How work needs shape the UK’s £1 billion AI market

The UK’s £1 billion consumer AI market also includes self-funded work tools

Some of the UK’s consumer AI spending comes from people buying tools partly for their jobs. They pay the bill themselves, but the work those subscriptions support may also benefit an employer. Economists studying the market see its large occupational differences as evidence that usefulness at work helps explain why some individuals pay.

The market has grown quickly enough to make that distinction commercially important. Economists at the University of Warwick and Imperial College London estimate that 5.1% of UK adults, or about 2.8 million people, paid for AI assistants in June 2026. They estimate that annual spending had reached £1 billion from almost nothing in early 2023, while subscriber numbers had roughly doubled annually. To understand that growth, product and marketing teams need to distinguish who subscribes, what they pay and what value they expect.

Personal payments define the study’s scope

Those questions are the focus of the economists’ analysis. They used anonymised bank and card transactions to measure individuals’ purchases of standalone AI products, products bought separately through their own subscriptions. Purchases through mobile app stores were included, so subscriptions bought through a phone’s app marketplace counted alongside other personal payments.

The researchers were Dr Nikhil Datta and Dr Peter John Lambert, both affiliated with the University of Warwick’s Department of Economics and the Centre for Economic Performance at the London School of Economics, and Dr Patrick Schneider of Imperial Business School. Their transaction records support estimates of paid participation and spending. They also show how purchases differ across occupations, demographic groups, locations and brands.

Because the records track personal purchases, the participation rate describes adults who pay for access themselves. Free access and licences funded directly by employers fall outside that measure. A large subscription market can therefore coexist with a relatively small share of adults personally subscribing, even as people gain access to AI through several payment arrangements.

Occupational differences point to work-related value

Within this personally funded market, the researchers found that subscription rates varied substantially by occupation. IT professionals had the highest reported rate, at about 20%. The other groups show how widely personal purchasing varies across jobs, including within professional and knowledge-based work.

The researchers connect these differences to the tasks people face at work. They interpret the pattern as a sign that workers have more reason to buy access when AI can help with those tasks. Datta explains: “People whose jobs involve tasks that AI can help with are more likely to pay for it themselves, so that suggests people are buying these tools partly for what they can do for them at work.”

His interpretation makes occupation useful for understanding demand while keeping the distinction between buying access and using it clear. A payment records a decision to subscribe; the occupational pattern suggests that usefulness at work contributes to some of those decisions. Product and marketing teams can use that pattern to identify groups with higher paid participation, then investigate the tasks those customers value.

But the grouped estimates need care when choosing a target audience. University lecturers and finance professionals share a reported range, as do school teachers, writers, artists and doctors; the figures do not give each occupation a separate rate within its group. Meanwhile, lower participation among electricians, nurses and social care workers describes their purchasing behaviour without assigning a reason to it. These differences support treating occupations as distinct customer groups rather than assuming a uniform professional audience.

Regional differences add another view of demand

Occupation is one way to divide the market; location provides another. The researchers found that London had the highest regional subscription rate in June 2026, at 7.9% of adults, compared with the national 5.1%. The North East had the lowest regional rate, at 3.8%. The analysis places London’s stronger participation alongside broader digital spending patterns and its concentration of professional and technical work.

The regional findings describe where personal purchasing is more common, while the occupational findings describe which workers are more likely to pay. Both can help teams choose where to investigate demand, but the regional comparison remains a separate descriptive finding. It shows how a national total contains markets with different levels of paid participation.

Spending share reflects plan prices and customer preferences

Differences within the paying audience also matter when comparing providers. The researchers found that ChatGPT remained the largest brand by subscription spending during the period studied, while Claude gained share rapidly. Between December 2025 and July 2026, Claude’s share rose from 7% to 29%, as ChatGPT’s fell from 74% to 53%.

These figures measure the money customers direct to each provider. Subscriber share measures the proportion of paying customers choosing a brand. Spending share can rise when a provider attracts more customers, sells more expensive subscriptions, or does both, so understanding the competitive shift requires attention to plan choice as well as customer numbers.

For Claude, the study links its gains to increased uptake of higher-priced plans, particularly among technical and professional workers. The researchers found stronger support for Claude among software and IT workers, engineers, finance professionals and academics. This connects the occupational and spending findings: usefulness at work may encourage a purchase, while the chosen provider and subscription tier determine how much it contributes to revenue.

The researchers found a different pattern for ChatGPT, which teachers, healthcare workers and human resources staff tended to favour. So an occupation’s likelihood of paying and its preferred provider are separate questions. A customer group may be commercially attractive because many people subscribe, because they choose more expensive plans, or because both patterns occur together.

Schneider describes how quickly the balance changed: “Like many early-stage markets, the picture can change very quickly. ChatGPT still leads at present, but its early dominance is shrinking. Claude captured almost a third of consumer subscription spending by July, up from just 7% seven months earlier,” he said. His comparison captures a sharp redistribution of subscription spending while ChatGPT retained the lead.

New subscribers are broadening the paying audience

While providers compete for spending, the mix of people entering the paid market is also changing. Datta says the earliest paying customers were disproportionately men and high earners, whereas women now account for nearly half of new subscribers. The incoming audience is broader, even as participation across the existing customer population remains uneven.

The gender figures show how that change can coexist with an overall gap. The researchers reported that men were 27% more likely than women to subscribe, while new subscribers were approaching an even gender split. The incoming group shows the direction of change; the overall comparison reflects everyone who has joined the paying population so far. A more balanced flow of new customers can gradually change that population.

Beyond gender, the researchers found strong differences by age and income. Adults under 30 were five times more likely to pay than those over 60, and the top fifth of earners were almost twice as likely to subscribe as the bottom fifth. These figures compare one group’s subscription rate with another’s. They help teams compare paid participation while keeping the analysis focused on personal purchasing.

Free users face a separate decision to pay

The differences among paying customers leave a further growth question: what will persuade people using free AI tools to buy a subscription? Lambert describes millions of people making recurring monthly payments and a market that developed in little more than three years. His rounded “one in twenty” description captures how small the personally paying share of adults remained despite that rapid growth.

For someone already using a free tool, buying a subscription requires a reason to value the paid offering. The size of that future market depends on how many users make this decision and which plans they choose. Lambert puts the question directly: “The big question is how many people currently using free AI tools will eventually decide they are worth paying for?”

In conclusion

The UK’s £1 billion consumer AI market is also, in part, a market for self-funded work tools. For business leaders, that makes personal subscriptions relevant beyond consumer demand. Employees may already be paying for capabilities they find useful at work, even where employers have not funded access.

The occupational differences offer a starting point for understanding that demand, rather than a ready-made purchasing plan. Higher subscription rates suggest where workers see value, but payment records do not establish productivity gains or show which tasks benefit. Leaders considering company-funded access need to connect employees’ purchasing decisions with evidence of usefulness, cost and appropriate data handling.

For AI providers, growth depends on more than attracting subscribers. Provider preferences vary across occupations, and higher-priced plans can shift spending share without an equivalent shift in customer numbers. Understanding which tasks justify payment, and which justify a more expensive plan, will matter as the paying audience broadens.

The next phase will test how far that willingness to pay extends. For employers and providers alike, the central question is what value makes AI worth funding, who receives that value and who should pay.