Microsoft unveils AI data centre with closed-loop cooling system: Can it become sustainable?

Microsoft CEO Satya Nadella has defended the company's growing AI infrastructure footprint by highlighting a new data centre that he says uses no more water each year than a neighbourhood restaurant.

Speaking during his keynote address at Microsoft Build 2026, Nadella presented Microsoft's Fairwater data centre in Wisconsin as an example of how future AI facilities could reduce environmental impact while supporting the growing demand for artificial intelligence services. The comments come as technology companies face increasing scrutiny over the resources required to power and cool AI systems, particularly water and electricity consumption.

What is Microsoft's Fairwater AI data centre?

The Fairwater facility is a 315-acre campus located in Mount Pleasant, Wisconsin. Microsoft has previously described it as one of its most advanced AI infrastructure projects. According to Nadella, the facility was designed differently from traditional data centres. Instead of spreading equipment across large horizontal buildings, Fairwater uses a two-storey vertical design that stacks thousands of graphics processors and computing systems closer together. Microsoft says this approach helps reduce latency while making more efficient use of land, power and cooling resources.

The company had committed approximately $7.3 billion to the project as of September 2025, making it one of Microsoft's largest AI infrastructure investments.

How does the water-saving system work?

The centrepiece of Microsoft's environmental argument is Fairwater's cooling system. According to Nadella, engineers filled the facility's cooling infrastructure with water only once during construction. The water is then continuously recycled through a closed-loop cooling system rather than being regularly replaced with fresh supplies. This means the facility does not need to draw large amounts of water from local communities during normal operations. Microsoft claims this approach allows the data centre's annual water consumption to remain comparable to that of a single local restaurant. As AI systems become larger and more energy-intensive, cooling has become one of the biggest environmental challenges facing the technology industry. Closed-loop systems are increasingly being explored as a way to reduce water demand.

The rapid growth of artificial intelligence has triggered a global data centre construction boom. Microsoft's Azure cloud platform now operates across more than 500 data centres in over 80 regions worldwide. According to Nadella, Microsoft has added more data centre capacity in the last 18 months than it did during the entire first decade of Azure's existence.

However, that expansion has also sparked criticism from some local communities and environmental groups.

Microsoft's broader sustainability argument

Nadella acknowledged that technology companies must do more than simply build data centres. According to the Microsoft CEO, companies need to demonstrate that new facilities benefit local communities through job creation, tax revenue and support for local organisations. He also argued that protecting local electricity prices should be a priority when building large-scale AI infrastructure.

"We have to earn trust," Nadella suggested during his remarks, pointing to the need for long-term relationships with communities hosting data centres. Microsoft says sustainable growth will become increasingly important as AI demand continues to rise globally.

Why AI infrastructure is expanding rapidly

The growth of generative AI tools such as ChatGPT, Copilot, Gemini and Claude has created unprecedented demand for computing power. Training and operating advanced AI models requires massive clusters of specialised chips, particularly graphics processing units (GPUs). These systems consume significant amounts of electricity and generate large amounts of heat, making efficient cooling essential.

Industry analysts expect global investment in AI infrastructure to continue rising over the coming years as technology companies compete to build larger and more capable AI systems. This has placed data centre sustainability at the centre of discussions around the future of artificial intelligence.

Can AI growth and sustainability coexist?

Microsoft believes the answer is yes. The company argues that innovations such as closed-loop cooling systems, more efficient building designs and renewable energy investments can help reduce the environmental impact of AI infrastructure.

However, independent experts and environmental groups are likely to continue examining these claims as more facilities become operational. For now, Fairwater has become Microsoft's showcase project for demonstrating how future AI data centres could operate with lower water consumption while supporting the growing computing demands of artificial intelligence. Whether similar approaches can be scaled across the industry's rapidly expanding network of AI facilities remains one of the key questions facing the technology sector.