AI Adoption in UK Businesses: The 2026 Figures

Part of our topic guide on AI Skills for Business.

By James Cotton, Founder, iO-Sphere

What the 32% counts, and who it leaves out

BIST's AI adoption survey 2026 (published 29 October 2026) counts a firm as using AI when it currently uses, or is piloting, any one of five named technologies, from text generation to agentic AI. BIST is the Department for Business, Innovation, Science and Trade. A firm piloting a single tool counts as a user.

The base is UK private-sector businesses with five or more employees, interviewed from 7 January to 21 April 2026, with public administration and defence, education, and health and social work excluded. No business with fewer than five employees and no public body is in any of its figures, so none of them describes all UK businesses.

Across all firms with five or more staff in January to April 2026, 32% used AI, 6% planned to, 24% neither used it nor had plans, and 38% called it "not relevant to the organisation", down from 51% in 2025.

Each answer came from the person responsible for technology in the business. The figures are that person's report for the firm, which leaves out what staff do with AI on their own.

The 2025 report, published by the Department for Science, Innovation and Technology (DSIT) on 28 January 2026, said: "A limitation of this approach is that the survey will not provide insights into shadow AI adoption", meaning use the business has not approved. The 2026 follow-up interviews found individuals often taking up basic tools "without seeking approval from senior team members".

The 2025 wave, interviewed from February to May 2025, asked the same questions about the same five technologies and found 16%. The report says adoption among firms with five or more staff doubled "in the last 12 months", to 32% in January to April 2026.

AI adoption by firm size, sector and nation

The table gives the share of UK private-sector businesses with five or more employees using AI in January to April 2026, by size band, beside the 2025 wave (February to May 2025).

Employees	Used AI, January to April 2026	2025 wave
5 to 9	27%	14%
10 to 49	34%	16%
50 to 249	47%	23%
250 or more	52% (130 interviews)	36%
All with five or more	32%	16%

In the report's words, "For micro, small and medium-sized businesses the proportion adopting AI nearly doubled", and it calls the increase for large businesses significant.

By sector, information and communication led at 61% among firms with five or more staff in January to April 2026, ahead of real estate at 45% and of finance (101 interviews) and business services and administration at 39% each. Finance and real estate are reported apart for the first time, so neither has a 2025 figure.

The lowest four were construction and retail and distribution at 25% each, transport and storage at 24%, and hotels and catering at 23%. In those four sectors, 69% to 73% of firms with five or more staff neither used AI nor planned to in January to April 2026.

In January to April 2026, firms with five or more staff used AI at 32% in England and Wales, 30% in Northern Ireland (90 interviews) and 26% in Scotland; English regions ran from 30% in the North East (99 interviews) to 35% in the South West. The report found the regional variation of 2025 "hardly evident in 2026, with no significant differences in the proportion who had adopted AI."

How the survey was run, and how to quote it

DSIT commissioned BMG Research to run the survey, and BIST published the results. BMG ran 3,503 telephone interviews with firms of five or more staff from 7 January to 21 April 2026, averaging 15 minutes, sampled from a commercial business database with targets by size and sector.

Answers were weighted by size and sector, and the regional split was found "in line with the population", so no regional weight was applied. The results are shares of firms; the report puts no count of UK businesses behind them.

Of the 69,370 businesses called at least once for the January to April 2026 fieldwork, the response rate was 5%, against 9% in the 2025 wave. The survey planned 400 interviews with firms of 250 or more employees and achieved 130. Its 50 follow-up interviews, which set out to over-represent firms using AI, and 7 site visits give illustrations with no percentages.

Both waves used the same population, business database, telephone mode, respondent, weighting and adoption questions. The contractor changed (IFF Research and Technopolis Group ran 2025), interviews shortened from 20 minutes to 15, seven questions were new or amended, and the barrier and planned-investment answer lists changed. So adoption compares across the waves; the barrier and investment figures do not.

Quote the survey with its base and period (UK private-sector firms with five or more employees, January to April 2026), with "doubled" as the report's word, and with each barrier figure tied to the stage of the firms that chose it. gov.uk files it under research and analysis, and its collection page says "The survey is now annual".

Why ONS and other surveys report other numbers

The Office for National Statistics (ONS) puts AI use at 30% of all businesses and 38% of those with 10 or more employees in September 2026, in its bulletin of 8 October 2026, which labels them "official statistics in development".

They differ from BIST's survey of UK private-sector businesses with five or more employees, January to April 2026, because ONS runs a voluntary online survey that includes firms with fewer than ten employees, leaves out agriculture, finance and insurance, public administration, public education and health, and covers only Great Britain for construction, retail and services.

ONS also counts AI through six named technologies. Its article of 20 July 2026 warns that its headline "does not distinguish between light use, user-level use, and more embedded, production-level adoption", and that "estimates of AI adoption can vary substantially depending on the measurement approach".

Worked out by us from ONS's published dataset, its all-business figure rose from about 18% in March 2025 to about 26% in March 2026 (months both survey waves covered), a rise short of the doubling the report gives for its own measure. ONS's 2026 adoption levels by firm size sit close to the survey's.

The British Chambers of Commerce, in research with Atos, reported in a release of 18 March 2026 that 54% of 668 responding firms, around 94% of them small and medium-sized, were actively using AI.

Microsoft's UK Stories feature of 20 October 2026 cites its UK AI Diffusion Report, which puts generative AI use among people aged 15 to 64 at 43.7% in Q2 2026, a share of people measured from usage telemetry.

Inside adopting firms: how many staff use AI, how often, on what

Among firms with five or more staff using AI in January to April 2026, 77% said fewer than half their staff used it, and on average 31% did, which the report calls "consistent with 2025". The share is highest at the smallest adopters, 38% on average at firms with 5 to 9 employees, and lowest at the largest.

By sector, more than half of staff used AI at 43% of information and communication adopters and 37% in business services, against 4% in hotels and catering and 9% in retail (five or more staff, January to April 2026).

Adopters used AI often: 57% constantly and 27% at least weekly, 84% weekly or more in all. Text generation leads the technologies: 30% of all firms with five or more staff used it in January to April 2026, against 9% for computer vision and image tools, the next. Adopters most often used off-the-shelf AI applications (62%) and automated tasks within jobs, short of whole jobs (58%).

Among firms with five or more staff using or planning to use AI in January to April 2026, administrative and support work led the tasks at 82%, then creative and content work (75%) and data and analytics (63%). Administration (81%) and marketing (71%) led the business areas, with HR (33%) and finance (30%) further down.

What AI has changed for firms so far

Asked to rate AI's importance on an 11-point scale from 0, "not important at all", to 10, "critically important", 14% of firms with five or more staff using AI in January to April 2026 gave it 8, 9 or 10, and 44% gave it 3 or below.

Half of those adopters (50%) reported higher employee productivity since adopting, down from 56% in 2025, while 4% reported a fall; 7% reported a rise of 20% or more, against 16% in 2025. On operating costs, 64% reported no impact, 22% a slight reduction and 2% a "significant reduction", while 9% reported an increase (five or more staff, January to April 2026).

The report concludes: "AI delivered measurable but largely incremental improvements." It adds: "Overall, AI was functioning primarily as a support tool, augmenting existing activities rather than redefining business models."

The report also picks out 101 "intensive adopters", 2% of firms surveyed (five or more staff, January to April 2026): firms that had built bespoke AI or embedded it in their processes, rated it highly important and used it constantly. On average 60% of their staff used AI, against 31% across all adopters, and 81% reported higher employee productivity, against 50%.

The group is selected on how deeply it uses AI, so these contrasts describe it and leave the cause open.

What held firms back from adopting AI

Firms chose barriers from a prompted list and could pick several. Those already using AI were asked which factors "prevented you from adopting AI in your organisation in the past"; firms planning to adopt, and those with no plans, were asked what prevents them now. For adopters, every answer looks back to before they adopted.

From that list, 67% of firms with five or more staff using AI in January to April 2026 chose "limited AI skills, expertise or knowledge" as something that had held them back. Limited knowledge of how AI could help the business followed at 64%, and bias or accuracy concerns at 56%; 5% faced no barriers.

Among firms planning to adopt (236 interviews, the smallest group), 78% chose limited AI skills from the same list as a current barrier. Among firms with no plans, the business case led: 65% chose no clear business case or proven return on investment, with limited AI skills second at 63% (five or more staff, January to April 2026).

Asked for the single most significant barrier on the list, adopters (five or more staff, January to April 2026) put limited knowledge of how AI could help the business first (21%), just ahead of limited AI skills (19%), and first across all 3,503 firms taken together (21%).

The report's reading: "Barriers to adoption were primarily related to capability and understanding rather than access to technology."

ONS asks a different question: which factors "in the last three months" had "prevented or delayed" adoption, with options for firms that had not tried AI or were not held back. Its Wave 165 dataset shows the level of expertise in AI as the most-named factor for businesses with 10 or more employees in September 2026, at 13.7%. The question design accounts for the gap with the survey.

What firms plan next, and what they said about training

Over the next 12 months, firms with five or more staff using or planning to use AI in January to April 2026 put off-the-shelf AI applications first among planned investments (61%), ahead of automating tasks (57%) and embedding AI in existing systems (54%). Reskilling and workforce development came fifth, at 41%, behind research and development.

They expect use to spread: about a third (33%) expect more than half their staff to use AI within one to two years, where currently 22% of adopters have more than half using it (five or more staff, January to April 2026).

The figure for training already given comes from ONS: in June 2026, of businesses with 10 or more employees that use AI or are not sure whether they do, 11% had trained more than half their workforce in AI, says its article on AI in UK businesses, and its dataset shows 31.1% had trained none.

BIST's 50 follow-up interviews, which leaned towards firms using AI, give illustrations. "Hands-on, experiential learning was preferred", many were teaching themselves from webinars and newsletters, and training was seen as "something which needed to be ongoing". Interviewees found gaps among older, more experienced staff, and described them, like managers, as more reluctant to engage with AI.

In one case study, a 40-staff conference and events venue, the parent organisation introduced Copilot about six months before the interview, and completing an online training programme was compulsory to receive a licence. Since then, "few colleagues were making regular use of the tool"; the operations manager called the training "quite basic and simple, just a couple of hours of training".

He thought the rollout would have benefited from "more practical, facilitated training" and "clearer guidance on specific tasks it could be used for within different roles". It is one site and one person's account.

What the figures mean for a firm choosing training

Benchmark your firm on its size band and sector, and on the share of its own staff who use AI, which the firm-level figure hides: a firm counts the same whether a handful of its people use AI or most of them do.

The barrier the figures name most is capability: AI skills, and an understanding of what AI could do in this business. Training addresses that, starting from where your own firm's gap sits, found with a skills gap analysis you run yourself.

Our view is that the training to buy, whether team training or an apprenticeship, is built on your team's own tasks, since a company asking for AI skills is usually asking for people who get more of their own work done with AI. It should run as ongoing, hands-on training, the form the interviewees preferred.

For workforce AI skills, the funded apprenticeship routes are AI Transformation (Level 4) and Data & AI Essentials (Level 3).

Before any training starts, turn the survey's measures into your own baseline: count how many of your people use AI each week and on which tasks, and ask each team what held it back. Count the same things again once the training has run, and set the two counts side by side.

Frequently asked questions

How much do UK firms spend on AI?

47% of firms with five or more staff using or planning to use AI that could estimate it spent nothing on AI in 2025 (BIST's survey, January to April 2026). The survey's data tables give a median of £100; the mean was £7,500, or £25,000 at large firms (26 interviews). The report links the £0 answers to AI built into existing software and to free or low-cost tools.

Do firms check what AI produces?

81% of firms with five or more staff using AI reported at least some human input or checking of its output in January to April 2026, down from 84% in 2025, and the share with none rose from 2% to 5%. The follow-up interviews found responsible use resting on informal habits, with no documented AI use policy or governance yet in most cases.

Is AI cutting jobs at UK firms?

6% of businesses with 10 or more employees using AI reported reduced headcount as a result, and fewer than 1% an increase, in September 2026 (ONS bulletin, 8 October 2026). BIST's survey (five or more staff, January to April 2026) found "no firm evidence of reductions in workforce to date", and interviewees expected the size and make-up of their workforces to stay broadly similar to 2030.