Schneider Electric introduces AI designs for power systems

The designs developed with NVIDIA significantly accelerate time to deployment and aid operators in adopting AI-ready infrastructure solutions.

Schneider Electric announced new reference designs developed with NVIDIA that significantly accelerate time to deployment and aid operators as they adopt AI-ready infrastructure solutions.

The first reference design delivers the industry’s first and only critical framework for integrated power management and liquid cooling control systems, including Motivair by Schneider Electric liquid cooling technologies, and enables seamless management of complex AI infrastructure components. It includes interoperability with NVIDIA Mission Control—NVIDIA’s AI factory operations and orchestration software, including cluster and workload management features. The control systems reference design can also be utilised with Schneider Electric’s data centre reference designs for NVIDIA Grace Blackwell systems, enabling operators to keep pace with the latest advancements in accelerated computing, with seamless control of their power and liquid cooling systems. 

The second reference design focuses on the deployment of AI infrastructure for AI factories of up to 142 kW per rack, specifically NVIDIA GB300 NVL72 racks, in a single data hall. Created to provide a framework for the next-generation NVIDIA Blackwell Ultra architecture, the reference design includes information on four technical areas: facility power, facility cooling, IT space and lifecycle software. The design is available under configurations for the American National Standards Institute (ANSI) and the International Electrotechnical Commission (IEC) standards.

The first controls reference design can also be utilised with Schneider Electric’s data centre reference designs for NVIDIA Grace Blackwell systems — enabling operators to keep pace with the latest advancements in accelerated computing while having seamless control of their power and liquid cooling systems. 

Engineered to move at pace

As AI advances, today’s data centre operators rely on reference design frameworks to overcome the speed and deployment challenges of high-density, GPU-accelerated AI clusters. By providing validated, proven, and documented data centre physical infrastructure designs, Schneider Electric enables operators globally to design and incorporate next-generation power and liquid cooling controls infrastructure before the newest AI infrastructure solutions even arrive, while optimising for cost, efficiency, and reliability. Schneider Electric’s fully engineered reference designs are laying the foundation for the latest AI factories, empowering data centre operators to meet the moment and be ready before it even happens.

Jim Simonelli, Senior Vice President and Chief Technology Officer at Schneider Electric, said, “Schneider Electric is streamlining the process of designing, deploying, and operating advanced AI infrastructure with its new reference designs. Our latest reference designs, featuring integrated power management and liquid cooling controls, are future-ready, scalable, and co-engineered with NVIDIA for real-world applications—enabling data centre operators to keep pace with surging demand for AI.”

Scott Wallace, Director of Data, said, “We are entering a new era of accelerated computing, where integrated intelligence across power, cooling and operations will redefine data centre architectures. Centre Engineering at NVIDIA. With its latest controls reference design, Schneider Electric connects critical infrastructure data with NVIDIA Mission Control, delivering a rigorously validated blueprint that enables AI factory digital twins and empowers operators to optimise advanced accelerated computing infrastructure.”

‘Plug-And-Play’ End-To-End Controls System

The groundbreaking controls reference design connects edge devices and facility controls for energy management and liquid cooling across NVIDIA GB300 NVL72 and NVIDIA GB200 NVL72 deployments leveraging NVIDIA Mission Control. Using a “plug-and-play” architecture based on the MQTT protocol, it bridges operational technology (OT) infrastructure and information technology (IT) systems, allowing operators to, for the first time, harness data from every layer to optimize performance.

With a focus on seamless interoperability between building and AI infrastructure management software, the controls reference design establishes redundant systems for power and cooling, and introduces new guidance for measuring AI rack power profiles. Ultimately, it ensures the highest standards of uptime, reliability and peak performance for AI deployments by enabling precise, real-time management of critical power and cooling resources.

The result is an end-to-end controls system that offers:

A standardised format interface publishing power management and liquid cooling controls data for consumption by local applications and upstream data consumers and tools, including AI infrastructure management software, digital twins, AI/ML and other enterprise systems. 

The controls architecture is designed to manage redundancy across cooling and power distribution infrastructure, including coolant distribution units (CDUs) and remote power panels (RPPs), ensuring resiliency in both white and grey space environments.

 New guidance for measuring AI rack power profiles, with a focus on rack peak power and power quality monitoring.

NVIDIA GB300 NVL72 reference design for high-density AI clusters

Schneider Electric’s reference design for NVIDIA GB300 NVL72 supports the deployment of NVIDIA GB300 NVL72 based clusters with a maximum rack density of 142 kW, such as NVIDIA DGX SuperPOD with DGX GB300 systems. The data hall is purpose-built and optimised to host three NVIDIA GB300 NVL72-based clusters powered by up to 1,152 GPUs using liquid-to-liquid CDUs and high-temperature chillers.

The reference design also includes Schneider Electric’s industry-leading ETAP and EcoStruxure IT Design CFD models, allowing users to leverage digital twins to simulate specific power and cooling scenarios to optimise designs for unique applications. It builds upon a previous blueprint for the NVIDIA GB200 NVL72 as Schneider Electric continues to collaborate with NVIDIA to provide fully engineered, tested models in anticipation of the new NVIDIA GB300 NVL72 platform.

The reference designs continue Schneider Electric’s long-established collaboration with NVIDIA to meet the data centre industry’s most pressing AI demands. In addition to these new reference designs, Schneider Electric has developed nine AI reference designs for various scenarios, including prefabricated modules, retrofit data centres, and AI infrastructure purpose-built for NVIDIA GB200 NVL72 and NVIDIA GB300 NVL72-based clusters.