Manufacturing Intelligence Report: The intelligent factory needs more than artificial intelligence
Manufacturers have spent years connecting machines, integrating enterprise systems and collecting production data, creating an environment in which artificial intelligence can begin to influence how factories operate. Yet as AI moves beyond experimentation and becomes embedded in production, an increasingly important question is being overlooked. Are factories designed to support intelligence at scale?
The challenge is no longer simply whether AI can detect equipment failures, optimize production schedules or identify quality problems. Those capabilities are becoming increasingly accessible. What remains harder is ensuring intelligent systems understand the operational environment, can access the information required to make reliable decisions and deliver results without introducing additional complexity.
This is the central argument of The intelligent factory has an architecture problem, a new Manufacturing Intelligence Report 2026 from The Manufacturing Tech Brief. Drawing on interviews with manufacturing and technology leaders, the report examines how industrial AI is changing the requirements for infrastructure, data integration, operational software and cybersecurity.
The findings challenge a longstanding assumption that increasing connectivity and deploying more sophisticated applications will inevitably create smarter factories. While these investments have improved operational visibility, they have not necessarily provided the foundations for intelligence to work effectively across different systems and facilities.
A factory may have predictive maintenance software, advanced production analytics and AI-enabled quality inspection, but if those capabilities operate independently, the result may be another generation of technological silos.
Architecture becomes an operational decision
As AI begins influencing production rather than simply analyzing it, previously supporting technologies are becoming central to manufacturing performance. Industrial networks, computing infrastructure and data platforms increasingly determine how quickly information can be processed, where decisions can be made and whether intelligent applications operate reliably.
The implications are significant for manufacturers trying to establish consistent digital strategies across multiple plants. Different facilities may operate equipment installed decades apart, use different generations of industrial software and have varying levels of connectivity. Introducing AI requires an approach that accommodates those differences without allowing every implementation to become a separate engineering project.
Jeremy Foster, Senior Vice President and General Manager of Cisco Compute, illustrates the changing importance of infrastructure. “The network is to a modern AI factory what power distribution was to a manufacturing facility 50 years ago,” he says.
As AI applications become more dependent on real-time information and increasingly influence physical processes, industrial connectivity is evolving from a supporting technology into a critical factor determining production performance, with network reliability and resilience shaping what manufacturing systems can achieve.
Computing architecture is also evolving to accommodate different operational requirements, with workloads such as model training and enterprise analytics demanding substantial centralized resources, while applications including machine vision and real-time quality inspection require processing capabilities closer to the production equipment generating the data.
Decisions about where intelligence operates therefore need to be driven by operational requirements rather than a general preference for cloud, edge or on-premises infrastructure. Manufacturers must also distinguish between collecting information and making it useful. Industrial equipment generates enormous quantities of data, but transmitting everything to a central platform increases costs without necessarily improving decisions. A more effective architecture processes information where it is relevant while ensuring important findings reach the people and systems capable of acting on them.
Connected data is not connected intelligence
One of the report’s most important arguments concerns the difference between operational visibility and operational understanding, as manufacturers increasingly rely on manufacturing execution systems (MES), enterprise resource planning (ERP), product lifecycle management (PLM), digital twins and specialist engineering applications that may provide accurate information within their own domains while remaining largely unaware of the consequences of events elsewhere.
Consider a predictive maintenance system identifying deterioration in a critical machine. The technical diagnosis may be accurate, but deciding what to do requires considerably more information. What is the equipment producing? Which customer orders depend on it? Is alternative production capacity available? How would an intervention affect delivery commitments?
Rather than consolidating everything into one application, manufacturers need to preserve context as information moves between specialist systems, allowing equipment condition to be understood alongside production schedules, product information and wider business requirements. This is changing the role of technologies such as MES, which can connect equipment data with production orders, materials and process conditions, while digital twins are evolving beyond visual representations into environments that support operational decisions. The objective is to ensure that information retains its meaning as it moves between systems, enabling manufacturers to understand not only what is happening in production but also the consequences for the wider business.
Product information presents comparable challenges. An engineering change can affect procurement, inventory, production and potentially products already operating at customer sites. Understanding those consequences requires information to remain connected across the lifecycle without assuming every department uses the same software.
The report presents the intelligent factory less as a collection of connected applications and more as an environment capable of coordinating decisions across specialist systems. This approach is illustrated by a global electronics manufacturer operating more than 150 factories, where a real-time event architecture connected manufacturing information with supply chain planning, allowing changes that previously took days to gather, evaluate and distribute to be reflected in production plans in under an hour. The improvement came not from replacing existing systems but from enabling information and decisions to move more effectively between them.
Trust and expertise shape the intelligent factory
As manufacturers give AI greater responsibility within operational workflows, the consequences of inaccurate information and poorly controlled systems become more serious. A dashboard containing an incorrect inventory figure may prompt an employee to investigate, but an automated system using the same information could initiate an inappropriate replenishment decision before anyone recognizes the problem. Data quality and governance therefore become essential to ensuring that intelligent systems understand what information means, where it originated and whether it is reliable enough to support a particular decision.
The challenge extends into cybersecurity, particularly as intelligent applications become more closely integrated with production equipment. Many industrial assets cannot be patched, restarted or subjected to conventional security testing without disrupting operations, making it essential to design architectures that can contain threats and maintain production when individual systems fail. Manufacturers must also establish clear boundaries around the authority given to AI, ensuring that increasingly sophisticated systems can support operational decisions without introducing unacceptable risks to safety, reliability or performance.
However, greater intelligence does not necessarily mean fully autonomous manufacturing, and some of the most credible applications involve AI supporting established engineering workflows rather than replacing experienced employees. In computer-aided manufacturing, for example, AI can assist with individual programming tasks while engineers review the results, while simulation allows teams to evaluate proposed operational changes before introducing them into live production. This approach recognizes that manufacturing decisions often carry physical consequences and that faster analysis does not eliminate the need for verification and accountability.
Dayan Rodriguez, Corporate Vice President of Manufacturing & Mobility at Microsoft, captures the broader change in the report. “The manufacturers building sustainable AI strategies have stopped thinking about AI as a tool and started thinking about it as part of how their business operates,” he says.
Integrating AI into established workflows also creates opportunities to preserve and extend valuable manufacturing knowledge. Experienced engineers understand equipment behavior, process limitations and unusual operating conditions that may not be adequately represented in formal documentation, and intelligent systems can help make that expertise available across a wider workforce. Rather than removing the need for skilled employees, the opportunity lies in supporting their decisions, accelerating learning and improving consistency across different facilities.
The next stage of intelligent manufacturing is unlikely to involve a universal technology platform or a single model for autonomous factories, because different production environments will continue to require different levels of automation, computing capability and human oversight. The manufacturers making the greatest progress may therefore be those that connect intelligence to the operational context it needs, embed it into established workflows and ensure decisions remain trustworthy as technology becomes an integral part of production.
The intelligent factory has an architecture problem
Discover how manufacturing and technology leaders are rethinking industrial AI, edge computing, connected operations, data governance and cybersecurity.
Drawing on industry interviews and practical examples, this 36-page report examines the architectural decisions shaping the next generation of intelligent factories.