# Siemens tests AI inspection and planning in a working factory Canonical URL: https://www.theroboteconomy.news/articles/siemens-erlangen-ai-inspection-production-planning/ Publisher: The Robot Economy Editorial byline: Mara Voss, Deployment Desk Published: 2026-10-20T08:52:08.551Z Updated: 2026-10-20T08:52:08.551Z Editorial information: https://www.theroboteconomy.news/about/#editorial-desks New applications at Erlangen bring virtual inspection preparation and AI planning into live production tests. Reported savings are specific to those workflows. ## Summary Siemens disclosed new AI inspection-preparation and production-planning applications tested with NVIDIA in live manufacturing at Erlangen. Its efficiency figures concern specific workflows. Comparing those applications with the broader automation target distinguishes current process assistance from future factory autonomy. ## Key facts - Siemens announced new Erlangen AI applications on 19 October 2026. - Synthetic images prepare circuit-board visual inspection before physical prototypes are available. - The planning agent presents scenarios to human planners and validates proposals in a factory simulation before implementation. - Siemens describes engineering-effort reductions and cost savings as up-to figures; they are not measured whole-factory gains. ## Lead image Image: https://www.theroboteconomy.news/images/articles/siemens-erlangen-ai-inspection-production-planning.webp Description: An industrial robot grips a fan assembly at Siemens’ Erlangen electronics factory, in a company-supplied context photograph. Credit: Context photograph: Siemens AG; industrial robot supporting automated handling at Erlangen. ## Report Siemens says it and NVIDIA are testing new AI inspection and production-planning applications in live manufacturing at its Erlangen electronics factory in Germany. The 19 October disclosure identifies two practical uses: preparing inspection before physical prototypes exist, and helping human planners compare production options. Sources: [s1]() The companies announced their expanded manufacturing partnership in January 2026. Siemens’ current partnership timeline describes subsequent co-development and live-production testing at Erlangen, with the site serving as a route toward broader industrial deployment. Sources: [s3]() ### What changes on the factory floor For printed circuit-board assemblies used in drives and frequency converters, Siemens reports that virtual production preparation can reduce engineering effort by up to 85% and time to market by up to 40%. The workflow generates simulated images from digital models of the boards to train visual inspection systems before physical prototypes become available. Sources: [s1]() NVIDIA’s published technical workflow separately documents generating clean and defective circuit-board images from design data, then aligning with real photographs. It illustrates how simulated data can prepare inspection models; it does not establish that Siemens uses that exact implementation or independently verify its savings. Sources: [s4]() A separate planning agent generates options when new orders arrive, showing human planners how choices affect production. Proposed plans are checked in a digital twin, a simulation of the factory, before implementation. Siemens says this approach can enable cost savings of up to 20%. These are company-reported potential gains, not an independently audited result for every production line. Sources: [s1]() ### The wider automation goal remains ahead Reuters, reporting from Erlangen on 19 October, attributed to Siemens managing-board member Cedrik Neike an ambition to raise factory-manufacturing automation from 30–40% to 70–80%. That is a future target. It does not establish that today’s inspection and planning applications have already achieved that level. Sources: [s2]() The distinction matters for manufacturers evaluating industrial AI. Faster inspection preparation addresses the work required to introduce a product; planning assistance addresses how an existing factory responds to changing orders. Each can be useful without demonstrating that the entire factory runs autonomously. Sources: [s1]() [s2]() [s3]() The next useful comparison is between matched production tasks: time to prepare inspection, cost to revise a schedule, and the human interventions needed when products or materials change. Those measures would show how much of the reported benefit transfers beyond the initial applications. The reviewed announcements do not provide that common comparison. Sources: [s1]() [s3]() ## Conclusion Separates two disclosed live-production workflows from the executive whole-factory automation ambition, preserving upper-bound denominators and identifying matched-task measurements needed to evaluate transfer beyond the initial applications. ## Disclosures Sponsored: no Desk bylines, AI assistance and editorial policies: https://www.theroboteconomy.news/about/ ## Sources - s1: [Siemens partners with NVIDIA to advance AI-driven, autonomous factories]() — Siemens AG; company-announcement; 2026-10-19T00:00:00Z - s2: [Siemens targets up to 80% factory automation with AI, board member says]() — Reuters; independent-reporting; 2026-10-19T00:00:00Z - s3: [Siemens and NVIDIA partnership timeline:19 October 2026 live-production applications]() — Siemens AG; company-announcement; 2026-10-19T00:00:00Z - s4: [Physical AI Defect Image Generation: version1.0.1 technical workflow]() — NVIDIA; technical-document; 2026-06-24T01:02:07Z