UK SMEs increase AI spending as Anthropic edges out OpenAI, research shows
Capital on Tap research reported by TechRadar shows about one in eight UK SMEs now pay for AI tools, with adoption rising to 12.8% in 2026. Anthropic’s Claude leads spending, while more than a quarter of buyers use two or more providers.
The study by Capital on Tap, a small business credit card provider, estimated that AI adoption among UK SMEs rose from 1.1% in 2023 to 12.8% in 2026. That means roughly one out of every eight UK SMEs is paying for AI tools, according to the findings.
Anthropic was the most popular AI provider in the data, accounting for 54.1% of all AI spending by the SMEs analyzed. OpenAI and ChatGPT ranked second with 27.3%. Average quarterly spend grew to £75.60 per SME. More than 28% of SMEs were using two or more AI platforms.
The presence of tools including Perplexity, Replit and Lovable in the list suggests many of the SMEs analyzed are in software development, TechRadar reported. But publishing led AI adoption by sector at 26%, followed by radio and television and consulting at 21.6% each and telecommunications at 20.3%.
Damian Brychcy, chief executive officer of Capital on Tap, said the tools are changing what small teams can handle. “A three-person firm can now do its own research, marketing, bookkeeping and customer service to a standard that used to need a department,” he said. “More than one in eight of the small businesses in our data are already buying AI tools, and the pace is accelerating, not levelling off.”
The analysis also pointed to duplication in SME AI spending. Brychcy said more than a quarter of AI buyers now pay for two or more providers, and the balance between providers can flip inside a single quarter. That puts pressure on SMEs to “cut the duplication and be honest about whether each tool is earning its place,” he said.
The research leaves unclear whether SMEs are equipped to appreciate the risks of AI tools, and few are likely to have dedicated IT staff across the details. Brychcy advised businesses to check a tool’s output, keep sensitive business and customer data out of tools they have not vetted, and not hand a model decisions that require human judgment.