Release 2026-07 · data to July 2026
One in 20 adults now pays for an AI assistant
We track who pays for ChatGPT, Claude and other AI tools in de-identified records from 14 million UK bank and card accounts. We publish a new release every month. Supported by the Centre for Competitive Advantage in the Global Economy (CAGE), funded by the Economic and Social Research Council (ESRC).
- 5.1% of adults paid for an AI assistant in June 2026, about 2.8 million people
- £1.07bn Annual consumer spending on AI, at the rate of July 2026
- 29% Claude's share of AI subscription spending in July 2026
- 5.9× Ages 20 to 29 against 60 and over: likelihood of having paid for AI
- 1.9× Top income band against the bottom: likelihood of having paid for AI
What the data show
Exhibit 1 of 7
5.1% of adults paid for an AI assistant in June 2026, up from 2.3% a year earlier
Show the table
| Month | Best estimate | 95% interval |
|---|---|---|
| Jul 2026 | 52.0 | 39.9–57.4 |
| Jun 2026 | 50.6 | 38.0–55.6 |
| May 2026 | 48.2 | 36.1–52.9 |
| Apr 2026 | 45.0 | 34.2–49.8 |
| Mar 2026 | 43.7 | 33.0–48.4 |
| Feb 2026 | 39.5 | 30.6–43.5 |
| Jan 2026 | 36.4 | 28.7–40.1 |
| Dec 2025 | 32.9 | 25.7–36.3 |
| Nov 2025 | 30.6 | 23.9–33.9 |
| Oct 2025 | 30.7 | 24.8–33.8 |
| Sep 2025 | 29.4 | 23.0–32.6 |
| Aug 2025 | 28.0 | 21.9–31.3 |
| Jul 2025 | 25.8 | 20.1–28.8 |
| Jun 2025 | 23.1 | 18.6–26.1 |
| May 2025 | 20.9 | 16.4–23.6 |
| Apr 2025 | 18.5 | 14.0–21.2 |
| Mar 2025 | 16.0 | 11.4–18.7 |
| Feb 2025 | 14.4 | 10.4–17.0 |
| Jan 2025 | 14.2 | 9.6–17.3 |
| Dec 2024 | 13.1 | 9.2–15.9 |
| Nov 2024 | 11.4 | 8.6–14.1 |
| Oct 2024 | 11.1 | 8.1–13.7 |
| Sep 2024 | 10.2 | 7.3–12.8 |
| Aug 2024 | 9.3 | 6.8–11.8 |
| Jul 2024 | 9.3 | 6.3–11.8 |
| Jun 2024 | 8.7 | 6.1–11.1 |
| May 2024 | 8.2 | 6.2–10.3 |
| Apr 2024 | 8.2 | 5.8–10.4 |
| Mar 2024 | 8.7 | 6.6–10.7 |
| Feb 2024 | 6.8 | 5.2–8.8 |
| Jan 2024 | 6.7 | 5.0–8.6 |
| Dec 2023 | 6.2 | 4.4–7.8 |
| Nov 2023 | 4.7 | 4.0–5.9 |
| Oct 2023 | 4.5 | 3.7–5.9 |
| Sep 2023 | 3.9 | 3.4–4.6 |
| Aug 2023 | 3.6 | 3.2–4.5 |
| Jul 2023 | 3.2 | 2.8–4.1 |
| Jun 2023 | 3.0 | 2.6–4.1 |
| May 2023 | 3.1 | 2.8–4.3 |
| Apr 2023 | 2.6 | 2.4–2.8 |
| Mar 2023 | 1.8 | 1.7–2.0 |
| Feb 2023 | 0.6 | 0.5–0.7 |
| Jan 2023 | 0.0 | 0.0–0.1 |
| Dec 2022 | 0.0 | 0.0–0.1 |
| Nov 2022 | 0.0 | 0.0–0.0 |
| Oct 2022 | 0.0 | 0.0–0.0 |
| Sep 2022 | 0.0 | 0.0–0.1 |
| Aug 2022 | 0.1 | 0.0–0.1 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Exhibit 2 of 7
Claude's share of AI subscription spending rose from 7% in December 2025 to 29% in July 2026
Show the table
| Month | ChatGPT | Claude | Gemini | Other |
|---|---|---|---|---|
| Jul 2026 | 52.5 | 29.4 | 5.1 | 13.0 |
| Jun 2026 | 52.8 | 27.6 | 5.3 | 14.2 |
| May 2026 | 56.8 | 22.7 | 5.1 | 15.4 |
| Apr 2026 | 58.8 | 21.3 | 5.4 | 14.5 |
| Mar 2026 | 65.1 | 15.9 | 5.2 | 13.8 |
| Feb 2026 | 70.0 | 10.3 | 5.7 | 14.0 |
| Jan 2026 | 72.0 | 8.1 | 5.9 | 14.0 |
| Dec 2025 | 73.5 | 7.4 | 5.8 | 13.3 |
| Nov 2025 | 74.0 | 6.9 | 5.4 | 13.7 |
| Oct 2025 | 74.1 | 7.2 | 5.4 | 13.2 |
| Sep 2025 | 75.8 | 6.5 | 5.1 | 12.6 |
| Aug 2025 | 76.8 | 5.8 | 5.0 | 12.4 |
| Jul 2025 | 76.0 | 6.1 | 4.7 | 13.2 |
| Jun 2025 | 76.4 | 5.7 | 4.1 | 13.8 |
| May 2025 | 77.5 | 5.1 | 3.8 | 13.6 |
| Apr 2025 | 76.2 | 5.5 | 3.7 | 14.5 |
| Mar 2025 | 74.5 | 5.6 | 3.9 | 16.0 |
| Feb 2025 | 73.4 | 5.7 | 4.5 | 16.4 |
| Jan 2025 | 73.5 | 6.1 | 4.5 | 16.0 |
| Dec 2024 | 71.8 | 5.7 | 4.1 | 18.4 |
| Nov 2024 | 72.1 | 5.4 | 4.4 | 18.1 |
| Oct 2024 | 70.6 | 5.4 | 4.6 | 19.3 |
| Sep 2024 | 71.1 | 4.6 | 4.4 | 19.9 |
| Aug 2024 | 71.1 | 4.5 | 4.3 | 20.2 |
| Jul 2024 | 71.3 | 3.8 | 3.3 | 21.6 |
| Jun 2024 | 71.1 | 2.7 | 2.4 | 23.8 |
| May 2024 | 72.3 | 2.4 | 2.1 | 23.2 |
| Apr 2024 | 71.9 | 2.9 | 0.7 | 24.6 |
| Mar 2024 | 72.9 | 4.0 | ||
| Feb 2024 | 75.8 | 1.8 | ||
| Jan 2024 | 75.3 | 2.0 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Exhibit 3 of 7
Adults aged 20 to 29 are 5.9 times as likely as adults aged 60 and over to have paid for AI, and 4.8 times after accounting for income, sex and region
Show the table
| Group | Ever paid for AI (%) | Adjusted (%) |
|---|---|---|
| 20 to 29 | 17.9 | 18.3 |
| 30 to 39 | 17.5 | 16.9 |
| 40 to 49 | 14.1 | 13.3 |
| 50 to 59 | 8.1 | 7.9 |
| 60 and over | 3.0 | 3.8 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Exhibit 4 of 7
In July 2026, men were 28% more likely than women to pay for an AI assistant
Show the table
| Month | Men | Women |
|---|---|---|
| Jul 2026 | 62.2 | 48.6 |
| Jun 2026 | 61.0 | 46.6 |
| May 2026 | 58.0 | 44.4 |
| Apr 2026 | 54.1 | 41.0 |
| Mar 2026 | 52.0 | 40.6 |
| Feb 2026 | 47.0 | 36.9 |
| Jan 2026 | 43.5 | 33.8 |
| Dec 2025 | 39.4 | 30.3 |
| Nov 2025 | 37.2 | 27.9 |
| Oct 2025 | 37.5 | 28.0 |
| Sep 2025 | 35.8 | 27.1 |
| Aug 2025 | 34.5 | 25.3 |
| Jul 2025 | 31.9 | 23.2 |
| Jun 2025 | 28.8 | 20.6 |
| May 2025 | 26.2 | 18.2 |
| Apr 2025 | 23.5 | 16.1 |
| Mar 2025 | 21.0 | 13.2 |
| Feb 2025 | 19.2 | 11.8 |
| Jan 2025 | 19.1 | 11.4 |
| Dec 2024 | 18.1 | 10.1 |
| Nov 2024 | 16.5 | 8.1 |
| Oct 2024 | 15.9 | 8.1 |
| Sep 2024 | 14.9 | 7.4 |
| Aug 2024 | 13.8 | 6.6 |
| Jul 2024 | 13.3 | 6.9 |
| Jun 2024 | 12.7 | 6.2 |
| May 2024 | 12.8 | 5.4 |
| Apr 2024 | 12.5 | 5.8 |
| Mar 2024 | 13.5 | 5.7 |
| Feb 2024 | 10.8 | 4.3 |
| Jan 2024 | 10.5 | 4.7 |
| Dec 2023 | 9.7 | 4.3 |
| Nov 2023 | 8.2 | 2.5 |
| Oct 2023 | 7.8 | 2.6 |
| Sep 2023 | 6.9 | 2.0 |
| Aug 2023 | 6.5 | 1.8 |
| Jul 2023 | 6.2 | 1.4 |
| Jun 2023 | 5.9 | 1.2 |
| May 2023 | 6.1 | 1.4 |
| Apr 2023 | 5.2 | 0.9 |
| Mar 2023 | 3.8 | 0.6 |
| Feb 2023 | 1.2 | 0.2 |
| Jan 2023 | 0.1 | |
| Dec 2022 | 0.1 | |
| Sep 2022 | 0.1 | |
| Aug 2022 | 0.1 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Exhibit 5 of 7
Adults with a monthly income over £4,000 are 88% more likely than those under £1,500 to have paid for AI
Show the table
| Group | Ever paid for AI (%) | Adjusted (%) |
|---|---|---|
| Under £1,500 | 10.0 | 11.1 |
| £1,500 to £2,000 | 11.6 | 10.6 |
| £2,000 to £2,500 | 14.9 | 13.4 |
| £2,500 to £3,000 | 15.6 | 14.4 |
| £3,000 to £4,000 | 17.7 | 16.8 |
| Over £4,000 | 18.8 | 18.8 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Exhibit 6 of 7
Paid AI adoption varies twelvefold across occupations
Show the table
| Occupation | Share paying for an AI assistant (%) |
|---|---|
| IT professionals | 20.3 |
| University & college lecturers | 16.0 |
| Finance & insurance professionals | 15.8 |
| Project & procurement managers | 13.8 |
| Marketing & PR professionals | 12.8 |
| Writers, artists & producers | 9.7 |
| Surveyors & town planners | 8.2 |
| Postal & telecoms workers | 7.7 |
| Chartered engineers | 7.5 |
| School teachers | 7.5 |
| Security officers | 7.0 |
| Medical doctors | 7.0 |
| Pharmacists | 6.8 |
| Psychotherapists & counsellors | 6.8 |
| Musicians | 6.7 |
| Human resources professionals | 6.5 |
| Fitness instructors | 6.1 |
| Actors & stage performers | 5.8 |
| Scientists & R&D engineers | 4.9 |
| Banking staff | 4.9 |
| Taxi & delivery drivers | 4.8 |
| Manufacturing & industrial staff | 4.7 |
| Allied health professionals | 3.8 |
| Natural & social scientists | 3.6 |
| Social workers | 3.4 |
| Accountants & auditors | 3.4 |
| Civil servants | 3.3 |
| Veterinarians | 3.1 |
| Dentists & dental nurses | 3.1 |
| Construction workers | 2.9 |
| Electricians | 2.9 |
| Firefighters | 2.5 |
| Nurses & midwives | 2.5 |
| Social care workers | 1.7 |
| All open-banking accounts | 3.6 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Exhibit 7 of 7
London leads the UK in paid AI adoption
Show the table
| Group | Paid AI users per 1,000 adults |
|---|---|
| London | 78.5 |
| South East | 53.5 |
| East of England | 50.0 |
| North West | 48.2 |
| South West | 45.6 |
| Northern Ireland | 45.5 |
| Scotland | 44.4 |
| West Midlands | 44.1 |
| Wales | 43.5 |
| East Midlands | 43.1 |
| Yorkshire and The Humber | 42.0 |
| North East | 37.6 |
| All adults | 50.6 |
Source: Datta, Lambert and Schneider (2026), data pack 2026-07.
Map boundaries: Source: Office for National Statistics licensed under the Open Government Licence v.3.0. Contains OS data © Crown copyright and database right 2025.
The paper
Who Pays for AI? documents how fast paid AI is spreading, how adoption differs by age, gender, earnings, occupation and place, and how consumers split their spending between ChatGPT, Claude and other tools.
How we find AI payments
Every subscription leaves a line on a bank statement. We match direct payments to a registry of more than 200 AI products using the statement text, the exact price and the product's launch date. Subscriptions billed through the Apple and Google app stores hide the product name, so we estimate them from recurring charges at AI prices. We then weight the accounts to represent UK adults.
| Date | Description | £ |
|---|---|---|
| 03 JUL | TESCO STORES 2041 | -23.45 |
| 03 JUL | OPENAI *CHATGPT SUBSCR | -20.00 |
| 04 JUL | TFL TRAVEL CH | -8.10 |
| 05 JUL | CLAUDE.AI SUBSCRIPTION | -18.00 |
| 06 JUL | NETFLIX.COM | -5.99 |
| 08 JUL | APPLE.COM/BILL | -19.99 |
| 09 JUL | PRET A MANGER | -6.75 |
| 11 JUL | APPLE.COM/BILL | -2.99 |
- Direct payment to an AI provider
- App store charge at an AI price, estimated
Illustration, not real records. Highlighted lines are direct payments for AI. Hatched lines are app store charges at an AI price, which we estimate rather than observe.
Get the monthly data pack
Download the national series and the breakdowns by age, gender, income and region as a spreadsheet and CSV files. The data pack is free. Enter your email address and, on your first download, tell us how you plan to use the data.