Nvidia and Schneider Electric team up for AI data center innovation

The rapid growth of artificial intelligence is driving unprecedented demands on data center infrastructure, prompting new solutions for power and cooling challenges.

Strategic partnership details: Schneider Electric and Nvidia have joined forces to create an innovative data center reference design that addresses the intense computational requirements of modern AI systems.

• The new design supports liquid-cooled, high-density AI clusters capable of handling up to 132 kW per rack, specifically optimized for Nvidia’s latest AI chips including the GB200 NVL72 and Blackwell series
• The collaboration integrates liquid-to-liquid coolant distribution units and direct-to-chip liquid cooling options, making complex cooling systems more manageable
• Schneider Electric’s software tools enable customization while maintaining adherence to high-density standards

Technical innovations: The partnership introduces significant advancements in data center power management and cooling efficiency.

• A new Galaxy VXL UPS (Uninterruptible Power Supply) features a modular design supporting up to 1.25 MW for critical loads, expandable to 5 MW when four units operate in parallel
• The compact UPS design reduces embedded carbon throughout its lifecycle, from manufacturing to installation and maintenance
• The reference design simplifies the implementation of liquid cooling systems, addressing one of the major challenges in scaling AI infrastructure

Industry impact: The collaboration represents a significant step forward in addressing the growing demands of AI computing infrastructure.

• Jensen Huang, Nvidia CEO, emphasizes the importance of stable and resilient infrastructure for advancing accelerated computing and AI development
• Marc Garner, Schneider Electric SVP, highlights the need to continuously adapt infrastructure to support new GPU technologies as they come to market
• The partnership aims to enhance energy efficiency and sustainability across hyperscale, colocation, and enterprise data centers

Looking ahead: The growing computational demands of AI workloads will likely continue to push the boundaries of data center design, making innovations in cooling and power management increasingly crucial for sustainable AI development. The success of this partnership could set new standards for how data centers are designed and operated in the AI era.