# Schneider Electric and NVIDIA expanded AI-factory engineering around power, cooling and rack systems Schneider Electric announced expanded AI-factory R&D with NVIDIA on June 11, 2025, covering power, cooling, controls and high-density racks rather than a completed facility. Canonical: https://rangoon.ai/insights/archive-schneider-nvidia-ai-factory-engineering-june-2025/ Author: Rangoon Editorial (https://rangoon.ai/insights/editorial/) Published: 2026-10-24 Updated: 2026-10-24 Event date: 2025-06-11 Tags: [Schneider Electric](https://rangoon.ai/insights/tags/schneider-electric/), [NVIDIA](https://rangoon.ai/insights/tags/nvidia/), [AI-factory engineering](https://rangoon.ai/insights/tags/ai-factory-engineering/) Source note: Historical reporting: Schneider Electric’s dated June 11, 2025 release announced expanded R&D with NVIDIA for AI-factory power, cooling, controls and high-density racks. NVIDIA’s GB200 NVL72 product record was read as current technical context for the rack platform referenced by Schneider, not as evidence of the June partnership date, a completed reference design, a deployed facility or supplier performance in every configuration. Image: AI-factory engineering layers. Original Rangoon explanatory diagram: Source boundary: dated primary release reporting; original editorial analysis. No vendor interface, photograph, measurement result, or endorsement implied. ## Key takeaways - Schneider Electric announced expanded R&D with NVIDIA on June 11, 2025 for AI-factory power, cooling, controls and high-density racks. - The release named a modular rack designed to support NVIDIA GB200 NVL72 and described electrical and liquid-cooling reference designs. - A reference design and component platform guide planning; neither source documents a completed AI factory, customer deployment or measured site result. ## The June release expanded engineering work across several facility layers [source 1](https://www.se.com/ww/en/about-us/newsroom/news/press-releases/schneider-electric-accelerates-the-development-and-deployment-of-ai-factories-at-scale-with-nvidia-68432d51b96642343f0f890b/) [source 2](https://www.nvidia.com/en-us/data-center/gb200-nvl72/) Schneider Electric announced on June 11, 2025 that it was collaborating with NVIDIA on research and development for power, cooling, controls and high-density rack systems for AI factories. The release was published during NVIDIA GTC Paris and framed the work around European AI-infrastructure ambitions. That list matters because it spans several layers that are often compressed into the phrase “AI data center.” Power equipment supplies and conditions electricity; cooling moves heat; controls connect monitoring and operation; a rack packages computing hardware and its local interfaces. The announcement says the companies were advancing R&D initiatives and would continue accelerating infrastructure work and reference designs. It does not say that every component had reached a final commercial configuration, that a full AI factory had been built, or that a customer facility was already operating the complete stack. The event is strongest as evidence of a stated co-development programme aimed at a specific class of high-density compute infrastructure. ## The GB200 reference connected facility design to a rack-scale compute platform [source 1](https://www.se.com/ww/en/about-us/newsroom/news/press-releases/schneider-electric-accelerates-the-development-and-deployment-of-ai-factories-at-scale-with-nvidia-68432d51b96642343f0f890b/) [source 2](https://www.nvidia.com/en-us/data-center/gb200-nvl72/) Schneider Electric said a new Open Compute Project-inspired rack system had been developed to support NVIDIA’s GB200 NVL72 platform and that the companies had co-developed electrical and liquid-cooling reference designs. NVIDIA’s current GB200 NVL72 product page describes a liquid-cooled rack-scale system with 36 Grace CPUs and 72 Blackwell GPUs connected through an NVLink domain. This technical record clarifies why facility power and liquid cooling were material to the partnership: the referenced compute platform is a rack-scale system rather than a conventional air-cooled server by itself. It does not retroactively prove that this exact current web-page configuration, performance figure or product availability applied unchanged on June 11, 2025. Nor does it prove that Schneider’s rack solution was installed with every GB200 system. The public sources support a design relationship around named components and interfaces; deployment choices, certification and integration details remain configuration-specific. ## Reference designs are planning artefacts, not operating-facility evidence [source 1](https://www.se.com/ww/en/about-us/newsroom/news/press-releases/schneider-electric-accelerates-the-development-and-deployment-of-ai-factories-at-scale-with-nvidia-68432d51b96642343f0f890b/) [source 2](https://www.nvidia.com/en-us/data-center/gb200-nvl72/) The announcement uses the language of reference designs and an approved coolant-distribution-unit vendor. A reference design can give operators an engineering starting point for electrical paths, cooling topology, rack placement and control interfaces. It is useful because high-density AI infrastructure has interdependent thermal, electrical and software constraints. Yet an approved component or a published design does not establish a project’s site power availability, utility interconnection, pipework quality, commissioning result, workload behavior, maintenance response or uptime. Schneider also announced a prefabricated modular EcoStruxure Pod Data Center and a rack infrastructure product on the same day. Product availability and a design template can shorten planning work, but the release does not identify a customer order, facility location, construction milestone, installed capacity or measured operating result. Reporting should preserve the engineering significance without treating a blueprint as a finished, independently validated AI factory. ## Supplier specifications and roadmap language need configuration-level follow-up [source 1](https://www.se.com/ww/en/about-us/newsroom/news/press-releases/schneider-electric-accelerates-the-development-and-deployment-of-ai-factories-at-scale-with-nvidia-68432d51b96642343f0f890b/) [source 2](https://www.nvidia.com/en-us/data-center/gb200-nvl72/) NVIDIA’s current product page contains performance and efficiency claims for GB200 NVL72, including comparisons with H100 infrastructure and statements about liquid cooling. Those claims are NVIDIA’s specifications and comparison conditions; they are not measurements of Schneider Electric’s June 2025 partnership work or of a particular customer site. The page also reflects a living product surface that includes later GB300 material, making date discipline especially important. Schneider’s release names future-facing R&D, product rollouts and reference designs, but it does not publish a facility-level energy-use reading, water figure, capital cost, construction timeline or independently audited performance outcome. It also does not endorse a particular AI model or application. The historical event supports a precise account of companies announcing expanded engineering collaboration across facility infrastructure and accelerated compute. Claims about a delivered project, grid effect, cooling efficiency or AI output need later evidence tied to a named design, site and operating period. ## Sources 1. [Schneider Electric: AI factories at scale with NVIDIA](https://www.se.com/ww/en/about-us/newsroom/news/press-releases/schneider-electric-accelerates-the-development-and-deployment-of-ai-factories-at-scale-with-nvidia-68432d51b96642343f0f890b/) 2. [NVIDIA GB200 NVL72 product record](https://www.nvidia.com/en-us/data-center/gb200-nvl72/) ## Related reading - [NVIDIA’s memory codesign post highlights a practical boundary between capacity and usable agent throughput](https://rangoon.ai/insights/nvidia-memory-full-stack-codesign-2026/) - [NVIDIA supply-chain capacity plans make packaging, test and optics explicit infrastructure constraints](https://rangoon.ai/insights/nvidia-supply-chain-capacity-reservations-2026/) - [TensorRT-LLM 0.17 added Blackwell and an experimental PyTorch workflow](https://rangoon.ai/insights/archive-tensorrt-llm-0-17-blackwell-pytorch-february-2025/) ## More information - [Documentation index](https://rangoon.ai/llms.txt) - [AI access and policies](https://rangoon.ai/ai/) - [Source repository](https://github.com/hypler-dev/rangoon)