In the context of global growth, there is a need for shorter time-to-market, cost reduction, flexibility, scalability, and efficiency. To achieve this, the process industry is moving towards modularized systems[1], [2]. To further improve efficiency and production yield, the use of Model Predictive Control (MPC) is also being investigated. However, the use of MPC in the production environment has so far been limited due to the software effort required for its complex integration. However, we can harvest the potential of modularized software and leverage the power of distributed software systems, there is an opportunity to make our software flexible, scalable, easy to implement, and manufacturer independent. This would also allow us to integrate MPC into the lifecycle of our plants. While including the MPC in the factory life cycle, it is important to take care of the sustainability goals and resource efficiency. Also, as the MPC is a controller application it should have low cycle times and be secure. All this could be achieved via the edge-devices. This paper demonstrates simple containerized MPC applications for modular plants on edge devices. We present the current state of the art and discuss the underlying automation architecture. This is based on cognitive architecture in combination with modular plant concepts. This is demonstrated by implementing a proof-of-concept demonstrator. The results are compared with the existing cloud-based approach. The results exhibit that this approach is comparable to the cloud-based approach. The paper discusses the lessons learned from the demonstrator and outlines topics for future research.