AI Data Centres Could Strain Power and Water by 2030: Study
A UN-backed study warns that AI-powered data centres could consume 945 TWh of electricity by 2030, putting growing pressure on energy and water resources.
AI may be transforming industries, but its future could depend on two resources that are becoming increasingly scarce: electricity and water.
A new UN-backed assessment has raised concerns about the environmental demands of the data centres that power modern AI systems. As companies race to build larger models and deploy AI across more services, the infrastructure supporting that growth is consuming vast amounts of energy and water.
Data centres could nearly double electricity use by 2030
According to researchers from the UN University Institute for Water, Environment and Health, global data centres could consume around 945 terawatt-hours (TWh) of electricity by 2030. That is more than double the approximately 448 TWh used in 2025.
The report highlights that if data centres were treated as a country, they would rank among the world's largest electricity consumers. As AI adoption accelerates, energy demand is expected to rise even further.
Water use is another growing concern.
The study estimates that the water required to operate and cool data centres could equal the minimum annual household needs of around 1.3 billion people in Sub-Saharan Africa.
Why does AI require so many resources?
Unlike traditional software, AI systems rely on thousands of powerful processors working together to train and run models. These chips consume significant amounts of electricity and generate large amounts of heat.
To keep equipment operating safely, data centres depend on cooling systems that often require both energy and water. As AI models become larger and more complex, the pressure on local resources increases. While operators continue improving efficiency, experts warn that rising demand could outpace those gains.
Warning signs are already appearing
The impact is already visible in some parts of the world. In Ireland, data centres accounted for around 21% of metered electricity consumption in 2023, leading authorities to pause new grid connections around Dublin until 2028.
Concerns have also emerged in water-stressed regions such as Mexico and Uruguay, where new data centre projects could place additional strain on local supplies. Beyond electricity and water, the report warns that rapid hardware upgrades may generate millions of tonnes of electronic waste by the end of the decade.
Can AI growth remain sustainable?
Researchers argue that AI's environmental footprint can still be managed through better planning and transparency. Measures such as improving cooling systems, increasing server efficiency, using cleaner energy sources and aligning new facilities with local resource availability could help reduce pressure on communities.
The report expresses that AI's future is not just a question of computing power. As demand for intelligent systems grows, access to reliable electricity and water may become some of the industry's most important challenges.