Schneider Electric, AVEVA pilot digital twin to boost AI data center energy efficiency

Schneider Electric, AVEVA, Nscale and Start Campus have completed a pilot of an end-to-end digital twin designed to improve energy efficiency in AI data centers by optimizing power, cooling and operations, using NVIDIA’s Omniverse DSX Blueprint.

The pilot integrates Schneider Electric, AVEVA and ETAP technologies on a unified digital platform, linking AI workload requirements with power, cooling, electrical and operational models. The approach is designed to accelerate AI Factory deployment and enable infrastructure changes to be tested virtually before implementation.

The pilot covers Nscale’s operations at the SIN01 data center within Start Campus’s 1.2-gigawatt SINES Data Campus in Portugal, which is designed to accommodate more than 66,000 NVIDIA Rubin GPUs.

Digital twin tests infrastructure changes:

Built on OpenUSD, the digital twin integrates engineering and operational data with power and cooling simulations to provide a unified view of data center performance. It enables teams to test scenarios involving power, cooling and operational processes, assess upgrades and validate designs before implementation, helping reduce procurement risks.

By connecting IT and operational technology data, the platform also supports continuous optimization of infrastructure performance and energy use.

Nscale expects the digital twin to help accelerate time to token delivery, improve token efficiency and increase flexibility while reducing risks associated with infrastructure upgrades.

Nscale CTO David Power said the digital twin allows the company to model operations from power and cooling through to token delivery, helping identify optimization opportunities before making changes on site.

Industry collaboration boosts efficiency:

Schneider Electric Executive Vice President of Software and Services Pankaj Sharma said the collaboration connects Nscale’s workload and IT requirements with data on Start Campus’s physical infrastructure to optimize performance, energy efficiency and token production.

Start Campus CEO Robert Dunn said the shared digital model enables the company to simulate rack density, power infrastructure and liquid cooling before deployment, helping accelerate implementation and improve efficiency.

Building on NVIDIA collaboration:

The pilot builds on Schneider Electric, AVEVA and ETAP’s work with NVIDIA on digital twins and AI Factory development. The three companies joined the Alliance for OpenUSD last year to advance digital twins and 3D modeling.

They are also developing reference architectures and integrated power management and liquid cooling solutions for next-generation AI factories, using NVIDIA Omniverse libraries to simulate real-world environments and improve energy efficiency.