On October 27, 2026, Schneider Electric, AVEVA, Nscale and Start Campus announced a successful Immersive Twin pilot focused on Nscale’s operations at SIN01, part of the SINES Data Campus in Portugal. The companies describe an OpenUSD-based digital twin that connects workload requirements and facility data with power, cooling and process simulations.
What the SIN01 digital twin connects
The reported model brings together workload requirements, engineering and operational data, and power and cooling simulations. It also links IT (information technology) and OT (operational technology) data to support what-if scenarios involving power, cooling and process behavior before physical changes at the site.
The announcement describes connections to Schneider Electric, ETAP, AVEVA and third-party technologies. The model’s practical purpose is to let teams examine how a proposed change might affect facility systems before they make that change on site.
What NVIDIA’s broader DSX architecture includes
NVIDIA’s general DSX architecture describes a path from CAD and BIM models to validated OpenUSD assets, called SimReady assets, with metadata for electrical systems, cooling and connection points. Its simulation layer covers network, grid, thermal, site and electrical domains.
An AI-accelerated surrogate model can return approximate results alongside full-fidelity solvers. In the runtime layer, the AI Factory Digital Twin (AIF-DT) includes Omniverse Kit rendering, application streaming, simulation-data routing, an AI agent, a data-lake database and USD storage. These components describe the broader Blueprint architecture; the reported SIN01 pilot is described through its workload, facility, power, cooling and process modeling.
NVIDIA’s DSX system architecture lays out the reference design in more detail.
The changes the pilot is intended to help teams assess
Start Campus CEO Robert Dunn described using a digital twin to model how increased rack density affects the power train and liquid-cooling loop before physical changes. That approach is intended to help teams assess new hardware generations and agree on a configuration before building it.
Nscale’s stated goals for the work include faster time to token, greater token efficiency, improved operations, more flexibility and lower risk when changing its technology stack.
