani, director of the United Nations University Institute for Water, Environment and Health and the report’s lead author, said the world still tends to treat AI as a purely digital tool. But in reality, he said, it depends on large-scale infrastructure that consumes energy, water and land.
The report also warns that the growth of AI could add more pressure on scarce natural resources and increase electronic waste. As more data centres are built, governments may face tougher choices between supporting digital growth and protecting environmental sustainability.
Cooling is one of the biggest concerns. Data centres need large amounts of water and electricity to prevent servers from overheating. As AI tools become more complex and widely used, the demand for cooling is expected to rise sharply. That could create new stress in regions already facing water shortages or power constraints.
The researchers say the challenge is not to stop AI development, but to make it more responsible. They argue that governments and companies need to plan for the environmental impact of digital infrastructure from the start. That includes cleaner power, better cooling systems, smarter energy use and stronger recycling of electronic equipment.
The findings arrive at a time when AI is being promoted as a tool for productivity, innovation and economic growth. But the report is a reminder that every digital advance has a physical footprint. If the current trajectory continues, the environmental price of AI could become much harder to ignore.
Instead of seeing AI only as a software revolution, the report urges policymakers to view it as an infrastructure challenge. That shift, the researchers say, is essential if the world wants to balance technological progress with climate and resource security.
