Scaling Industrial AI from Months to Days, Siemens and AWS Joining Hands

This session shows how Siemens and AWS have transformed industrial AI deployment at a Siemens electronics factory, reducing deployment times from months to days through a standardized, template-based architecture built on Siemens Industrial Edge and AWS cloud services. Marvin, advanced optical inspection and closed loop manufacturing expert at Siemens, and Henning Rudolph, principal partner development manager for industrial at AWS, walk through the full journey — from the origins of the Siemens-AWS partnership in 2015 to live production use cases running today. The session covers four vision AI use cases implemented on the factory shop floor: automatic optical inspection of component placement, flux inspection using 3D cameras, AI-assisted solder quality inspection to reduce false positives and manual rework, and coating inspection using a low-cost, cloud-supported approach. All four use cases run on the same technology stack, developed jointly by Siemens and AWS. A central theme is the shift away from isolated, silo-based AI deployments toward a repeatable, template-driven MLOps cycle spanning edge to cloud. Data collected from machines and cameras on the shop floor is stored in Amazon S3, processed using AWS Lambda and ECS container infrastructure, and redeployed to edge devices via a unified AI asset manager. This approach allows factory operators — not just data scientists — to manage and monitor AI models across production cells. The factory, a World Economic Forum Lighthouse Factory, now runs over 100 AI models on the shop floor. The first use case took five to six months to implement. Subsequent deployments using the established blueprint were completed in two to three days. The template approach has also enabled cross-factory scaling, with other Siemens factories adopting use cases through a shared platform and common architecture. The speakers also discuss the vision for autonomous manufacturing, where AI models operating in closed loop may eventually control and optimize one another, with generative AI expected to play a growing role in routine process optimization.