Factory data often sits inside operational systems that are separate from enterprise IT. Cisco and Rockwell Automation are targeting that divide with a software-defined manufacturing architecture that links plant-floor systems with higher-level computing environments.
The companies’ Full-Stack Software-Defined Manufacturing architecture combines Cisco’s software-defined networking with Rockwell Automation’s software-defined automation. Cisco detailed the approach recently, describing it as infrastructure for moving industrial data between operational technology (OT) environments and computing systems used for analytics and AI.
The reference design addresses one of the technical requirements behind industrial AI: getting operational data from machines, sensors, controllers, and production systems into environments where it can be processed. Cisco and Rockwell said the architecture creates secure data paths between plant-floor equipment and compute resources located at the edge or in the cloud.
Cisco and Rockwell also identify isolated industrial data as a barrier to wider use of analytics and AI in manufacturing. Their architecture is designed to connect industrial control systems with enterprise networks and higher-level computing resources so plant-floor data can be used by edge, data-centre, and cloud applications.
Industrial environments can generate large volumes of data from connected production cells, machinery, materials, products, and sensors. Cisco said difficulties accessing accurate and contextualised plant-floor data can limit analytics and the training and operation of AI and machine-learning systems.
Accessing machine data is only one part of that process. Industrial AI also depends on managing and combining information from production systems so sensor and control data can be used alongside engineering and enterprise information.
The US National Institute of Standards and Technology (NIST), in its 2026 roadmap for AI and machine learning in smart manufacturing, identified industrial data management, integration with heterogeneous sensing and control systems, and interoperability between AI platforms and established manufacturing software as continuing deployment challenges.
The architecture uses Cisco networking and Rockwell Automation’s industrial systems to connect those OT data sources with higher-level applications. It supports on-premises, hybrid, and public cloud environments alongside edge computing infrastructure.
Moving factory data into AI systems
Software-defined networking provides the underlying connectivity between industrial and enterprise systems. Cisco said the architecture is designed to carry real-time data between sensors, actuators, controllers, edge devices, and cloud platforms while applying network segmentation, access policies, threat detection, and controls for remote connections.
Rockwell describes data and edge platforms as the infrastructure connecting equipment, control systems, and enterprise applications for industrial AI. Those platforms can supply data for model training while also supporting inference and workload orchestration across distributed production environments.
The flow of information is not limited to moving plant-floor data into central cloud systems. Rockwell said industrial data platforms can also deploy trained AI models back to edge environments, allowing inference to run closer to the equipment generating the data.
The companies are using edge computing for workloads that need processing closer to production systems. Cisco’s architecture includes edge infrastructure, data-centre resources, and multicloud connectivity for AI and machine-learning workloads that use operational data.
Cisco and Rockwell identify low latency and network reliability as requirements for more autonomous industrial operations. Their architecture therefore combines edge computing with data-centre and multicloud resources rather than placing every AI workload in one central environment.
Time-sensitive workloads can remain close to production equipment, while data needed for model development, analytics, and engineering can move into higher-level computing systems.
Supported use cases include predictive maintenance, AI models running at the edge, digital twins, cloud-based engineering, and cybersecurity spanning edge and cloud environments. The architecture also covers data collection and contextualisation for analytics across industrial and enterprise systems.
Predictive maintenance provides one example of how operational data can be used by industrial AI. Rockwell’s GuardianAI uses equipment data to establish normal operating baselines and detect deviations, while its separate work with Augury connects machine-health insights with maintenance recommendations and execution workflows.
Digital twins provide another example of how operational and higher-level systems intersect. Cisco and Rockwell include digital twins and cloud-based engineering among the workloads supported by the architecture, while Rockwell uses digital twin software for simulation and virtual commissioning of automated systems.
Rockwell also describes industrial AI systems that use real-time production data for adaptive automation and closed-loop optimisation. Cisco and Rockwell’s reference architecture provides the network, edge, and data infrastructure connecting those workloads with plant-floor systems.
Rockwell is separately connecting more of its manufacturing software with AI systems. On August 11, the company announced an API-based integration between its Plex Quality Management System and FactoryTalk Analytics VisionAI, connecting AI-based visual inspection with quality-management workflows.
Inspection results generated by FactoryTalk Analytics VisionAI can be recorded in Plex QMS for traceability, product serialisation, and inspection history. Rockwell said its wider AI work also covers cloud-based manufacturing execution systems, edge AI, and digital twins.
IT/OT convergence raises integration challenges
Cisco and Rockwell’s manufacturing architecture also builds on their Converged Plantwide Ethernet, or CPwE, work. The companies have developed CPwE reference architectures for more than a decade to connect EtherNet/IP industrial automation systems with wider enterprise networks.
CPwE provides tested designs for industrial networking and forms the network foundation for the newer software-defined manufacturing architecture. Cisco said the newer design extends that model to support software-defined automation, AI workloads, industrial observability, and additional security controls.
Connecting industrial environments to AI infrastructure also requires dealing with equipment that was not designed for modern data integration. NIST noted that connected manufacturing systems can operate alongside older machinery that lacks standard communications protocols or digital interfaces.
That coexistence can complicate interoperability between newer AI platforms, industrial control systems, and enterprise software, particularly where legacy equipment cannot exchange data through modern interfaces.
Connecting industrial equipment with enterprise and cloud systems also changes the security requirements of the network. Cisco identified increased connectivity between IT and OT as an expansion of the potential attack surface, particularly where legacy industrial systems cannot be upgraded or patched as easily as conventional IT equipment.
The reference design incorporates Cisco Industrial Threat Defense and security controls covering endpoints, networks, applications, and data flows. The architecture supports network segmentation, granular access policies, remote-access security, and threat detection across connected industrial environments.
Cisco and Rockwell also cite ISA/IEC 62443 and NIST among the standards relevant to the security architecture. Their design includes controls around endpoint protection, recovery, and communications between industrial and higher-level systems.
The companies formally launched the Full-Stack Software-Defined Manufacturing reference design in India in July. A demonstration environment at Rockwell Automation’s Gurugram facility allows manufacturers and partners to test the architecture before applying it across production lines or plants.
The July collaboration also included a training programme through Cisco Networking Academy and Rockwell’s Learning+ programme. It focuses on networking, cybersecurity, and digital skills for employees working with software-defined industrial environments.
See also: IT/OT convergence brings OT security into ServiceNow
Want to learn more about the IoT from industry leaders? Check out IoT Tech Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including AI & Big Data Expo and the Cyber Security Expo. Click here for more information.
IoT News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
