Schneider, NVIDIA, AVEVA set roadmap for AI data centers
The pressure to scale artificial intelligence is reshaping the infrastructure that supports it.
Schneider Electric, NVIDIA and AVEVA presented a technical roadmap to standardize the lifecycle of AI-designed data centers.
The growing demand for AI processing capacity is pushing the data center industry to rethink its technical foundations. Compared with traditional infrastructure, these environments require higher compute densities, new power architectures and advanced cooling systems, driving structural changes in their design and operation.
In this context, Schneider Electric, NVIDIA and AVEVA presented a technical roadmap aimed at standardizing the lifecycle of next-generation AI infrastructure. The announcement focuses on the development of validated blueprints that enable the design and operation of gigawatt-scale “AI factories,” addressing simulation and maintenance challenges in modern data centers.
Simulation and digital twins.
The approach incorporates the use of digital twins to create virtual replicas of facilities in order to test and optimize performance before construction. This involves assessing variables such as thermal dynamics, airflow and control systems in simulated environments, helping reduce risk and accelerate deployment. In parallel, AI tools are integrated as assistants capable of autonomously detecting and resolving technical failures.
Infrastructure technical design.
The proposed design for data centers based on NVIDIA Vera Rubin establishes a technical framework aimed at supporting high-density compute workloads, integrating power and cooling in a coordinated way.
Validated by Schneider Electric together with NVIDIA, the model includes the use of higher voltages to optimize electrical distribution, cooling systems capable of operating at up to 45°C to reduce energy consumption, and a rack organization that enables resource sharing without compromising electrical stability.
These capabilities have previously been evaluated through digital simulations focused on thermal and energy behavior.
Operations-focused AI.
In parallel, the digital twin architecture developed by AVEVA with NVIDIA introduces a full lifecycle approach for the design, simulation and operation of large-scale AI infrastructure. Within this framework, Schneider Electric promotes the creation of “SimReady” assets and digital twins on NVIDIA Omniverse, integrating engineering tools that enable multidomain simulations prior to physical construction.
In practice, this approach not only accelerates “time-to-token,” but also enables optimized design decisions, reduced risk and improved operational efficiency from early project stages.
Additionally, Schneider Electric has advanced testing with the NVIDIA Nemotron model to enable agentic AI capabilities, using real-time IoT data to diagnose issues, identify root causes and autonomously recommend actions. This development aims to speed up incident resolution and reduce unnecessary operational interventions.
These developments are part of an ongoing collaboration between the companies, which also includes initiatives such as participation in the Alliance for OpenUSD, the promotion of 800 VDC power architectures and the development of reference designs for advanced AI computing platforms, reinforcing a joint technical approach to the evolution of data centers.
|
|
