AI arrives late at the small business, and worse than the cloud did
In the OECD a large firm is more than three times as likely to use AI as a small one. With cloud the difference was twofold. The distance is widening, not closing.
Executive summary
In 2024, 40% of OECD firms with 250 or more employees used artificial intelligence, against 20.4% of those with 50 to 249 and 11.9% of those with 10 to 49. The gap is more than threefold.
This is not the usual digital divide. Small firms are about half as likely as large ones to buy cloud services, but less than a third as likely to use AI: the distance is specific to this technology.
The stated blocker is not price. In the European Union, 70.9% of firms that considered adopting AI and did not point to a lack of in-house expertise; only 20.7% say it is not useful for them.
Some 86% of employers worldwide expect AI to transform their business before 2030 and 63% identify the skills gap as the main barrier. The constraint is people, not licences.
Context
Every recent technology wave reached large firms first and then got cheap enough to reach small ones. Cloud followed that path and its size gap narrowed over the years.
With artificial intelligence the opposite is happening. The cost of access fell faster than in any previous wave and the size gap is wider, because what is scarce is not the tool but someone who can integrate it.
Figure 1 · AI use by firm size across the OECD, 2024
% of firms in each size bandTable 1 · The size gap, technology by technology
% of firms in each size band| Indicator | Small | Medium | Large | Source |
|---|---|---|---|---|
| AI use, European Union, 2025 | 17.0% | 30.4% | 55.0% | Eurostat |
| AI use, OECD, 2024 | 11.9% | 20.4% | 40.0% | OECD |
| CRM use, Brazil, 2025 | 29% | 41% | 56% | Cetic.br |
Business implications
Pick a process, not a tool. Adoption that works in a small firm starts with the task that eats the most hours, not with the most complete platform on the market.
Buy integration time, not licences. Software cost stopped being the barrier. The cost of whoever makes it work inside the process, however, is still there.
Measure in hours recovered, not active users. It is the only metric a small-business owner can verify without a dashboard, and the only one that justifies investing again next quarter.
Methodology & data
Three sources with published methodology: the OECD ICT Access and Usage by Businesses database, consulted in July 2025; Eurostat's EU ICT usage survey, with fieldwork in the first months of 2025; and Cetic.br's ICT Enterprises 2025 survey of 4,174 Brazilian firms. All cover firms with ten or more employees.
No equivalent regional measurement exists for Latin America; the Brazilian data are used as a reference for the largest market, not as a regional average.
References
OECD (2025). AI adoption by small and medium-sized enterprises. G7 Presidency discussion paper. · Eurostat (2025). Use of artificial intelligence in enterprises, Statistics Explained. · World Economic Forum (2025). Future of Jobs Report 2025. · CGI.br/NIC.br - Cetic.br (2026). ICT Enterprises 2025: main results.