Research16 July 2026·By Clerq·11 min read

The state of AI adoption in UK small and mid-sized businesses.

A curated roundup of the UK AI adoption statistics that survive scrutiny. Every figure carries its publisher, report, year and a link to the primary source, plus a plain reading of what it means for operators.

UK AI adoption statistics range from 16% to 54% depending on who was asked, how AI use was defined and when the fieldwork ran. None of those figures is wrong. Each answers a different question. This page collects the numbers worth citing, links every one to its primary source, and adds a one-line operator's reading, which is Clerq's interpretation, not the source's claim.

Direct answer

The picture in one paragraph

As of the most recent survey rounds, ONS found 25% of UK businesses currently using some form of AI (late December 2025), DSIT found 16% of businesses with at least five employees using one or more AI technologies (published January 2026), and BCC and Atos reported 54% of UK firms actively using AI (March 2026). Adoption rises sharply with business size, is growing on every repeated measure, and has so far had little reported effect on workforce size. The figures come from different populations and definitions and should never be combined into one trend line.

The headline UK adoption numbers

Three regularly cited UK sources currently measure business AI adoption. They disagree, and the disagreement is informative.

25%of UK businesses reported currently using some form of AI technology in late December 2025, up 15 percentage points since the question was introduced in late September 2023. Clerq's reading: on the broadest official measure, adoption has roughly gone from niche to a quarter of the economy in about two years.ONS BICS, Jan 2026

16%of surveyed UK businesses with at least five employees were using one or more AI technologies, with larger businesses reporting higher use than micro businesses. Clerq's reading: when the bar is a defined AI technology in use rather than any AI touchpoint, the number roughly halves.DSIT, Jan 2026

54%of UK firms were actively using AI in March 2026, in a survey conducted with Atos where the respondent base skews towards engaged member businesses. Clerq's reading: among firms engaged enough to answer a chamber survey, AI use is now a majority behaviour.BCC + Atos, Mar 2026

The spread from 16% to 54% is not measurement error. ONS asks a broad "any form of AI" question across a large business panel in its Business Insights and Conditions Survey. DSIT's AI Adoption Research asks about specific AI technologies among businesses with five or more employees. The BCC and Atos survey samples chamber-connected firms. Different denominators, different definitions, different answers.

Adoption by business size

44%of businesses with 250 or more employees reported currently using AI, against 25% of businesses overall, in late December 2025. Clerq's reading: the size gradient is the single most consistent finding in UK adoption data; capability follows headcount and budget.ONS BICS, Jan 2026

DSIT's research points the same way: large businesses reported higher use than micro businesses. For a founder-led business in the £1m-20m range, the practical implication is that the constraint is rarely the technology. It is the capacity to scope, integrate and govern it, which larger firms buy with specialist headcount and smaller firms must buy as a defined project.

The barriers businesses actually report

71%of surveyed UK businesses reported lack of identified need as a barrier to AI adoption. Clerq's reading: most businesses have not mapped which of their workflows AI could carry, so they cannot see a need that is usually there.DSIT, Jan 2026

60%of surveyed UK businesses reported limited AI skills, expertise and knowledge as a barrier. Clerq's reading: the skills gap lands hardest on scoping and governance, not on using the tools, which is why unscoped projects stall.DSIT, Jan 2026

These two barriers explain more of the adoption gap than any technology constraint, and they are the pattern behind why AI projects stall in UK businesses: the use case is undefined, so the need never becomes visible and the skills are never pointed at a specific workflow.

What adoption is doing to jobs

95%of SME AI users told BCC and Atos that AI had not affected their workforce size over the previous year, and 86% said job roles were unchanged. Clerq's reading: current SME adoption is augmentation of existing teams, not substitution; the capacity gain is being absorbed, not banked as headcount cuts.BCC + Atos, Mar 2026

4%of businesses currently using AI told the ONS their overall workforce headcount had decreased as a result of those technologies in late December 2025. Clerq's reading: measured job displacement from AI remains marginal in official UK data, whatever the headlines imply.ONS BICS, Jan 2026

Momentum: what is coming next

15%of UK businesses reported plans to adopt some form of AI within the next three months in late December 2025, up 2 percentage points from late September 2025 and the highest since the question was introduced. Clerq's reading: the pipeline of intending adopters is the strongest official signal that the adoption curve has further to run.ONS BICS, Jan 2026

The global context for UK operators

Enterprise-focused global research is context, not a UK SME benchmark, but two findings are worth holding alongside the UK data. McKinsey's The state of AI in 2025: Agents, innovation, and transformation reports 88% of respondents' organisations using AI regularly in at least one business function, while 23% report scaling an agentic AI system somewhere in the enterprise. Clerq's reading: even among the world's most-resourced adopters, using AI is nearly universal while operating scaled agentic workflows remains a minority achievement, so the gap is execution, not access.

Stanford Digital Economy Lab's Enterprise AI Playbook, built from 51 successful implementations, found the hard problems were usually organisational readiness, process redesign, trust and data infrastructure rather than the model. BCG's AI Radar 2026 reports that nearly all surveyed CEOs expect AI agents to produce measurable returns in 2026, which is an expectation rather than a result, and Deloitte's State of AI in the Enterprise 2026 identifies customer support as the function where respondents expect the highest agentic impact. Clerq's reading across all three: expectations are running ahead of measured results, which makes disciplined baselines and measurement more valuable, not less. Whether any of this justifies investment in your business is a payback question, covered in AI ROI for UK SMEs: how to model payback.

How to read these numbers

Before quoting any adoption statistic in a board pack, check four things about it. Most public misuse of these figures fails at least one.

The denominator. "UK businesses" can mean all registered businesses, businesses with five or more employees, or a membership base. The 16% and 54% figures differ mostly because of who was counted, not because adoption tripled between surveys.
The definition. "Using AI" ranges from one employee using a chatbot to a governed workflow in production. Surveys that ask about any use will always report higher numbers than surveys that ask about specific technologies or scaled systems.
Self-selection. Voluntary business surveys over-sample engaged, digitally confident firms. Treat member-panel results as an upper bound on the wider population.
The fieldwork date. Adoption is moving quickly; a figure from 2023 describes a different market. Always cite the survey window, not just the publication year.

A statistic without its denominator is an opinion with a percentage sign.

What the data does not tell you

An honest roundup should also state what is missing, because the gaps are where bad citations breed.

There is no reliable UK figure for SME adoption of agentic AI specifically. McKinsey's 23% agent-scaling figure describes global enterprises, not UK SMEs, and no official UK survey yet separates agents from broader AI use.
There is no official series linking SME AI adoption to measured financial returns. Adoption statistics count usage, not payback; the business case still has to be built workflow by workflow.
There is no single official SME-specific adoption series. ONS covers all businesses with a size breakdown, DSIT covers businesses with five or more employees, and membership surveys cover their panels. Anyone quoting "the UK SME adoption rate" as one number has picked a survey without telling you.
Sector-level SME detail remains thin. Most published sector splits come from small subsamples and should be treated as indicative at best.

The direction of travel across every repeated measure, though, is one way. That trajectory, and why it is structural rather than cyclical, is the argument of AI isn't a fad.

Frequently asked questions

What percentage of UK businesses use AI?

It depends on the survey. ONS found 25% of businesses currently using some form of AI in late December 2025. DSIT research published in January 2026 found 16% of surveyed businesses with at least five employees using one or more AI technologies. BCC and Atos reported 54% active use in March 2026. The surveys use different populations, definitions and methods, so the figures are not interchangeable.

What percentage of UK SMEs use AI?

There is no single official SME-specific series. The closest current figures are BCC and Atos reporting 54% of UK firms actively using AI in March 2026, in a survey where most respondents were SMEs, and ONS reporting 25% of businesses overall in late December 2025, with usage rising sharply by business size.

Is AI adoption in the UK growing?

Yes on every measure that repeats over time. ONS reports current AI use rose 15 percentage points between late September 2023 and late December 2025, and a further 15% of businesses were planning to adopt within three months, the highest since the question was introduced.

Is AI reducing jobs in UK SMEs?

Not at scale in the survey data so far. BCC and Atos found 95% of SME AI users reported no impact on workforce size over the previous year, and ONS found only 4% of AI-using businesses reported a headcount decrease from those technologies.

How reliable are AI adoption statistics?

Treat each figure as an answer to its own survey question, not a universal fact. Results vary with the population sampled, the definition of AI use, self-selection of respondents and when fieldwork ran. Always check the denominator before comparing two numbers.

References

Primary and authoritative sources

Every statistic on this page links to its primary source below. Interpretive one-liners are Clerq's own reading and should not be attributed to the source publishers.

Turn the statistics into a decision about your business.

Adoption data says what the market is doing. The two-week Diagnostic says what your workflows would gain, at what cost, under what controls. The roadmap is yours to use with Clerq or another builder.

See the Diagnostic

← All insights