Immersive Twin: A Digital Twin of an AI Factory in Sines
Schneider Electric, Aveva, Nscale and Start Campus have announced an end-to-end digital twin pilot built on the Nvidia Omniverse DSX Blueprint. It is a useful case for understanding how to integrate electrical, thermal and operational models in a single platform.
According to Automazione Plus, published on 28 October 2026, Schneider Electric, Nscale (a full-stack AI cloud platform) and Start Campus (a data center developer and operator) have announced one of the first end-to-end digital twin projects built with the Nvidia Omniverse DSX Blueprint. The article presents the Blueprint as part of Nvidia DSX Sim.
What the Immersive Twin is
The pilot is called Immersive Twin and brings together technologies from Schneider Electric, Aveva and Etap in a single digital platform. Its stated purpose is to link the requirements of artificial intelligence workloads to the models of the facility's power, cooling, electrical system and operations.
Two objectives are stated: speeding up the delivery of AI factories and improving so-called token efficiency, that is, how efficiently the infrastructure turns resources into computational output. The prototype is installed at the Start Campus data center campus in Sines, Portugal, and aims to optimize Nscale's operations.
What we don't know yet
The information available is limited to the headline and summary of the original article. There are no quantified results, timelines, information on the project's status (pilot or live deployment) or details on the architecture. The precise role of Aveva and Etap is also unclear: it is not possible to say which vendor contributes which model. The claims about delivery speed and token efficiency should therefore be read as vendor statements, not verified results.
Why it matters
The case is interesting less for the announcement itself than for the model it illustrates: a multi-vendor digital twin that combines electrical, thermal and operational simulation with load demand, and that is used both at the design stage and in operation.
It is not an isolated episode. In the same week, Dassault Systèmes and Vertiv announced a collaboration on the design and delivery of AI factories. The signal is that the "simulate before you build" approach, familiar in manufacturing, is extending to AI infrastructure, where power and cooling are central constraints.
The value of a digital twin lies in getting models that today live in isolation to talk to each other.
Practical takeaways for industrial teams
Even outside data centers, the pattern can be replicated in plants and facilities. A few methodological pointers:
- Start from load demand: explicitly link production (or compute) demand to the energy and cooling models, rather than sizing them separately.
- Integrate existing models: electrical, thermal and operational models often already exist in different tools; the key work is the common interface, not a rewrite.
- Use the same twin to design and operate: a model that stays alive in operation avoids the gap between engineering and operations.
- Ask for evidence: before adopting a similar approach, verify the claimed benefits, such as delivery times and efficiency, with measurable metrics.
In summary
As of 29 October 2026, the Immersive Twin in Sines is a concrete but still thinly documented example of a digital twin that holds power, cooling and operations together. Once data on the architecture, project status and measured results become available, it will be possible to assess how transferable the model is beyond the world of AI factories.