Skip to main content
Home » Artificial Intelligence » The next digital divide is not access to AI – it’s knowing what to do with it
Future of AI 2026

The next digital divide is not access to AI – it’s knowing what to do with it

Marco Bianchini

Economist, Centre for Entrepreneurship, SMEs, Regions and Cities
OECD

Small firms are no longer watching AI from the sidelines. Across OECD countries, the share of small businesses using AI rose from 7% in 2023 to 17% in 20251 — more than doubling in two years.


Medium-sized businesses show a similar pattern, rising from 14% to 30%2 over the same period. New AI tools based on large language models, now cheap and easy to access online, are changing what AI means for smaller firms.

For now, uptake appears broad but shallow

In 2025, the OECD supported the Canadian G7 Presidency by proposing a taxonomy of SME AI adopters, based on digital maturity, scope of AI adoption and the tools they use. The aim was to look beyond whether firms use AI and ask why and how they use it.

In a group of 2,000 digitally active SMEs surveyed by the OECD in 2026,2 61% reported using at least one AI-enabled application. Of these, 76% were “AI novices”: firms using standard, off-the-shelf tools for isolated tasks. Only 5% used customised AI, and just 3.6% deployed agentic AI. 

The OECD identifies four key enablers: connectivity data, algorithms and computer skills and finance

Off-the-shelf AI is most often used for marketing, cited by 70% of its users. Customised AI is most often used for demand prediction (39%).

More than half of surveyed SMEs (54%) reported at least moderate benefits from AI. But only 21% saw a significant or transformational impact.

The danger is a new digital divide. Across the OECD, 52% of large firms used AI in 2025, compared with 17% of small firms. The gap is widening too: the firm-size gap in AI use reached about 35 percentage points in 2025, up from 23 points in 2023.

Policy can help bridge the gap

The OECD identifies four key enablers: connectivity data, algorithms and computer skills and finance.

That means affordable high-speed access, shared data infrastructure, trusted AI tools, awareness campaigns, practical training and patient funding.

Support should be tailored. A shop testing chatbots doesn’t need the same help as a manufacturer building predictive systems. Many OECD governments recognise this challenge, and are working to make AI adoption simple, productive and safe for small businesses.


[1] OECD (2026), “Empowering SMEs in the age of AI: The 2026 OECD D4SME Survey”, OECD SME and Entrepreneurship Papers, No. 78, OECD Publishing, Paris, https://doi.org/10.1787/bf5a9816-en.
[2]OECD ICT Access and Usage By Business database, Accessed May 2026 (https://data-explorer.oecd.org/vis?df[ds]=DisseminateFinalDMZ&df[id]=DSD_ICT_B@DF_BUSINESSES&df[ag]=OECD.STI.DEP

Next article