Siemens showcases applications with Nvidia that could cut engineering effort by up to 85%
Key Facts
Siemens on 19 October 2026 showcased industrial AI applications developed and tested with Nvidia at its electronics plant in Erlangen, saying they could reduce the engineering effort required to prepare production by up to 85%. Development and testing took place during live production, according to a Siemens statement, bringing the partnership's technologies into an operating factory. The workflow presented includes virtually preparing production of printed circuit board assemblies using digital twins. This process allows quality inspections to be prepared before physical prototypes are available by working with a virtual version of the product. The company says the same workflow can shorten time to market by up to 40%, alongside reducing engineering preparation effort.
In production planning, Siemens said the use of AI agents could deliver cost savings of up to 20%. These agents generate scenarios for handling new orders, which are then validated using a digital twin of the factory before implementation. The Erlangen site already uses more than 100 AI algorithms across production, logistics, quality inspection and maintenance. Among the new applications, Nvidia's "Cosmos" models generate realistic training data for quality inspection systems, according to the company. The site's applications thus cover planning how to fulfil orders and preparing inspection systems, alongside existing uses in material flows and production equipment maintenance.
The Erlangen applications build on an expansion of the partnership announced by Siemens and Nvidia on 6 January 2026 at the Consumer Electronics Show to build an industrial AI operating system. According to Nvidia's announcement, Siemens is committing "hundreds of industrial AI experts", along with its hardware and software. Nvidia, in turn, is providing the computing infrastructure, simulation libraries, models and development frameworks needed for the collaboration. The operating system project was preceded by collaboration to connect Siemens' "Xcelerator" platform with Nvidia's "Omniverse" platform. The connection aimed to build digital twins combining physics simulation with real-time AI, providing the technical foundation for representing and testing industrial processes virtually.
The partnership extends to chip design, with the companies targeting a speedup of between 2 and 10 times in some design automation workflows, according to Nvidia's announcement. This goal relies on integrating Nvidia technologies into Siemens' tools for verification, planning and process optimisation. The announcement named Foxconn, Hyundai Heavy Industries, KION Group and PepsiCo among customers evaluating some of the partnership's capabilities. The collaboration also includes developing a repeatable blueprint for AI computing centres. This blueprint coordinates power, cooling and automation requirements to improve the efficiency of the infrastructure powering those centres.
The Erlangen presentation was preceded by results the companies announced from industrial simulation applications at other customers. In a June 2025 announcement, Nvidia said BMW Group and Siemens achieved a 30-fold speedup using "Simcenter STAR-CCM+" simulation software accelerated by its technologies. The result involved transient aerodynamic simulations of full vehicles, including the entire vehicle geometry. In a separate announcement in January 2026, Siemens said the use of digital twins at PepsiCo increased throughput by 20% during initial implementation. The PepsiCo result focused on production volume, while the automotive group's result concerned the speed of completing engineering simulations.
Development of the Erlangen applications is part of a programme to transform and expand the site, for which Siemens has announced an investment of 500 million euros. The programme includes establishing a technology campus for development and manufacturing, according to the company's statement on the site project. Across the broader technology partnership, the development roadmap includes completing GPU acceleration of Siemens' simulation portfolio. This aims to enable customers to run larger, more accurate simulations faster, according to Nvidia. The companies also intend to develop autonomous digital twins that provide engineering designs and real-time self-optimisation, using the "PhysicsNeMo" framework and open models.
Siemens said the companies planned to add further applications in the months following the Erlangen presentation in October 2026. Its statement said that "further applications are planned in the coming months". In parallel, the first phase of the site's development includes a centre for logistics, services and flexible automation, scheduled to be built by 2027. In the next phase, Siemens plans to establish a centre for power electronics development and innovation. This phase includes additional manufacturing space, with completion scheduled for 2030.