NVIDIA Details Industrial AI Push Across Europe’s Manufacturing Base

NVIDIA said on Oct. 21, 2026, that European industrial leaders are adopting its CUDA-X and Omniverse libraries, along with Cosmos and Nemotron open models, across engineering, factory operations and manufacturing workflows. The company also said its Factory Operations Blueprint is now in early access on GitHub, giving developers a framework for building factory manager and specialist agents that monitor operations, inspect defects and coordinate responses.

Factory operations and industrial agents

According to NVIDIA, the Factory Operations Blueprint, or FOX, is designed for always-on agents that can coordinate work across machines and connect to models including Cosmos and Nemotron, as well as data sources such as computer-aided design, product lifecycle management systems, simulation software and IT and operational technology systems. The company said the approach is intended to help industrial agents monitor factory operations in real time, correlate visual findings with machine logs and adapt from examples and documentation using Cosmos in-context learning without fine-tuning.

NVIDIA said Siemens is integrating agents built with FOX into Intelligence Center X, its agentic enterprise system, to help identify root causes and coordinate factory operations, with estimated cost savings of 20%. It added that Siemens’ Process Preparation X agents use NVIDIA Cosmos Data Factory tools and skills to generate defect images for circuit-board inspection, and that Siemens said the approach reduced calibration time of inspection processes by 40% at its Erlangen electronics factory.

Simulation, design and digital twins

The company said CUDA-X libraries are being used to accelerate simulation work so engineers can test design alternatives before building physical prototypes. At GTC Berlin, NVIDIA said it is releasing cuDSS for general availability to speed calculations used in structural stress, fluid-flow, electromagnetic and semiconductor-device simulations.

NVIDIA cited several industrial uses of its software stack. Mercedes-Benz uses Siemens’ Simcenter STAR-CCM+ running on NVIDIA GPUs for aerodynamic simulations, while also using NVIDIA Cosmos 3 and the Metropolis VSS blueprint for vehicle inspection and windshield assembly verification. Cadence is benchmarking automotive fluid-dynamics workloads on NVIDIA GPUs, and Autodesk is bringing PhysicsNeMo and Warp into NavPack, integrated with Autodesk Alias, to support vehicle-performance prediction. The company also said imec and Infineon are using accelerated computing for semiconductor simulation, with imec’s S-Litho model showing up to 40x speedup and Infineon reporting a 12x speedup in 3D breakdown simulation using Sentaurus Device with NVIDIA CUDA libraries.

Other examples cited by NVIDIA include Leonardo and Neural Concept working on aircraft design with Omniverse libraries, Dassault Systèmes using Nemotron-powered virtual companions and CUDA-X-accelerated solvers for vehicle-door design, and UniversalAGI working with Saab on wave-impact modeling using Warp. PhysicsX is working with GB1 on yacht design using Omniverse libraries, while Fermioniq’s Sofia agent is being used with Deltares and Dutch ministries to model river flow on Warp. Mistral also plans to use CUDA-X, Warp and Omniverse libraries to accelerate physics simulation and generate training data for engineering AI models.

Robotics and factory planning

NVIDIA said Omniverse and Isaac libraries are being used to build digital twins and robotics simulations before changes are made on the factory floor. It cited Audi’s battery-production digital twin for testing production concepts, as well as work with Dassault Systèmes and Agile Robots to simulate mobile humanoid robots and logistics processes in DELMIA virtual twins. Schaeffler is integrating digital twins of humanoid actuators with Isaac simulation frameworks, and Wandelbots is using NOVA with Isaac Sim to validate assembly workflows digitally.

The company also said Falkor, a KONGSBERG company, is integrating agents and Omniverse libraries into its platform, with Microsoft Azure NCv6 virtual machines powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs used to combine engineering data, operating information and 3D geometry for inspection and maintenance use cases. Schneider Electric and AVEVA are developing an AI factory digital twin for Nscale using Omniverse libraries to compare power, cooling and equipment configurations before physical changes are made.

Context and limitations

The announcement is framed as an ecosystem update around industrial AI, simulation and digital-twin workflows rather than a financial disclosure. NVIDIA did not provide product pricing, revenue figures or a market update in the material supplied. The reported savings and speedups are attributable to the individual companies and benchmarks named in the release, and they should be read as specific use-case results rather than broad industry-wide averages.