Why 76% of Italian SMEs Do Not Invest in AI (and Why It Is Not About Money)

76% of Italian small and mid-sized companies have neither invested in artificial intelligence nor plan to (Observatory on Digital Innovation in SMEs, Politecnico di Milano, 21 May 2026). The easy reading is that the money is not there. In my experience with these companies the brake is something else: in a small company innovation has no department, it has one person. If that person does not decide, nothing happens.

How many Italian SMEs actually invest in AI?

The gap is stark. 71% of large Italian companies have started at least one artificial intelligence project, against 15% of mid-sized and 7% of small ones. On paid generative AI licences the ratio is harder still: 84% of large companies use them, against 9% of SMEs (Artificial Intelligence Observatory, Politecnico di Milano, February 2026).

Meanwhile the Italian AI market is worth 1.8 billion euro and grew 50% in a year (same source). So the growth is real, but it is concentrated where the companies are large.

There is a third figure, and it explains the other two: only 7% of SMEs have launched structured AI training programmes, and 47% have carried out no research and development activity in the past three years (Observatory on Digital Innovation in SMEs, May 2026).

What does "it costs too much" actually mean?

It is the sentence I hear most often, and it almost never refers to money. Underneath it sit three different things, and each needs a different answer.

One: I cannot see the value. The owner is unable to measure the return, and not because they lack the ability: because nobody ever handed them a method for doing it. This does not call for persuasion, it calls for guidance. The ROI calculator exists precisely for this, and the calculation is done with them, not delivered to them.

Two: I do not believe it works. Scepticism about whether it actually applies is legitimate, and words do not dismantle it. A demo on their own process does, before the company pays for a prototype. Anyone asking you to sign before showing you anything is asking for an act of faith.

Three, and this is the heaviest: I am afraid of what happens to my people. One owner I work with has a process in front of him whose automation would free 84 hours a month of one person's time. He has not done it, and the reason is not the cost: he would not know what to give that person afterwards, and he fears loading them with stress they do not carry today.

That is not resistance to change, it is responsibility towards someone who has worked with you for years. It is also why an AI project in a small company is never only a technical project: 84 hours a month are two working weeks, and the question "what do we do with them" has to be answered before, not after.

Who decides on innovation in a small company?

In most of the small companies we work with, the research and development department does not exist. It was not cut: it was never opened, because R&D is perceived as a cost rather than a resource. The national figure confirms it, with 47% of SMEs carrying out no R&D for three years (Observatory on Digital Innovation in SMEs, May 2026).

The consequence is that innovation in a small company is almost always triggered by the will of the executive at the top. Not by a process, not by a dedicated budget, not by a committee. By one person who at some point decides.

The moment it happens has a recognisable shape: someone takes on the role of innovation leader. Sometimes it is an outside consultant, more often it is an internal manager or director who rises to that role because they want to bring AI into the company as a capability, not as a purchased tool. Without that figure, the project stays a line in the minutes.

In my experience on the ground, the external push today is stronger than it was a year ago. Agentic AI now sits inside almost every piece of software, and the office suite has been largely automated. Many service companies are starting to ask themselves whether this resistance still makes sense, and they are asking it unprompted.

Why does a small company have an advantage over a large one?

This is where the story of the laggard needs turning around, because the gap in the numbers is real but the structural advantage sits with the smaller company.

In a large enterprise AI is not decided by one person: it is decided by a committee. The chain of authority is long, the decision process is heavy, and around the budget sit the interests of several departments competing for it. A technology choice becomes an internal negotiation, and internal negotiations take time.

In a small company that chain is short. The person who decides and the person who pays are often the same, and that person can say yes on Tuesday and see the first result within the month. The bureaucratic lightness that looks like a limitation (no dedicated department, no formalised process) is exactly what allows fast movement, when the will is there.

Read this way, the 71% against 7% gap does not measure a technological difficulty. It measures how many times that person decided.

Where does a small company start?

From a single department, and from a process someone already owns.

A company we work with took its first step in administration and then extended it across three departments. The measured result is three working days saved every month, achieved against the view of colleagues who believed it was not possible. It is not a transformation, it is a piece of work that used to be done by hand and now is not.

The instructive part comes next. The same company tried to extend automation to a second, more technical process and stopped, because there it found gaps in the digitalisation of its own workflow. They are closing them methodologically before putting AI back on top, which is the right call: automating a process with gaps in it means automating the gaps too, faster and across more cases.

This is what I teach when a company asks me where to start. First you check whether the process really exists, then you decide what to automate. The reverse order is the most common way to waste a budget.

Want to know whether your company is ready?

If you recognise yourself in that 76%, the useful question is not "how much does AI cost". It is: is there anyone in here willing to own this as a capability, and which single process is costing me more than it should today.

That is the starting point, and the free Pre-Assessment is built to answer both questions together. It is the first step of our AI consulting path for small and mid-sized companies.

If you would rather form your own view first: there is the AI readiness test, the piece on who should own AI inside a small company, the one on how to work through a team's fears and, if the brake really is budget, the guide to incentives and funded training.

Frequently asked questions

How many Italian SMEs use artificial intelligence?

15% of mid-sized and 7% of small companies have started at least one AI project, against 71% of large companies (Artificial Intelligence Observatory, Politecnico di Milano, February 2026). Looking at future investment, 76% of SMEs report having neither invested nor planned to invest in AI (Observatory on Digital Innovation in SMEs, Politecnico di Milano, May 2026).

How much does a small company need to invest to start with AI?

Less than it expects, if it starts from a single process. The first useful step is not a budget, it is choosing the process: one someone in the company already owns, with a measurable cost in time. The calculation compares that cost in time with the cost of the tool, before anything gets signed.

Why are Italian SMEs behind on AI?

Not for lack of money. Most small companies have no research and development department, and 47% have carried out no R&D for three years (Observatory on Digital Innovation in SMEs, May 2026). Innovation therefore depends on the will of whoever runs the company. When that will is there, a small company moves faster than a large one, where the same decisions go through a committee.