Inside Salute’s Liquid Cooling Service for AI Data Centres

The shift to AI workloads is forcing data centre operators to rethink how they manage cooling systems, with liquid cooling becoming standard for high-density deployments. 

Salute, which provides lifecycle services for data centres, has now launched an operations service for direct-to-chip liquid cooling at Nvidia GTC Washington DC, targeting facilities running AI and high-performance computing workloads.

The company has secured partnerships with Applied Digital, Compass Datacenters and SDC, which have adopted the operational model for their infrastructure.

Erich Sanchack,CEO at Salute, says: “Salute has achieved a long list of industry firsts that have made us an indispensable partner for 80% of companies in the data centre industry.

“This first-of-its-kind DTC liquid cooling service is a major new milestone for our industry that solves complex operational challenges for every company making major investments in AI/HPC.”

The need is pressing, because AI and HPC data centres run at power densities that dwarf traditional enterprise and cloud facilities. 

Direct-to-chip liquid cooling – which circulates coolant directly onto processors rather than cooling the air around them – has become essential to manage the heat. 

But that creates operational headaches, as temperature spikes from cooling interruptions can damage equipment, whilst leaks present serious hazards when liquid mixes with electrical systems. 

Downtime translates directly into revenue loss and service level agreement penalties.

Salute’s frameworks for direct-to-chip cooling operations: Explained

The service includes a design and operational assessment tailored to each facility’s specifications, along with commissioning support to ensure systems work as intended. 

Salute also maintains a library of direct-to-chip liquid cooling practices that it continuously updates based on work with Nvidia, coolant distribution unit manufacturers, chemistry suppliers, leak management vendors, original equipment manufacturers, hyperscale cloud providers and more than 20 companies deploying AI workloads.

The operational model gives data centre teams emergency procedures, maintenance protocols and standard operating procedures for every aspect of direct-to-chip operations, including chemistry management, leak management, safety protocols, coolant distribution unit management and risk mitigation. 

These are written by operational experts who have managed some of the world’s largest supercomputers, according to Salute. 

The service includes training through classroom instruction, online learning and laboratory certification programmes, plus operational support to help data centre operators expand their teams and scale their AI and HPC environments.

Data centre providers who came on board early will use the service to support an estimated 260MW of capacity in the coming months, expanding to an estimated 3,300MW by the end of 2027. 

John Shultz, Chief Product Officer (CPO) for AI and Chief Learning Officer at Salute, sees that as validation: “This service has already proven to be a game changer for the many data centre service providers who partnered with us as early adopters,” he says. 

“By successfully mitigating the risks of DTC liquid cooling, Salute is enabling these companies to rapidly expand their AI/HPC operations to meet customer demand. 

“These companies will rely on this service from Salute to support an estimated 260 MW of data centre capacity in the coming months and will expand that to an estimated 3300 MW of additional data centre capacity by the end of 2027. 

“This is an enormous validation of the impact of our service on their ability to scale. Now other companies can benefit from this service to protect their investments in AI.”

Applied Digital and SDC adopt operational model across sites

Applied Digital, which builds digital infrastructure for AI deployments, has woven the service into its operations.

Laura Laltrello, CPO at Applied Digital, explains why: “Applied Digital is at the forefront of building innovative digital infrastructure that meets the explosive growth in AI demand. 

“High-density environments that utilize liquid cooling require an entirely new operational model, which is why we partnered with Salute to implement operational methodologies customised for our facilities and our customers’ needs,” she says. 

“We are delivering truly world-class operations for AI/HPC computing that our customers can trust to mitigate risk, lower TCO, ensure performance and protect their investments.”

SDC has rolled out the model across multiple sites globally as its customers expand AI and HPC operations. 

Walter Wang, Founder at SDC, says the service solves a critical problem: “Salute is making it possible for SDC’s customers to accelerate AI deployments with zero downtime, thanks to the proven operational model, real-world training and other best practices.

“Salute’s ability to perform this at scale under rapid timelines is critical to our customers, who are expanding their AI/HPC operations across multiple sites globally. 

“Salute has long been a trusted partner in the industry and this new service clearly makes them the go-to partner for companies building the foundation for AI around the world.”

Visit the website here.