Wednesday, 21 October 2026Support us

Aube.

News of progress
PilotSingle source

Siemens tests AI factory tools in live production at Erlangen

Languages for this article
Original · ENESFRITPT

Originally written in English. 5 languages available; yours is one click away.

Illustrative photo: Machine working in a factory while cutting material on a production line in the workshop
Illustrative photo: Shixart1985 · CC BY 2.0 · Wikimedia Commons

For a new circuit board, the first check used to wait for a physical prototype. Quality inspection preparation could take several weeks, because teams needed real boards to train the vision systems that check solder and components, and to set up the inspection equipment. At Siemens' Electronics Factory in Erlangen, in southern Germany, that wait is being cut. The new workflow uses AI to generate photorealistic images of circuit board assemblies from digital twins, which are virtual copies of a product. Engineers can prepare inspection systems before any physical prototype exists, and the applications are now being tested in live production at the plant.

The Erlangen plant was founded in 1971 and employs about 1,000 people. It makes power electronics and automation technology for industry and infrastructure, including parts used in machine tools and battery production. The site has run digital manufacturing technologies during live operations for more than a decade, and more than 100 AI algorithms are already in use in production, logistics, quality inspection and maintenance. The Siemens and NVIDIA team built the first applications in less than ten months, combining NVIDIA's AI, simulation and accelerated computing with Siemens' industrial software and automation expertise. The collaboration was announced at CES in Las Vegas in January 2026.

The planning tool works in a loop. When new orders arrive, an AI agent generates production scenarios that consider capacity, material availability, delivery dates and line constraints. It shows planners the effect of each option using defined key performance indicators. Each proposed plan is then evaluated in a digital twin of the production facility before it is implemented. NVIDIA Cosmos world models generate training data for quality inspection, and NVIDIA Omniverse libraries let teams test plans virtually with factory data. Siemens states that this process can support cost savings of up to 20% while helping manufacturers respond to new orders, material shortages and changes in customer requirements.

The figures need reading with care. The 85% cut in engineering effort and 40% reduction in time to market are stated as maximums for new printed circuit board assemblies, not as averages across the plant, and they are Siemens' own estimates. The tools are being tested at Erlangen to judge how they could be deployed at other manufacturing sites, but the sources name no other site and give no date for that move. Siemens and NVIDIA plan to build further applications for product design, manufacturing and delivery operations in the coming months.

For the engineers and planners at Erlangen, the gain is measured in weeks saved on inspection preparation and in plans checked before they reach the line. For Siemens and NVIDIA, the next step is to extend these tools beyond one site, and the timing of that step is not yet set. Siemens also plans to expand the Erlangen site through 2030, backed by an announced investment of €500 million in its transformation. For now, the change is concrete in a single factory, and the test is whether the same gains hold on lines that were not built with these tools in mind.

up to 85%Cut in engineering effort for new PCB assemblies with virtual production preparation, per Siemens

Sources — read the originals(Paris time)

Engineering.comEN
0000

Read next

Comments

Loading the thread…

Sign in to leave a comment. Sign in