AI adoption by small businesses: OECD 2025 figures
The OECD prepared a paper for the G7 on how small and medium-sized businesses adopt AI and what holds them back. AI adoption by small businesses is far lower: 11.9% of small companies in OECD countries use AI, against 40% of large ones. Below are the figures by company size, industry, skills and training, and the four conditions for adoption that the OECD names.
Key figures
11,9%
of small companies (10–49 employees) in OECD countries use AI; for medium-sized ones (50–249) it is 20.4%, for large ones (250 and more) 40%, data for 2024 or the latest available year
14%
of companies with 10 or more employees in OECD countries used AI in 2024; in 2020 it was 5.6%, and more than 50% used cloud services in 2024
1,9–6,1%
of companies in G7 countries used AI in core production functions in 2024, from Japan to the US; survey years and methods differ by country
29%
of small and medium-sized companies with generative AI use it in their core business
almost 45%
of companies in ICT use AI, more than 25% in professional services, 7.2% in construction and 7.8% in hotels and restaurants
50%
of small and medium-sized companies in four G7 countries say their staff lack the skills for generative AI; in the D4SME survey, more than 50% named not knowing how to use it, and 80% in Japan
fewer than 30%
of small and medium-sized companies with generative AI train their staff to use it: from 11.3% in Japan to 29.4% in Canada
0.2–1.3 percentage points
is what AI may add to annual labor productivity growth in G7 countries over the next ten years, according to estimates the OECD cites
The report in brief
- Published by
- Organisation for Economic Co-operation and Development (OECD)
- Published on
- 9 December 2025
- Original title
- AI adoption by small and medium-sized enterprises: OECD discussion paper for the G7
- Length and language
- 62 pages, in English
- Checked
- 16 October 2026
- Next edition
- a one-off paper; no new edition has been announced, so follow OECD publications on AI and small business
“…large firms using AI (40%) is more than three times that of small firms (11.9%)”Executive summary, p. 6
How the report was made
This is a discussion paper by the OECD Secretariat, prepared at the request of Canada’s G7 presidency for a ministerial meeting in December 2025 (p. 2). It draws on the OECD ICT Access and Usage by Businesses database for companies with 10 or more employees for 2020–2024 (p. 9), national surveys of G7 countries, the 2024 OECD survey on generative AIgenerative AI. Generative AI is a family of neural networks that create new content: text, images, music, video and program code, from a description in plain language. in small and medium-sized businesses and the 2025 D4SME survey.
The OECD names four conditions for adoption: connectivity, AI resources (data, algorithms, computing), skills and financing (pp. 3 and 6). Companies fall into four groups by maturity, from beginners to those who have built AI into their work (p. 6).
What the report does not measure
Countries can be compared only roughly: survey years and definitions differ, and the latest data for the US and Japan in the database is from 2021. The D4SME surveys are not representative, and micro-enterprises with fewer than 10 employees are not in the main statistics.
The financial barrier is described in words, without the share of companies that complain about it (ch. 4, p. 36). The paper does not measure the effect of AI on the revenue of a single company.
What this means for a company in Russia
Russia in the report: the report has no data on Russia, because the country is not an OECD member and is not in its database. Of the countries outside the EU, the figures cover the G7 countries, including the US, Canada and Japan.
- Small companies use AI more than three times less often than large ones: 11.9% against 40% (p. 10). The OECD figures do not transfer directly to Russia, because the methods differ. For a small company, it pays to start not with development but with a list of ready-made AI tools for business.
- Half of small and medium-sized companies in four G7 countries complain about a lack of skills (p. 32), and fewer than 30% train their staff (p. 35). Build training into the deployment plan from the first month: corporate training closes this gap faster than hiring.
- Only 29% of small and medium-sized companies with generative AI use it in their core business (p. 11); the rest try it on support tasks. Choose one process that brings in revenue and test the result on it: this is how a pilot on one process works.
What to compare it with
- Eurostat: AI in EU enterprises 2025: the same gap by company size in the EU, 2025 data
- AI in US companies: BTOS survey 2026: the share of US companies using AI, counted by a different method
- EU Digital Decade 2026: EU targets for AI in companies by 2030 and the reasons companies do not adopt it