Schneider Electric and NVIDIA Form Strategic Partnership for AI Factories

Global energy technology leader Schneider Electric has expanded its collaboration with NVIDIA and industrial software company AVEVA to develop next-generation artificial intelligence data center infrastructure. Developments announced at NVIDIA GTC in San Jose cover the entire lifecycle from design to operation of high-efficiency "AI Factory" structures at gigawatt scale.

New reference design: NVIDIA Vera Rubin
The NVIDIA Vera Rubin reference design developed under this collaboration provides NVIDIA with a validated roadmap for power and cooling infrastructure for rack-scale systems.

This design includes:

-480 VAC power distribution,

-45°C supply temperature for higher efficiency,

-Optimized IT room architecture,

-Different operating modes in GPU cabinets (MaxP and MaxQ)

and other critical infrastructure requirements aimed at improving performance and efficiency in data centers.

New era in AI Factories with digital twin
The lifecycle digital twin architecture developed through collaboration between AVEVA and NVIDIA aims to accelerate commissioning of large-scale AI Factories and improve GPU efficiency.

Thanks to this structure created on the NVIDIA Omniverse platform:

-Critical parameters such as power distribution, thermal dynamics and airflow can be simulated,

-Different scenarios are rapidly tested to optimize design,

-System validation can be performed before physical installation.

This approach accelerates engineering processes while significantly reducing "time-to-token" duration.

Autonomous operations with agent-based artificial intelligence
Schneider Electric also announced testing of the NVIDIA Nemotron model to improve alarm management in data center operations.

The next-generation agent-based artificial intelligence system:

-Analyzes IoT data in real time,

-Performs root cause analysis,

-Provides automated action recommendations.

This technology aims to increase operational efficiency while reducing incident response times.

"Error margin narrows in AI infrastructure"
Manish Kumar noted that as artificial intelligence workloads increase, the margin for error in data center design is narrowing, stating:
"To deliver artificial intelligence at scale, electrical, cooling and digital architectures must work in an integrated manner. Digital twins make it possible to optimize infrastructure before installation."

Vladimir Troy emphasized that gigawatt-scale AI Factories require a new infrastructure standard, noting that this collaboration is critical in terms of energy efficiency and predictability.

Strong ecosystem for AI data centers
The announcements represent a continuation of the long-term collaboration between Schneider Electric and NVIDIA. In this context:

-Integration of ETAP software into NVIDIA Omniverse,

-OpenUSD-based digital twin work,

-800 VDC power architecture support,

-High-density data center solutions

Joint innovation efforts continue in numerous areas.