How artificial intelligence is reshaping manufacturing operations from the factory floor to the cloud

Artificial intelligence is moving beyond the buzz in manufacturing, delivering real value through applications like predictive maintenance, quality inspection, and production optimization. In a recent conversation, industry experts Alex West and Anna Ahrens explored how AI is transforming operations today and where it’s heading next.

This blog highlights that discussion, covering the practical benefits of AI, the challenges of scaling, and the growing importance of tools like Edge AI and Generative AI. It also looks at key enablers for success, from data quality and integration to cross-functional collaboration, and where the biggest growth areas are emerging.

Whether you're just starting with AI or looking to scale, this blog offers insights into what’s working, what’s next, and how to stay ahead in the evolving world of smart manufacturing.

How is AI adding value for customers in manufacturing right now?

AI is creating value, and not just today. Applications like predictive maintenance, quality inspections, and production optimization have been around for years and have proven their return on investment (RoI). In fact, some predictive maintenance solution vendors provide guaranteed RoI for projects before they even start. Customers report 25-30% savings in maintenance efforts with predictive analytics. With AI-driven quality inspection, it is possible to achieve up to a 95% reduction in quality inspection errors and up to a 45% reduction in quality variability, according to our quarterly IIoT projects database. In the same database, we see reported productivity improvements ranging from 10-30%, achieved through data analytics, lead time reduction, and overall equipment effectiveness (OEE) increases from 2-15% via condition monitoring and data analytics, as well as labor efficiency increases. Not to forget the workforce savings on repetitive tasks reduction with the help of AI.

Generative AI (GenAI) and AI assistants/chatbots are new tools that will deliver value in documentation handling, troubleshooting, and programming, increasing workers' efficiency and significantly simplifying engineering tasks. Due to the general awareness of AI possibilities and the boom created by GenAI, we see manufacturers investigating and investing in AI projects. However, the real implementation and scale of AI projects are still very low.

How can businesses make money from AI?

The value of AI projects grows with scale. We see rather low numbers, around 10%, of manufacturers implementing AI at scale. The major challenge of AI projects lies in their complexity on several levels: data quality and availability, OT-IT integration, and cloud integration are required.

The most important factors are management support, organizational structure readiness for AI (dedicated leads, cross-department collaboration, and centralized efforts), and expert knowledge. While the first group is related to digital maturity and can be solved in some cases with additional equipment like sensors and gateways or with AI and analytic tools/partners involvement for data creation and handling, the second group of hurdles relates to company culture and strategy and should be solved internally. For example, AI often does not have its own budget and sits with IT, while a cross-domain approach is indispensable. Domain experts' involvement is as critical as data scientists at all stages of the project lifecycle. A centralized, cross-department, and even cross-site approach is important for scaling. The same applies to data governance, which is often neglected by manufacturers.

Who stands to profit from AI in manufacturing?

Due to the complexity of technology and missing expertise, services and partnerships are extremely important. According to Omdia’s AI in Manufacturing survey, less than a quarter of customers are developing and integrating AI projects in-house; all other projects are partner-led, often reliant on cloud offerings. The share of system integration and consulting in the AI manufacturing value chain is significant, and solution vendors should consider this in their offerings.

What's the status of Edge AI in manufacturing?

Edge AI is the future of manufacturing AI. Edge AI processes data on-site, close to machines, rather than sending it to the cloud. This is critical for real-time closed-loop applications, security, and bandwidth efficiency. Customers indicate in our surveys that they are choosing edge solutions due to data privacy/IP requirements and to reduce cloud usage costs.

We have three important drivers here: the development of edge hardware (GPUs, other AI accelerators), the development of AI models that can be deployed on restricted devices, including small LLMs, and the emergence of edge platforms, which provide centralized management of devices and workloads for edge systems. This all drives the adoption of edge AI.

For vendors, Edge AI unlocks new business models by embedding AI in industrial and compute devices with the possibility to sell them as part of an edge platform, providing competitive advantages, additional services like predictive maintenance or control optimization, and securing recurring revenue with the movement to the subscription/SaaS business model.

Currently, there is a lot of uncertainty in this area, among customers as well as on the vendor’s side. Especially for critical applications relevant to worker safety, such as automated control and control optimization, questions arise: How reliable are available AI-based solutions? How to integrate edge AI into legacy devices and systems? Where should edge AI be placed? If embedded on a device, how to avoid interference with other automation functions and applications? Who and how will update and patch the edge embedded devices and applications? There will be an increased need for orchestration and management solutions and efforts to prove the reliability of systems with edge AI to end customers.

Which areas in manufacturing are seeing the most growth due to AI?

Machine vision in combination with robotics is an area with major growth (but has limited volume). Applications include DL-based identification of products (defect analysis, pick and place, vision and lidar-based pallet identification and transportation, autonomous route adjustment, 3D vision-based navigation systems and optimization, quality inspection, cobot collaboration, and autonomous adjustment of material handling based on vision and analytics). For these applications, edge AI would be the preferable location due to cost, security, and latency requirements.

GenAI/low-code-based programming is promising, though in an early stage.

How is Generative AI being used in manufacturing?

Generative AI is emerging as a powerful tool for automation vendors, with applications in Copilots for Engineers and Maintenance (code generation, troubleshooting guidance). Another area is generative design and simulation, enhancement in supply and demand chain optimization.

One of the defining features of GenAI is its multimodality—an ability to work with different data formats: text, drawing, numbers, sound. This explains the large potential of GenAI for manufacturing in particular. Of course, customer interaction, requirements handling, documentation, and compliance handling are huge implementation areas.

Determinism is a weak point of GenAI, or a weak point of manufacturing. Unlike rule-based AI, generative AI is not 100% reliable and should be implemented for the right use cases which have some fault tolerance. Even then, continuous improvement and development, model re-training, and results evaluation with domain experts are unavoidable.

What’s next for AI in manufacturing over the next 5 years?

We expect AI in manufacturing to become more autonomous, scalable, and integrated into industrial workflows. We will see more AI deployments and edge AI use cases. Recently, agentic AI has become a more prominent topic in manufacturing—we expect to see progress in agent-based use cases and frameworks.

The AI ecosystem, especially integration and maintenance services for AI applications, will grow around it. Most importantly, we hope to see AI helping with the workforce and skills problem in manufacturing: taking over repetitive tasks, simplifying complex and domain knowledge-based use cases like coding and engineering, and making manufacturing an even more attractive world to work in.

As manufacturers continue to explore the potential of AI to drive efficiency, innovation, and growth, informed decision-making is more important than ever. Omdia’s deep expertise in industrial AI, edge computing, and digital transformation provides the strategic insights you need to stay ahead. Connect with our analysts to learn how our research can support your AI strategy and help you navigate the evolving landscape of smart manufacturing.