News Categories

Liquid cooling manifold functions as key flow distribution piping device inside server racks. Situated between cold plate CDU and cold plates, it evenly distributes coolant to server modules and dissipates heat from high-density electronic components, maintaining stable device performance. Sophisticated internal channel design ensures smooth and balanced liquid circulation, laying solid foundation for reliable data center liquid cooling system operation.

As AI workloads continue to surge and push server heat density to new heights, liquid cooling has become the industry-standard solution for safeguarding reliable data center operation. However, many operators still hold the misconception that upgrading existing facilities to deploy a full data center liquid cooling system requires excessive capital investment, large-scale construction and mandatory service downtime.

As a global leader in critical digital infrastructure and continuity solutions, Vertiv has launched Vertiv™ CoolLoop RDHx chilled water rear door cooling worldwide. Designed for AI and high performance computing workloads, it delivers efficient rack-level cooling and works seamlessly with cold plate liquid cooling. As a robust data center liquid cooling solution, it greatly improves overall heat dissipation performance.

A real-world data center operation case reveals a key industry transformation challenge. A senior precision air conditioning engineer with eight years of experience, after being reassigned to a liquid cooling facility, instinctively tried detecting temperatures at air outlets — only to realize liquid cooling for data center environments have no air outlets at all. This incident illustrates that the biggest hurdle for air cooling O&M engineers transitioning to liquid cooling is not technical capability, but ingrained operational habits and muscle memory from the air cooling era.

Cooling Distribution Unit (CDU) serves as the core thermal management device for modern liquid cooling infrastructure. It transfers heat generated by IT equipment to cooling media and circulates the liquid to complete overall heat dissipation. Through built-in pumps and pipeline networks, the CDU delivers coolant to heat-producing devices such as servers and storage hardware, absorbing operational heat and forming a continuous cooling cycle. As foundational equipment for high density server cooling and high-performance computing scenarios, CDU supports reliable and efficient heat dissipation for upgraded and newly built data center facilities.

Driven by rapid AI computing growth and high-density server deployment, traditional air cooling can no longer support high-heat-flux chip heat dissipation. Liquid cooling has become the mainstream solution for modern data centers, thanks to its high efficiency, low energy consumption and low noise performance. As the core hub of liquid cooling infrastructure, the Cooling Distribution Unit (CDU) determines system operation efficiency, stability and energy-saving performance, serving as the key enabler for scalable liquid cooling high density computing.

Fueled by advances in AI, machine learning and high-performance computing (HPC), data center thermal density continues to surge, outstripping the cooling capacity of traditional air systems. Liquid cooling has become the preferred solution for high-power-density scenarios, with performance advantages over conventional air cooling clearly reflected in liquid cooling vs air cooling data center comparisons.

Driven by rapid AI computing iteration and rising server power density, traditional air cooling solutions can no longer sustain the heat dissipation demands of modern data centers. Liquid cooling has emerged as the new industry standard for infrastructure upgrade, delivering ultra-high heat exchange efficiency and significant energy-saving benefits. As core infrastructure equipment for high-density computing, liquid cooling systems have become essential technical knowledge for data center operation and maintenance teams. Standardized deployment and scientific daily maintenance are critical to eliminating operational risks and maximizing system performance, while answering the industry’s key concern over How much energy does liquid cooling save and optimizing overall data center liquid cooling PUE.

With the rapid iteration of AI computing, machine learning and high-performance computing, modern data centers are facing surging thermal density, making traditional air cooling unable to adapt to high-load chip heat dissipation demands. As the core hub of liquid cooling systems, the Cooling Distribution Unit (CDU) undertakes key heat exchange tasks between primary and secondary loops, serving as the core guarantee for safe and efficient operation of high density server cooling scenarios. Integrating professional design principles, engineering calculation, standardized selection and full-cycle operation and maintenance, CDU has become an essential core device to support the stable construction of liquid cooling for AI GPU data center.

Two-phase cold plate cooling represents an advanced form of cold plate liquid cooling. Using fluorinated compounds as refrigerants can significantly boost the reliability of liquid cooling systems. Once leaked, these fluids turn into non-conductive gas and will not damage servers. Most fluorinated refrigerants have low boiling points; they vaporize after absorbing heat inside cold plates, greatly enhancing heat dissipation efficiency, making this technology ideal for high density server cooling and high-demand AI data center cooling scenarios. According to NVIDIA’s technical documents, applicable refrigerants include R134a, R513A, R515B, R471A, 1233zd, R1234yf and 3M Novec HFE-7000 series. When R134a is adopted for two-phase cold plate cooling, it handles 2.5kW heat load per GPU and 175kW per rack, with a maximum heat dissipation capacity of 8 W/(cm²·℃).