Schneider Electric advances designs and AI tools with NVIDIA for gigawatt-scale AI factories

Schneider Electric collaborated with NVIDIA and AVEVA to advance methods for developing and managing Artificial Intelligence (AI) data center infrastructure at a gigawatt scale. The development focuses on validated designs, digital twin integrations, and AI-powered operational tools intended for large-scale AI facilities.

The collaboration aims to address operational complexities and enhance efficiency in managing power and cooling systems within expansive AI data centers. These efforts include lifecycle digital twin architectures designed to optimize performance and support evolving AI workloads through simulation and digital management.

The NVIDIA Vera Rubin reference design was created to validate power distribution and cooling for NVIDIA’s rack-scale system architecture, incorporating higher supply voltages and thermal control parameters. AVEVA’s software contributes lifecycle digital twin capabilities integrated within the NVIDIA Omniverse environment, facilitating system simulation, validation, and iterative design optimization. Additionally, Schneider Electric tested the NVIDIA Nemotron model to enable agentic AI functionality for alarm management, which autonomously interprets system alarms and recommends corrective actions using real-time Internet of Things (IoT) data.

Work completed includes the validation of the Vera Rubin design with electrical system and Computational Fluid Dynamics (CFD) models and embedding AVEVA’s engineering and operations software within the Omniverse DSX Blueprint. The testing of agentic AI for alarm management also represents a move toward autonomous, software-defined operation within data center management.

Manish Kumar, Executive Vice President at Schneider Electric, said, “Delivering AI at scale requires tightly integrated electrical, cooling and digital architectures that can support both unprecedented performance demands while maintaining peak energy efficiency. By combining advanced software, digital twins and validated reference designs, operators can simulate and optimize infrastructure before a single rack is deployed.” Vladimir Troy, vice president of AI infrastructure at NVIDIA, said, “Together, NVIDIA and Schneider Electric are providing the power, cooling, and digital twin architectures needed to accelerate time-to-token for our customers worldwide.”

These initiatives build upon prior joint efforts including support for NVIDIA’s evolving rack systems and integration with OpenUSD and Omniverse assets. These developments are intended to facilitate future deployments and operations of AI data centers through validated models and simulation tools.