Nvidia unveils expanded digital twin plan for AI factories

The Omniverse Blueprint platform integrates power, cooling and networking modeling of future data centers.

Nvidia announced a major expansion of its Omniverse Blueprint -an integrated project to create digital twin AI factories, which is now available in preview mode. The announcement was made at Computex 2025 in Taipei.

The goal of the project is to provide a single platform for engineering design, simulation and optimization of AI factory infrastructure – from power and cooling to network architecture. The ecosystem already includes Siemens, Delta Electronics, Jacobs, Cadence, Schneider Electric, Vertiv and ETAP, which are integrating their digital models into the OpenUSD environment.

What is the Omniverse Blueprint for AI factories

The platform is built on the Nvidia GB200 NVL72 reference architecture and uses OpenUSD -an open format for describing 3D scenes. This enables the creation of unified digital twins of data centers, including all components from power supply to ventilation to compute racks.

key features

• Simulation at the level of physical laws (SimReady) – an exact digital copy of the infrastructure;
• Integration with Cadence Reality and ETAP – thermal and electrical modeling before construction;
• Support for large-scale AI workloads – from generative AI to industrial simulations;
• Accelerated design and reduced downtime risks

Partnerships and SimReady standardization

The SimReady project is creating a single standard for modeling components for digital twins. It gives developers and data center owners a set of requirements and procedures to create critical infrastructure in 3D, including thermal and electrical load behavior.

Companies such as Siemens and Delta Electronics are building their digital equipment models in OpenUSD. Jacobs is involved in end-to-end workflow testing, while Vertiv and Schneider Electric are supplying power and mechanical system models.

“AI factories are rapidly changing the structure of the entire digital infrastructure. With Omniverse Blueprint and SimReady, customers can test power efficiency in advance,” said Tanj Khandelwal, CEO of ETAP.

How it helps AI development

• Analysis of airflow in a data center
• Modeling power consumption under AI loads
• Training physical AI agents in an industrial environment

Nvidia says it is digital twins with physically accurate behavior that will be the basis for designing the next generations of data centers and AI services.

“Cadence Reality’s integration with Omniverse Blueprint is transforming the engineering cycle – we can design AI factories faster and more efficiently,” said Ben Gu, Cadence’s vice president of R&D.

The article Nvidia unveils expanded digital twin plan for AI factories was first published on ITZine.ru.