How many Italian companies use AI in 2026: the data by size and sector

01 How many Italian companies use artificial intelligence?

In 2025, 16.4% of Italian companies with at least 10 employees used at least one artificial intelligence technology. The figure comes from Istat’s yearly survey of digital technologies in companies, published on 15 December 2025: it was 8.2% in 2024 and 5.0% in 2023. In two years the share more than tripled, yet it remains a minority: 83.6% of companies used none of the technologies counted.

What “using AI” means in official statistics: using at least one of eight technologies. They range from extracting information from text documents to generating text, code, images, video and audio, from speech and image recognition to machine learning for data analysis, workflow automation, and robots or vehicles that move on their own. The survey asks which technologies a company uses and in which business areas, not how much: a company trying out one tool counts like one that has built it into its processes.

Sources: Istat, Imprese e ICT 2025; Eurostat, dataset isoc_eb_ai (updated 15 June 2026). Industry and non-financial services. Rounded to one decimal: the EU 2025 figure is 19.95%.

The 2025 jump needs one caveat. From that year the European questionnaire, which Istat applies, also counts the generation of images, video and audio, which was not asked in 2024. Part of the growth may come from the broader question. The figure is still the most solid available, with a 95% confidence interval between 15.5% and 17.3%.

02 Is Italy behind Europe on AI?

Yes, but less than a year ago. In 2025 the EU average is 19.95%: Italy ranks 18th of 27, below Germany (26.0%), Spain (20.3%) and France (18.2%). Denmark (42.0%), Finland (37.8%) and Sweden (35.0%) lead; Romania, Poland and Bulgaria close the list, all below 9%.

Speed is what matters most. In 2025 Italy gained 8.2 points, Europe 6.5. The gap to the average fell from 5.3 to 3.6 points, and in relative terms Italy went from 61% to 82% of the European level.

The EU Digital Decade target asks for 75% of companies to use AI by 2030. At 16.4%, Italy has covered 21.9% of the way, by Istat’s own calculation.

03 How much does company size matter?

Almost entirely. In 2025 AI is used by 53.1% of companies with at least 250 employees and by 14.2% of those with 10 to 49: more than one large company in two, one small company in seven. Medium-sized companies sit in between: 25.9% between 50 and 99 employees, 30.8% between 100 and 249.

Source: Istat, Imprese e ICT 2025, Prospetto A. Growth: 2025 value divided by 2023 value, our calculation.

The growth column holds the least obvious finding. The companies that accelerated most are neither the largest nor the smallest, but those with 50 to 99 employees: from 5.6% to 25.9% in two years, more than four and a half times.

In percentage points, though, the gap between large companies and SMEs is widening: about 20 points in 2023, 37 in 2025. Istat notes that for most other digital indicators the size gap is closing. For AI it is the opposite.

Source: Eurostat, isoc_eb_ai. The 50 to 249 class is the one Eurostat uses.

Against Europe, large Italian companies are almost level (53.1% against 55.0%). The gap sits with small and medium companies, just under 3 points in each class. And since the national average (16.4%) is much closer to the small companies’ figure than to the large ones’, it is the small companies that set the pace of the whole country.

04 Which sectors use AI the most?

The ones that live on information. In 2025 AI is used by 51.3% of companies in information and communication and by 35.7% of those in professional, scientific and technical services (law and accounting firms, engineering, consulting, research). Construction (10.4%), transport and storage (11.6%) and accommodation and food services (11.7%) are at the bottom.

Source: Eurostat, isoc_eb_ain2 (Istat data for Italy), companies with at least 10 employees. “Energy, water and waste” combines sections D and E, which Eurostat does not publish separately for Italy because the estimates are unreliable.

Set against Europe, the table shows where the gap lies. It is not uniform. In trade, construction, accommodation and food services and transport, Italy is level with the European average, within half a point either way. The gap is concentrated in knowledge services: more than 11 points in information and communication, almost 11 in real estate, almost 5 in professional services. And in manufacturing, 2.5 points.

Within manufacturing the picture is patchy.

Sources: Istat (Italy) and Eurostat (EU). Machinery and electrical equipment and transport equipment are left out: for Italy Eurostat flags the 2025 estimates as unreliable.

Italian food companies are above the European average (15.3% against 12.5%), chemicals and pharmaceuticals are in line. The widest gaps are in basic metals and metal products (8.8% against 14.3%) and in wood, paper and printing (8.3% against 15.6%).

By area the map is the expected one: 19.3% in the North-West, 17.6% in the North-East, 15.2% in the Centre, 12.2% in the South and the Islands.

05 What do Italian companies use AI for?

Mostly for reading. Among companies that use AI, 70.8% use it to extract information from text documents, 59.1% to generate text, code, images, video or audio, 41.3% for speech recognition. Further behind: machine learning for data analysis (20.0%), image recognition and workflow automation (about 18% each), autonomous robots and vehicles (5.9%).

Sources: Istat, Imprese e ICT 2025 (table 9a); Eurostat, isoc_eb_ai. A company can use more than one technology.

The most interesting figure is in the first row. In extracting information from documents, Italy is level with Europe (11.6% against 11.8% of companies). The Italian gap lies entirely in the other technologies: generating text and images, machine learning, automation. Italian companies that choose AI use it first of all to get at what is already written in their documents.

By business area, companies that use AI apply it mostly to marketing and sales (33.1%), to organizing administrative processes (25.7%) and to research and development (20.0%), followed by production (17.5%), IT security (12.1%), accounting and finance (10.2%) and logistics (6.1%).

One share is growing faster than the others: 33.4% of companies that say they use AI name none of these areas, against 15.5% in 2024. Almost all are small (83.3% have fewer than 50 employees). Istat reads it as an adoption “increasingly widespread but still not very structured”: AI is there, but nobody has decided what it is for.

06 Why do other sources give different numbers?

Because they count different things. Alongside Istat there are other serious sources, and none contradicts the others once you look at whom they count and what they ask.

Sources: Istat; Bank of Italy, Annual Report on 2025 and Occasional Paper no. 1005; Osservatorio Artificial Intelligence, Politecnico di Milano, February 2026.

The Bank of Italy counts more than Istat for two reasons: it starts at 20 employees, so it leaves out the smallest companies, where AI is least common; and it counts as adopters those who use AI in a limited or experimental way. Its figure is most useful for the trend: the share doubled in 2025 to 27% and rose to 32% in early 2026. It ranges from 28% of companies with 20 to 49 employees to 63% of those with 500 or more, from 27% in manufacturing to 36% in services.

The same survey shows how much AI matters, for now: 70% of companies that use it have seen no effect on labor productivity (59% among those using it extensively) and 84% no effect on headcount. Half expect a positive effect on productivity in the next three years.

07 How many people use AI at work in Italy?

It depends on who asks, and how. Estimates range from one employed person in eight to almost one worker in two, and the distance is not an error: these are different questions put to different people.

Sources: Eurostat, isoc_ai_iaiu; Microsoft Italy, 9 May 2024, and Work Trend Index 2024; Osservatorio Artificial Intelligence, February 2026; ECB blog, 26 August 2026.

Eurostat’s measure is the strictest: an official household survey, representative of all employed people, manual jobs included, and limited to generative AI in the last three months. The others question samples of workers, often office workers, and count any AI tool. All of them agree on two points: use is growing fast, and Italy sits below the European average.

For Italy, the Eurostat survey adds an uncomfortable detail: in 2025, 19.9% of people aged 16 to 74 had used generative AI tools in the previous three months, against 32.7% across the EU. Only Romania did worse.

08 Why are people moving faster than companies?

One reason is simple: to use AI, a person needs an account; a company needs a decision. The result is use that bypasses company tools. According to the Politecnico di Milano observatory, only 19% of workers who use AI do so exclusively with tools their company provides, and eight in ten use tools the company does not provide. In 2024 Microsoft measured the same thing for Italy: 73% used AI tools outside the company perimeter, against 78% worldwide.

Large companies are reacting, unevenly. Among those studied by the observatory, 84% have bought licenses for at least one generative AI tool, four in ten have written usage guidelines, 24% ban tools the company does not provide, and just 9% have structured AI governance. On the AI Act obligations, more than one company in two has started AI literacy initiatives, but only 15% has a structured compliance project.

For SMEs the observatory’s numbers are lower: trials have been started by 15% of medium-sized and 7% of small companies; 9% use paid generative AI tools and another 9% free ones. Meanwhile the Italian AI market reached €1.8 billion in 2025, up 50% on 2024, and SMEs account for 18% of it.

What gets lost in this gap, work done in personal accounts that never becomes company knowledge, is the subject of one of our articles: Personal ChatGPT is not your company’s memory.

09 Why do many companies not adopt AI?

Among companies that do not use AI, 11.5% have at least considered it (4.6% in 2023). Those that looked into it and stopped point first of all to a lack of skills.

Sources: Istat (table 9c); for the EU, Eurostat, Statistics Explained, which publishes only these four reasons.

Two readings. First: Italian companies that considered AI cite each obstacle less often than the European average, skills included. Second, and more important: in Italy 11.5% of companies that do not use AI have considered it, in Europe 14.0% (our calculation on Eurostat data: 11.1% of companies considered AI without adopting it, out of 79.1% that do not use it). The Italian lag starts before the obstacles: many companies have not yet reached the point of asking the question.

The Bank of Italy finds a similar picture among companies with at least 50 employees, in early 2026: about a third of those that do not use AI, or use it little, see no particular obstacle. Among the others, the main reason is that AI seems to fit their business poorly; for those already using it a little, the lack of skills is what holds back wider use.

10 What do these numbers mean for an SME?

• You are not behind, you are average. Among companies with 10 to 49 employees, one in seven uses AI (14.2%); among those with 50 to 99, one in four (25.9%). But the average is moving fast: it more than tripled in two years.
• Your people are probably using it already. The problem is not introducing AI, it is that today it runs through tools the company did not choose: only 19% of users stick to company tools.
• The first use is nearly always the same: documents. 70.8% of companies that use AI apply it to extract information from text. It is also the use in which Italy is level with Europe.
• Deciding what it is for is half the work. A third of companies that use AI cannot name the area it serves: spontaneous use, not a project. And among companies with at least 20 employees that use it, 70% have not yet seen effects on productivity.
• Skills are obstacle number one. 58.6% of companies that stopped cite a lack of skills. Since 2 February 2025, moreover, anyone using AI systems in a company must support the AI literacy of their staff (AI Act, article 4).

Where to start, in order:

1. Ask people what they already use, and for what. Ten minutes of anonymous questionnaire say more than any estimate.
2. Pick one measurable use built on your documents: the answers people now ask a colleague for, the datasheets, the procedures.
3. Decide which data can go where, and give people a company tool before banning personal ones.
4. Train people on the chosen tool and on its limits.
5. Measure before and after, on a few things: time to find a piece of information, repeated questions to colleagues.

To choose the tool, we wrote a guide with twelve questions to ask vendors: How to choose an AI assistant for your company. And if your first use is your documents, that is the work ekory does: it connects the folders you already use and answers with the document and page attached.

11 Methodology and limits

• Sources. Istat (press release, tables and methodological note of Imprese e ICT 2025, 15 December 2025); Eurostat (datasets isoc_eb_ai, isoc_eb_ain2 and isoc_ai_iaiu, updated in June 2026, and the Statistics Explained articles); Bank of Italy (Annual Report on 2025, 29 May 2026; Occasional Papers no. 1005, March 2026, and no. 946, June 2025); Osservatorio Artificial Intelligence of the Politecnico di Milano (press release of February 2026); Microsoft (Work Trend Index 2024); European Central Bank (blog of 26 August 2026). All consulted on 10 October 2026.
• Who is counted. The Istat survey covers companies with at least 10 employees in industry and services, excluding agriculture, mining and finance: a population of 218,446 companies, 17,041 respondents, collected between May and July 2025. Micro-enterprises are left out: they are 94.9% of Italian companies and employ 42.3% of the workforce (Istat, 2022 data). No official AI figure exists for them.
• What “using” means. For Istat and Eurostat one of the eight technologies is enough, at any intensity. From 2025 the questionnaire includes generating images, video and audio: part of the growth may come from there.
• Precision. For the total, the 95% confidence interval runs from 15.5% to 17.3%. Estimates for single industries are less precise: we left out those Eurostat flags as unreliable for Italy (machinery and electrical equipment, transport equipment, and energy and waste taken separately).
• Our calculations. Italy’s rank (18th of 27) orders Eurostat’s 2025 figures for the 27 member states. Growth by company size is the ratio between Istat’s 2025 and 2023 values. The share of non-users that considered AI is, for the EU, 11.1 divided by 79.1 (Eurostat): 14.0%. The same calculation for Italy (9.4 out of 82.3) gives the 11.5% published by Istat.
• Workers. Measures of individual use cannot be added to, or directly compared with, measures of company adoption: they count people, not companies, with different questions. The observatory’s press release does not state its sample; an article written with the observatory’s researcher describes an online survey of 3,000 workers in Italy, France and the United Kingdom, and mentions “one worker in three” using only company tools, against the 19% of the press release: we used the press release.
• What we did not publish. Estimates of AI use in micro-enterprises, for which no official source exists, and a few percentages circulating online without a primary source we could check.

Frequently asked questions

Sources

1. Istat, Imprese e ICT, anno 2025, press release, tables and methodological note, 15 December 2025.
2. Eurostat, Use of artificial intelligence in enterprises, Statistics Explained, data extracted December 2025.
3. Eurostat, dataset isoc_eb_ai and isoc_eb_ain2, updated 15 June 2026.
4. Eurostat, Use of artificial intelligence by individuals, Statistics Explained, data extracted March 2026, and dataset isoc_ai_iaiu.
5. Bank of Italy, Relazione annuale sul 2025, box “Intelligenza artificiale: adozione ed effetti sulle imprese”, 29 May 2026.
6. T. Ropele and A. Tagliabracci, The economic impact of artificial intelligence: evidence from Italian firms, Bank of Italy, Occasional Paper no. 1005, March 2026.
7. L. Bencivelli, S. Formai, E. Mattevi and T. Padellini, Embracing the digital transition: the adoption of cloud computing and AI by Italian firms, Bank of Italy, Occasional Paper no. 946, June 2025.
8. Osservatorio Artificial Intelligence, Politecnico di Milano, Intelligenza Artificiale in Italia, press release, February 2026.
9. Microsoft and LinkedIn, 2024 Work Trend Index Annual Report, 8 May 2024, and Microsoft Italy press release, 9 May 2024.
10. European Central Bank, AI adoption and the productivity promise: what workers report, ECB Blog, 26 August 2026.
11. Istat, Annuario statistico italiano 2025, chapter 14, Imprese.
12. EconomyUp, AI e lavoro: l’impatto reale tra efficienza produttiva e rischio di debito cognitivo, 13 February 2026 (survey method of the observatory).