Europe keeps talking about industrial AI. Manufacturing SMEs are still being left alone.

In Italy, and across much of Europe, we keep saying that manufacturing is strategic for competitiveness, resilience, sovereignty and growth.

Yet, the way Europe approaches innovation for manufacturing SMEs still feels strangely unserious.

We fund pilots and we hold panels on AI readiness, but when it comes to the companies that actually make up the industrial backbone of our economies, the support model is still often disconnected from how these businesses operate in the real world, with the result that manufacturing SMEs still experience AI as something distant, expensive, intrusive, and vaguely threatening.

That perception is often described as a “resistance to change”, without accounting for the shortfalls of market and policy environment in making innovation understandable and accessible. If a factory owner believes an innovation will end up policing operators, if a production manager assumes it will require a multi-year integration project, if an engineer expects another black-box dashboard that nobody on the line will use, those reactions did not appear out of nowhere. They were produced by years of weak translation between technology and operations and too many projects designed around funding rather than adoption logic.

This is where the current European conversation still falls short as it assumes that once awareness exists, adoption will follow. That assumption is false in manufacturing SMEs.

An SME does not need another generic message about transformation. It needs to know, concretely, whether a solution will reduce scrap, improve throughput, shorten cycle times, stabilise production, or make planning less reactive. It needs to know how much data is required, whether new sensors are needed, whether IT will be overwhelmed, whether operators will understand the output, whether plant managers can trust the logic, and how long it takes before the system produces something useful.

This is also why the market has a supply-side problem as there are still too few solution providers building for industrial scalability in the way SMEs actually need. A great deal of industrial technologies, and more specifically “industrial AI”, is still delivered either as generic software that never really adapts to factory-specific reality, or as consulting-heavy customization that turns every project into a long, expensive reinvention. Neither model is good enough, as SMEs need configurable industrial products and systems that can adapt to the specifics of a production context without requiring months of redesign or a resident data science team.

That point matters because manufacturing variation is real and solution providers need to account for different plants, product mixes, constraints, data quality, bottlenecks, and pretending these differences do not exist leads to shallow solutions. But accepting them should not automatically mean punishing the customer with slow, high-friction delivery. The real challenge is mass productization with enough configurability to reflect the reality of industrial operations. That is where Europe should be placing much more pressure on the market in order to package industrial intelligence in a way that can scale.

And this is where the education gap becomes critical.

If Europe is serious about strengthening manufacturing SMEs, education cannot remain an afterthought attached to a grant or a showcase event, but it must become part of industrial policy. Cedefop’s 2025 findings are clear that AI is reshaping jobs and skills across Europe. Yet the public narrative around AI in manufacturing is still too often left to the extremes of techno-utopian promises or workforce anxiety. In between, there is far too little practical education aimed at helping SMEs understand what AI changes and how it can be deployed in ways that support people rather than displace their judgment.

That missing middle is expensive. It slows down demand, distorts procurement, and creates a vacuum where fear fills in the blanks. On the shop floor, AI is often imagined as a control mechanism before it is understood as decision support because too few initiatives have explained industrial AI in the language of production, quality, maintenance, and daily work. If people only encounter AI through headlines, suspicion is a rational response.

There are better models emerging elsewhere, and Europe should pay attention. In places such as Singapore’s Jurong Innovation District, advanced manufacturing is being developed as a shared ecosystem rather than left to the initiative of individual factories alone: demonstration environments, digital twin validation, automation, robotics, and talent development are combined to reduce adoption risk and make innovation more tangible. A similar logic is visible in East Asia more broadly. In South Korea, Changwon is pushing AI manufacturing at industrial district level, supported by public policy, demonstration programmes, research concentration, and a long-term plan to turn the wider manufacturing base into an intelligent platform rather than a collection of isolated experiments. Even corporate-led initiatives are moving in the same direction, as Samsung and the Korea Federation of SMEs are once again opening a smart factory support programme for around 150 SMEs in 2026, now with an explicit AI track.

What these cases have in common is clear: industrial AI is being treated as enabling infrastructure and capability-building at ecosystem level, not as sporadic experimentation delegated to single companies.

To be fair, Europe is not doing nothing. The European Digital Innovation Hubs network has expanded with a reinforced AI focus, and by mid-2025 the network had carried out nearly 16,000 digital maturity assessments, more than 14,500 of them on SMEs. That is valuable in the sense that it shows there is institutional recognition that SMEs need structured support, but this does not by itself solve the adoption bottleneck. https://digital-strategy.ec.europa.eu/en/policies/edihs

That last point deserves saying plainly: manufacturing SMEs do not need luxury innovation.

They need practical systems that fit their economics. Solutions that help operations improve without first demanding a major transformation project and can work with existing data environments, including imperfect ones. Solutions that can speak to the owner worried about margins, the production manager worried about targets, the engineer worried about process behaviour, and the operator who simply wants to know what is happening and what to do next. If a solution cannot make sense across those levels, it is unlikely to stick.

That is exactly where a large share of industrial digitalisation still fails: not because the concept was wrong, but because the solution never became part of ordinary decision-making. If AI remains an overlay rather than an operational layer, adoption will continue to stall.

In summary, what is actually missing in Italy and Europe that would help Manufacturing SMEs increase technological adoption?

1)     Education initiatives designed for industrial roles, not generic digital literacy

2)     Ecosystem mechanisms that reduce first-adoption risk through demonstration, coaching, and trusted local channels

3)     A stronger market push toward productised industrial solutions that are configurable, explainable, and economically viable for SMEs

Until then, Europe will keep producing a strange contradiction: we will say SMEs are the backbone of the economy, then continue offering them innovation in formats built for someone else.