LiquidStack

New accelerator generation. Same liquid cooling infrastructure. Different thermal and flow demands. A GPU refresh can change hydraulic behavior and make yesterday’s control settings less appropriate, even when the cooling equipment itself hasn’t changed. Learn what operators should reassess as AI hardware and workloads evolve.

Your liquid cooling system may be telling you something long before an alarm sounds. A gradual rise in pump current or a shift in differential pressure can signal changing conditions weeks before performance deteriorates. What should data center operators be watching? Explore the signals and operating practices that help sustain liquid cooling performance as AI infrastructure evolves.

LiquidStack’s GigaModular CDU is among the first offerings to qualify as an NVIDIA DSX Ready CDU. Our involvement began before product qualification. LiquidStack collaborated with NVIDIA AI Infrastructure during the development of the CDU requirements supporting the initial NVIDIA DSX Ready program, bringing practical experience from designing and deploying direct-to-chip liquid cooling for high-density infrastructure. The qualification helps AI factory builders identify a trusted cooling building block and evaluate its fit within the wider architecture with greater confidence. Built as a centralized, scalable platform, GigaModular provides a coordinated approach to cooling across multi-megawatt deployments. As a Trane Technologies company, LiquidStack combines specialist liquid-cooling engineering with broader thermal-management expertise, manufacturing scale and global service capabilities. Read the full story: https://lnkd.in/eURhC8iS #DSXReady #AIFactory #LiquidCooling #AIInfrastructure #DataCenters

Liquid cooling design decisions shape performance, uptime, and cost long after deployment. Get practical guidance from LiquidStack and Leading Edge Design Group on how to validate liquid cooling systems for stable, full-performance operation in high-density AI environments.

Introducing CDU 2.X, our brand new architecture-agnostic CDU built for today's AI hardware, with headroom for what comes next. CDU 2.X gives operators more than headline megawatts. It delivers 3,750 lpm at 3.5 bar available head pressure, creating hydraulic headroom as flow-per-kW and rack differential-pressure requirements rise. Its configurable architecture is designed to support current and successive AI compute platforms across end-of-row and rack-adjacent deployments, with site-specific options for power, redundancy, controls, piping and monitoring.

Introducing CDU 2.X, our brand new architecture-agnostic CDU built for today's AI hardware, with headroom for what comes next. CDU 2.X gives operators more than headline megawatts. It delivers 3,750 lpm at 3.5 bar available head pressure, creating hydraulic headroom as flow-per-kW and rack differential-pressure requirements rise. Its configurable architecture is designed to support current and successive AI compute platforms across end-of-row and rack-adjacent deployments, with site-specific options for power, redundancy, controls, piping and monitoring. LiquidStack can help configure CDU 2.X around your rack requirements, facility infrastructure, redundancy strategy, and operating model.