12 Days of Data Centre Christmas: December 2024
In December 2024, Microsoft has announced plans to implement water-free cooling systems across its new data centre developments from August 2024, marking a fundamental shift in its approach to thermal management for high-density AI workloads.
The new design eliminates water evaporation for cooling purposes through the implementation of chip-level cooling solutions, a technology that provides precise temperature control through a closed-loop system. It is expected to reduce water consumption by 125 million litres annually per data centre facility.
Steve Solomon, Vice President of Data Centre Infrastructure Engineering at Microsoft, explains the technology: “These new liquid cooling technologies recycle water through a closed loop. Once the system is filled during construction, it will continually circulate water between the servers and chillers to dissipate heat without requiring a fresh water supply.”
The initiative forms part of Microsoft's Data Centre Community Pledge, which outlines the company's commitment to local economies and communities where it operates facilities. Steve emphasises that “protecting local watersheds is an important part of this pledge – especially in areas where water stress is growing.”
Microsoft also plans to pilot the zero-water evaporated designs at new facilities in Phoenix, Arizona and Mt. Pleasant, Wisconsin, with operations scheduled to commence in 2026. The company has initiated the implementation of next-generation cooling technology across all new data centre designs from August 2024.
The shift to next-generation data centres is expected to help reduce Microsoft’s WUE to near zero for each data centre employing zero-water evaporation. As the fleet expands, this transition will contribute to further reductions in Microsoft's fleetwide WUE.
While the current fleet continues to operate with a combination of air-cooled and water-cooled systems, the company's commitment to water conservation remains clear.
How Nokia Will Support AI Workloads at Nscale Data Centre
Nokia is building an IP network for Nscale’s new sustainable data centre to support AI workloads. The data centre, which is powered by renewable energy and optimised for energy-efficient cooling, will enable cutting-edge AI services, including Graphics Processing Unit as a Service (GPUaaS), a technology which Nscale is a global leader in.
The new IP network stands to support growing global demand for AI-driven applications, which inevitably consume significant amounts of data. The leading telco says that its Ethernet-based data centre fabric provides scalability, programmability and low-latency performance.
The solution comes at the perfect time for Nscale, who secured US$155m in Series A funding in December 2024 to expand its operations across Europe and North America. It now aims to continue accelerating the deployment of large-scale GPU clusters, which are used for AI model training.
In this way, Nscale hopes to set a benchmark for sustainable AI and high-performance computing (HPC) infrastructure. As global demand for AI-driven applications continues to grow, the level of data consumption only increases.
“Our mission is to redefine the boundaries of AI and High-Performance Computing through innovative, sustainable solutions,” says David Power, CTO at Nscale. “Nokia’s data centre fabric enables us to scale our GPU clusters while maintaining the reliability and performance needed to serve our customers with cutting-edge AI services.
“The flexibility of Nokia’s solution ensures we can bring advanced AI capabilities to market faster.”
See also in December
5 Minutes With: Natham Blom, co-CEO of Iceotope Technologies
Vertiv & Compass Datacenters: Boosting Liquid Cooling for AI
Soitec: Silicon Photonics to Power Next-Gen AI Data Centres
Explore the latest edition of Data Centre Magazine and be part of the conversation at our global conference series, Tech & AI LIVE and Data Centre LIVE.
Discover all our upcoming events and secure your tickets today.
Data Centre Magazine is a BizClik brand
