In this Vertiv™ power and cooling infrastructure, which will be seamlessly integrated with the NVIDIA Blackwell platform, the architecture will simplify and accelerate the deployment of AI workloads in new and existing data centers, while reducing risk and facilitating standardization across locations. The architecture's comprehensive critical power infrastructure is designed to significantly reduce wasted power by aligning AI nodes with data center capacity blocks. The hybrid air-liquid cooling infrastructure leverages the interdependent impact of the two approaches to efficiently manage high-density heat dissipation. Furthermore, the design allows for optional Open Compute Project (OCP)-inspired systems, such as DC power trays.

The reference architecture is part of the Vertiv™ 360AI reference design portfolio, geared toward data center retrofits or new builds, and designed to help customers choose from multiple integrated solutions to power and cool AI platforms and other high-performance computing applications. Key benefits of the GB200 NVL72 reference architecture, developed in collaboration with NVIDIA, include:

Rapid deployments and retrofits: To facilitate the use of pre-configured modules and factory integration, VertivTM MegaModTM CoolChip delivers turnkey critical AI infrastructure up to 50% faster than on-site deployments.

Power management in less space: Employing Vertiv's advanced power technology, including the Vertiv™ Trinergy™ uninterruptible power supply (UPS) and the Vertiv™ EnergyCore lithium-ion battery cabinet, the design provides industry-leading reliability and energy management efficiency, occupying up to 40% less space than traditional solutions.

Efficient cooling: By integrating low GWP (Global Warming Potential) liquid and air cooling technologies on a large scale, including the Vertiv™ AFC chiller, Vertiv™ Liebert® chilled water-based room cooling systems, and Vertiv™ XDU refrigerant distribution units, annual cooling costs can be reduced by up to 20% compared to conventional solutions.

Dynamic workload management: Integrated load averaging via lithium-ion batteries and the new generation of UPSs provide support for dynamic GPU workloads.

Installation and operations services: Thanks to its industry-leading scale and reach, and its nearly 4,000 service engineers globally, Vertiv is the most reliable service partner for the data center lifecycle, as well as the expert in retrofits and new builds of highly complex systems.

As businesses incorporate the potential of AI at an unprecedented rate, Vertiv is shaping the future of critical power and cooling to support accelerated computing, through the most comprehensive portfolio of critical digital infrastructure that enables the operation of AI-ready infrastructures capable of handling their unique requirements and other accelerated computing applications.

The collaboration between Vertiv and NVIDIA establishes a roadmap for the future of technical development and facilitates the deployment of accelerated computing on a large scale.

More information