Politics Business Culture Technology Environment Travel World
Home Technology Feature
Technology · Exclusive

Can new cooling tech really solve AI's water problem?

Can new cooling tech really solve AI's water problem?
Technology · 2026
Photo · Kai Lindgren for European Pulse
By Kai Lindgren Technology Editor Sep 7, 2026 3 min read

The rapid expansion of artificial intelligence has brought a hidden environmental cost: water. Data centres, the warehouses of servers that power everything from cloud computing to AI models, generate immense heat that must be managed. In 2025, these facilities worldwide used an estimated 222 billion litres of water for cooling, according to consultancy Rystad Energy. Without intervention, that figure could nearly triple to 644 billion litres by 2030.

Tech giants, under mounting public pressure, insist they have a solution. Nvidia, the US chipmaker, claims its new DSX system for designing and managing AI data centres can almost eliminate water consumption at some facilities. The system uses closed-loop cooling, where liquid flows directly through servers, close to chips that can exceed 80°C. By starting with warmer water—around 45°C, compared to the industry standard of 32°C—Nvidia avoids the need for energy-intensive chilled air. Simple fans circulating air are often sufficient, though in hotter climates or during heatwaves, additional cooling may still be required.

But the trade-off is stark: using less water typically requires more electricity. “There’s a pretty direct trade-off between how much water is used and how much energy is used,” said Andy Masley, an independent researcher covering AI and data centres. The liquid in sealed pipes still needs to be cooled, often by blowing air over it, which consumes power.

Public pressure and corporate responses

Microsoft, Amazon Web Services (AWS), and Meta have all told AFP they use closed-loop systems that result in no net water loss. Yet their overall water consumption rose between 2022 and 2025 as they expanded operations. Microsoft improved its water-use efficiency by 25%, and AWS by 37%, according to their latest sustainability reports. However, comparisons are difficult because there is no industry-wide standard for reporting environmental, social, and governance (ESG) metrics.

Elon Musk’s SpaceX, which became a major data-centre operator after acquiring his AI company xAI, has never published an ESG report. In June, ratings agency MSCI gave SpaceX its lowest ESG score. This opacity fuels scepticism among critics who argue that companies are more concerned with public relations than genuine conservation.

“Because water is generally much cheaper than electricity,” companies have less incentive to cut water use on cost grounds alone, said Shaolei Ren, an engineering professor at the University of California, Riverside. “There are incentives,” he added, but these are driven more by public opposition to data centres across the United States.

Another challenge is retrofitting older facilities. However, Minh K. Le, who leads data-centre research at Rystad Energy, notes that older data centres are generally smaller and less powerful than the new mega-facilities, so they require less cooling in the first place.

The water used directly by data centres is only part of the story. Generating the electricity that powers them, and manufacturing the chips and servers, also consumes significant water. This broader footprint is often overlooked in corporate sustainability claims.

For Europe, the issue is particularly pertinent. As AI-driven data centres strain grids—Ireland is even weighing nuclear options—the water-energy nexus becomes a critical policy question. European nations, from the Netherlands to Spain, are already facing water scarcity, and the growth of data centres could exacerbate tensions. The industry’s promise of water-free cooling may offer a path forward, but the energy penalty could undermine climate goals.

Ultimately, the tech sector’s ability to deliver on its bold claims will be tested as public scrutiny intensifies. The challenge is not just technical but economic and political. As Masley put it, “There’s no free lunch.”

More from this story

Next article · Don't miss

Dubai trains 14,000 firms in agentic AI through new academy

Dubai Chambers has launched an e-learning academy to train 14,000 member companies in agentic AI. The initiative aims to help businesses adopt autonomous AI systems to streamline operations and improve decision-making.

Read the story →
Dubai trains 14,000 firms in agentic AI through new academy