The recent NVIDIA GTC 2026 conference in San Jose California was abuzz with more than 30,000 attendees and Schneider Electric as a Diamond sponsor. The event featured many revelations including the shift from experimental, proof of concept working model AI to production grade AI hitting a key inflection point. NVIDIA officially introduced the next-gen Vera Rubin platform to boost AI performance 3-4x over the Blackwell platform in AI factories.

The Vera Rubin platform will need high-density liquid-cooled systems with applications including industrial digitalization, combining generative AI with robotics, autonomous machines, and digital twins. 

Agentic AI 

The token economy was a hot topic: where tokens are recognized as the output of AI factories and the evolution of the Gigascale AI factory reference design to the DSX blueprint. NVIDIA positioned Agentic AI as the core of the “Next Industrial Revolution,” with Nemotron open models serving as the primary engine for autonomous reasoning and planning.

Rubin this year and years to follow  

To enable companies to prepare for the installation of Vera Rubin, Schneider Electric introduced the world’s most comprehensive reference design including special drawings, bills of material, schematics, and performance specifications. The new reference design supports NVIDIA’s new rack-scale systems (Vera CPUs, Rubin GPUs) with optimized power (480 VAC) and liquid cooling, delivering immense AI performance while maximizing tokens-per-watt.  

Created for the NVIDIA Vera Rubin NVL72 racks, the validated reference design covers power and cooling and is integrated with Schneider Electric’s controls reference designs and supports operation at MaxQ 188 kw/rack and MaxP 227 kw/rack that delivers the most tokens.

The latest liquid cooling tech

This design also allows higher Technology Cooling System (TCS) liquid temperature loop supply temperature of 45°C. Running at this temperature provides enhanced efficiency by enabling the operation of free cooling depending on the location, temperature and humidity. The reference design features the latest liquid cooling technology including the industry-leading 2.5MW Coolant Distribution Unit (CDU) designed to cool high-density data centers reliably, at scale. The MCDU-70 is the highest-capacity CDU available from Motivair and it presents a breakthrough flexible and scalable solution for meeting the rigorous demands of gigawatt-scale AI factories.

Schneider Electric demonstrated its 800VDC power infrastructure prototype done in collaboration with NVIDIA and specifically designed for next-generation AI data centers, including the NVIDIA Vera Rubin Ultra platform. The Rubin Ultra GPUs will be housed in the new Kyber rack architecture (up to 576 GPUs per rack) and racks can reach up to 1 MW to 1.2 MW of power density.

Optimizing the DSX blueprint

NVIDIA’s Omniverse DSX is a comprehensive, open blueprint for designing and operating gigawatt-scale AI factories, validated at the new AI Factory Research Center at Digital Realty’s site in Manassas, Virg. Schneider Electric’s AVEVA and ETAP collaborated with NVIDIA jointly to advance the design, simulation, build, operation, and optimization of the DSX Blueprint for AI Factories. Highlights of this collaboration include:

• An extension of the methodology and expertise that AVEVA is known for across many industrial sectors.
• With AVEVA Process Simulation, customers can model and run simulations of advanced liquid-cooling networks for AI Factories to refine designs and maximize cooling efficiency. 
• ETAP leverages Omniverse acceleration libraries for real-time electrical infrastructure performance tracking, thermal simulation, and predictive maintenance.
• Customers can use AVEVA Operations Control and Unified Operations Center to manage data center infrastructure comprised of electrical (UPS, switchgear, PDU, generator etc.), mechanical (chiller, CDU etc.), and safety systems into one scalable unified platform.

Using Nemotron Agentic AI Model for alarm management services

Schneider Electric announced during the event that it has successfully tested the NVIDIA Nemotron Agentic AI model to power a new Agentic AI alarm management capability. This service’s advancement addresses a longstanding challenge in the data center industry: interpreting alarms at a system level to identify root causes and determining the appropriate corrective actions.

Leveraging real-time streaming IoT data across multiple systems, Schneider Electric’s Agentic AI autonomously analyses, diagnoses, and recommends actions using a suite of integrated tools. Working alongside expert technicians, the technology delivers faster and more consistent issue resolution, reduces unnecessary dispatches, and enhances operational resilience. This milestone reinforces Schneider Electric’s commitment to redefine asset performance management through cutting-edge AI innovation.

The Superbowl of AI conferences

Many call the NVIDIA San Jose GTC the “Superbowl of AI Conferences” or the “Woodstock of AI” and this year did not disappoint (plus the real U.S. football Superbowl was played down the street in Santa Clara this year). The event was a culmination and announcement of the result of the many workstreams between NVIDIA and Schneider Electric for the advancement of AI data centers. Schneider will continue to collaborate with NVIDIA to make designing, simulating, and operating gigawatt-scale AI factories as easy as possible as the move to inference accelerates.