In a significant move to support the growing demand for AI infrastructure, Vertiv has partnered with NVIDIA to co-develop a comprehensive reference architecture designed for the NVIDIA GB200 NVL72 platform. This collaboration marks a pivotal step in transforming traditional data centres into AI factories capable of handling the immense computing needs of AI applications across enterprises.
The newly released 7MW reference architecture offers a blueprint for the deployment of NVIDIA’s advanced, liquid-cooled rack-scale platform. With the ability to support up to 132kW per rack, the architecture is built to optimise deployment speed, performance, and energy efficiency while allowing for scalability in both new and existing data centres. For businesses adopting AI technologies, this innovation promises not only increased flexibility but also cost reductions and enhanced performance.
Giordano Albertazzi, CEO of Vertiv, emphasized the strategic importance of this partnership, stating, “As a leader in critical power and cooling infrastructure, Vertiv is uniquely positioned to support the NVIDIA GB200 NVL72 platform,” he said. “Our collaboration will enable customers to deploy AI data centres faster and more efficiently while preparing them for future workloads.” This partnership, he added, aligns with Vertiv’s mission to help businesses tackle the growing complexity and energy demands of AI workloads.
The architectural design goes beyond mere power and cooling solutions. It includes hybrid liquid- and air-cooling systems designed to handle high-density heat generated by AI workloads. Moreover, the architecture is modular, allowing for rapid deployment and integration into existing infrastructure, a key advantage as companies retrofit their data centres for AI.
NVIDIA’s CEO, Jensen Huang, remarked on the evolving needs of data centres: “As the industry moves toward generative AI and accelerated computing, the architectures we need are far more advanced than those used for general-purpose computing. Vertiv’s cooling and power technologies are critical to realizing our vision of AI factories that generate digital intelligence for industries across the board.”
A key highlight of the collaboration is the introduction of Vertiv’s 360AI portfolio, designed to help companies navigate the complexities of AI infrastructure. It includes solutions like the MegaMod CoolChip, which speeds up deployment by up to 50%, and the Trinergy uninterruptible power supply system (UPS), which delivers reliable, energy-efficient power management in 40% less space than traditional systems.
For businesses looking to optimise their operations, the new reference architecture also incorporates energy-efficient cooling technologies that could cut annual cooling costs by 20 percent. Additionally, integrated load management tools ensure that AI workloads are handled dynamically, providing real-time operational flexibility.
As enterprises embrace AI at unprecedented rates, Vertiv’s partnership with NVIDIA is poised to reshape the future of data centre infrastructure, driving innovation and supporting the next wave of AI-driven growth. By addressing the critical challenges of energy consumption, scalability, and deployment speed, this collaboration sets the stage for businesses to future-proof their operations in a world increasingly powered by AI.
This development underscores the need for companies to rethink their infrastructure as AI becomes a central component of digital transformation strategies. With the right tools and infrastructure, businesses can harness the full potential of AI to drive efficiencies, reduce costs, and stay ahead in an ever-competitive landscape.
