The Office for National Statistics published new figures on 20 July 2026 showing self-reported AI use among UK businesses with 10 or more employees has risen from around 12% in late 2023 to around 35% now — nearly a threefold increase. It’s the number every trade publication is running with this week. The number that matters more sits underneath it: the average number of distinct AI technologies used per adopting business has moved only from 1.4 to 1.6 over the same period, and just 10% of adopters report extensive use. Large language models are the single most common category, used by 18% of all UK businesses, ahead of visual content tools (16%), machine-learning data processing (12%) and image processing (6%).
Adoption also splits sharply by size and sector. Only 28% of businesses with fewer than 10 employees report using any AI technology, against 49% of businesses with 250 or more staff. Information and communication leads every sector at 58% adoption; construction trails at 13%. Read together, the picture is a large number of firms that have picked up one AI tool — most often a chat interface bolted onto existing workflows — and stopped there, against a much smaller group actually building AI into how they operate.
So what
“AI adoption tripled” is true and also the less useful half of this story for anyone running a business. The gap that ONS is measuring — between having touched an AI tool and having built AI into your actual product or operations — is the gap that separates a ChatGPT subscription from a competitive advantage. If your business is one of the 35% that adopted early and is still sitting at 1.6 tools with no real integration, the opportunity isn’t “adopt more AI,” it’s turning shallow tool use into something that’s actually built into your software and workflows rather than bolted on beside them. That’s the work we do on AI-assisted development engagements, or get in touch if you want to work out what “deeper” should actually mean for your business.