The 21-point gap between small firms and big business

Official Office for National Statistics figures for June 2026 put AI use among UK businesses with 0–9 employees at 28%. Among firms with 250 or more staff, it is 49%. That 21 percentage point gap sits over a part of the economy that is not marginal: 5.42 million businesses employ between zero and nine people — 95.3% of every private sector business in the country. (DBT, Business Population Estimates 2025)

Appetite is not the problem. The Federation of Small Businesses recorded AI use among small firms rising from 20% in 2023 to 55% in 2025. The direction of travel is clear. The depth is not.

Why the numbers stay low

The most common reason given by businesses that haven’t adopted AI is not cost, and it is not risk. DSIT’s AI Adoption Research found 71% of non-adopters said they had no identified business need — meaning nobody had shown them what it would actually do for their business. 60% also cited limited skills, and 48% couldn’t find tools that suited them.

Cost bites hardest at the smallest end: 33% of micro businesses say software is too expensive, against 23% of medium-sized firms. (DBT/Ipsos, 2025)

Trust is the other brake, and it is tightening rather than easing. FSB found concern about AI risks among small firms rose from 73% in 2023 to 92% in 2025 — led by inaccuracy (54%), security and intellectual property (39%) and unclear training data (38%).

What actually reassures

Government research points at a missing ingredient rather than a missing product. The Department for Business and Trade found 26% of the smallest micro firms (1–4 staff) received no implementation support at all after buying new technology. Owners in that research put what they wanted plainly:

“Not going to sell us something… just talk us through and support us.”

Small business owner, DBT research

What it’s worth

The returns are measured, not promised. Among small firms already using AI, FSB reports 59% see productivity gains and 24% report higher revenue. And the smallest adopters commit hardest — ONS analysis indicates 17% of 0–9 employee adopters use AI extensively, against 9% of adopters in firms with 250 or more staff. The barrier is reaching the first working system, not what follows it.