How we know: Two Siemens releases, both quoting NVIDIA, plus a trade-press restatement. The figures are Siemens' 'up to' numbers.
ManufacturingCase study
Siemens says virtual ramp-up can cut PCB engineering effort at Erlangen by up to 85%
Synthetic images of circuit boards let the plant prepare inspection before the first physical prototype exists, a step that used to take several weeks. Siemens' false-call and yield figures are still targets.
ERLANGEN, Germany, Oct. 19, 2026 — Siemens said preparing new printed-circuit-board assemblies virtually at its Erlangen electronics factory can cut engineering effort by up to 85 percent and time-to-market by up to 40 percent, because inspection no longer waits for a physical prototype. Here's what happened, what it cost, and what went wrong.
Cost and outcome (as of Oct. 21, 2026)
- Outcome: Virtual production preparation "can reduce engineering effort by up to 85 percent and cut time-to-market by up to 40 percent," Siemens said. (Siemens, Oct. 19, 2026)
- Cost: Agent-based production planning "can enable cost savings of up to 20 percent." Siemens is investing €500 million in the site through 2030; Manufacturing.net, Oct. 19, 2026 put that at about $560 million.
- Targets: The PCB inspection project is "targeting" 40 percent fewer false calls and more than 5 percent first-pass yield improvement. (Siemens, Oct. 21, 2026)
The wait it removes
Siemens said preparing quality inspection "previously took several weeks, because teams had to wait for physical prototypes to train vision systems and manually configure inspection equipment." (Siemens, Oct. 19, 2026)
Now AI "generates photorealistic representations of PCB assemblies from their digital twins to train the vision systems," so engineers can prepare inspection before prototypes exist. The boards are the core of the drives and frequency converters made in Erlangen. (Siemens, Oct. 19, 2026)
In the companion release, Siemens said its Teamcenter and Process Preparation X software build a digital twin of the "golden board," and NVIDIA Cosmos generates synthetic images of good and defective boards so inspection machines can be calibrated first. (Siemens, Oct. 21, 2026)
Once production starts, an AI-supported secondary inspection step checks whether a defect flagged by the optical inspection system "represents a real defect or a false call." (Siemens, Oct. 21, 2026)
The goal is a factory that can "rapidly introduce near daily changes (3 to 5 times a week) without disruption." (Siemens, Oct. 21, 2026)
Planning by agent
Planners used to weigh capacity, materials, delivery dates and lines "based on their experience." Now an AI agent generates production scenarios when orders arrive and shows the impact on defined KPIs; proposals are checked against a digital twin of the whole plant before they're used. (Siemens, Oct. 19, 2026)
The plant
The Erlangen plant opened in 1971, employs about 1,000 people and has more than 100 AI algorithms in use across production, logistics, quality inspection and maintenance, Siemens said. (Siemens, Oct. 19, 2026)
Siemens board member Cedrik Neike said Erlangen has "an annual productivity increase of seven percent." The joint Siemens and NVIDIA team built the first new offerings "in less than ten months." (Siemens, Oct. 19, 2026)
The two companies said at CES in January 2026 that they would build "the world's first fully AI-powered manufacturing sites," and have since been co-developing and testing applications in live production at Erlangen. Neike said the plant gets "around 3,000 customer visits every year." Manufacturing.net, Oct. 19, 2026 reported that further applications are planned in the coming months. (Siemens, Oct. 19, 2026)
"Manufacturers shouldn't have to wait for the first physical circuit board to start preparing inspection," said NVIDIA's Rev Lebaredian. (Siemens, Oct. 21, 2026)
Siemens executive Mark Hindsbo said a "virtual" golden board lets electronics makers "change production dramatically quicker without taking the line out of production." (Siemens, Oct. 21, 2026)
What went wrong
- Ceilings, not averages. Every headline figure is "up to" or "can," with no baseline product or line given. (Siemens, Oct. 19, 2026)
- Inspection gains are targets. Siemens says the project is "targeting" its false-call and yield figures. (Siemens, Oct. 21, 2026)
- The vendor is the customer. Erlangen is Siemens' own "Customer Zero" for products Siemens and NVIDIA sell. (Siemens, Oct. 21, 2026)
- The 85% is also a target. Siemens' Oct. 21 release lists "up to 85 percent less manual engineering effort" among the project's targets. (Siemens, Oct. 21, 2026)
- What we don't know yet: how many PCB introductions have used the virtual workflow, and what the typical saving was.
Our follow-up question for Siemens
On how many new PCB assemblies has the virtual workflow been used, and what was the median engineering-effort saving?
Sources
- Siemens AG, “Siemens partners with NVIDIA to advance AI-driven, autonomous factories,” press release HQCOPR202610177477EN, Oct. 19, 2026. assets.new.siemens.com
- Siemens, “Siemens and NVIDIA advance AI-powered PCB inspection from Digital Twin to factory,” Oct. 21, 2026. news.siemens.com
- IEN Staff, “Siemens Cuts Engineering Effort by 85%, Time-to-Market by 40% at AI Pilot Plant,” Manufacturing.net, Oct. 19, 2026. manufacturing.net
© 2026 Real World Wire. Written October 29, 2026.