OpenAI UK accounts show £2.3m staff cost per employee
The company’s 2025 accounts reveal a rapidly expanding London operation whose headline average is dominated by share-based compensation.
OpenAI UK reported £258m in total staff costs for 2025, equivalent to about £2.3m per employee, based on an average headcount of 110. The figure is an accounting average rather than a typical salary: £167.8m came from share-based compensation, while wages and salaries totalled £87.4m.
The figures are contained in full accounts for OpenAI UK Ltd covering the year to 31 December 2025. Companies House records show that the filing was submitted on 13 October, making the details newly available as the company’s British operation attracts renewed attention.
Total staff costs rose sharply from £60.8m in 2024, when the company reported an average of 55 employees. The 2025 accounts therefore show both a larger London workforce and a much higher cost attached to it. Dividing total staff costs by average headcount produces the £2.3m figure, but does not show how pay was distributed between junior staff, senior researchers and executives.
Most of the total was not paid in cash through ordinary payroll. Share-based compensation accounted for £167.8m, compared with £87.4m in wages and salaries. The accounts also list £1.4m in pension costs and £1.4m in other employee benefits. The equity figure includes the effect of a corporate recapitalisation, so it should not be read simply as a year’s worth of newly granted shares.
OpenAI UK’s research and development spending rose to £117.8m from £48.2m in 2024. The accounts describe the company as a service operation within the wider group, with revenue earned from its parent through a cost-plus arrangement rather than primarily from direct sales to British customers.
That arrangement helps explain the scale of the numbers. OpenAI UK reported turnover of £293m in 2025, up from £79.2m, and a pre-tax profit of £31.7m. The reported turnover includes the costs recharged to the parent company with an additional margin, so it should not be treated as evidence that the London office independently generated £293m from British customers.
The accounts offer a glimpse of how frontier AI companies compete for a limited pool of technical staff. A large equity component can make offers attractive when a company expects its value to rise, but it also makes annual comparisons volatile and dependent on accounting valuations. The average therefore says more about the company’s compensation model and the concentration of high-value roles than about the pay of an ordinary London employee.
For Britain’s technology sector, the filing points in two directions. It suggests that London can host highly paid AI research and engineering work, while also highlighting the gap between frontier-lab compensation and the wider labour market. Whether that spending produces lasting domestic benefits will depend on how many jobs, skills and businesses remain in Britain, not on the headline average alone.
OpenAI did not publish a breakdown of individual compensation in the filing. The available accounts establish the total costs, average headcount and business structure, but they cannot show what the typical employee earned or how much of the share-based compensation will ultimately be realised.
One story, several perspectives
What is established
- OpenAI UK reported £258m in total staff costs for 2025 and an average workforce of 110 people.
- £167.8m of the total was share-based compensation and £87.4m was wages and salaries.
- OpenAI UK’s research and development spending rose to £117.8m.
- The company’s accounts describe a cost-plus arrangement with its parent company.
- The accounts do not show what the typical individual employee earned.
Left
Arguments The figures show why policymakers should scrutinise concentration of wealth and ensure that AI investment creates broad benefits through training, taxation and stronger worker protections.
Values The emphasis is on redistribution, fair access to opportunity and preventing a small group of highly paid specialists from capturing most of the gains from automation.
Consequences Supporters would accept some higher costs or regulation if it produced wider skills investment and reduced inequality, even if companies became less aggressive in recruiting.
Centre
Arguments Britain should welcome high-value research jobs while requiring transparency about public support, tax treatment and the real domestic economic contribution of multinational AI companies.
Values The focus is on evidence, proportional regulation and balancing competitiveness with accountability.
Consequences This could preserve the UK’s attractiveness to research firms while giving government better information about whether headline investment translates into lasting jobs and productivity.
Right
Arguments Exceptional compensation is a market signal that Britain should make it easier for high-growth technology companies to recruit, invest and reward scarce talent without excessive intervention.
Values The emphasis is on enterprise, property rights and international competitiveness rather than equalising outcomes.
Consequences A lighter-touch approach could attract more AI investment and tax revenue, but it may also widen pay disparities and leave the UK dependent on decisions made by overseas parent companies.
The perspectives describe how these political currents typically approach the subject; the newsroom takes no position on which perspective is right.
Fact-check Approved · Priya Shah — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The filing date is confirmed directly by Companies House, while the detailed staff-cost figures and accounting explanation are supported by reporting based on the filed accounts. The article explicitly distinguishes total staff costs from individual pay and avoids claiming that the average represents a typical employee.
- confirmed OpenAI UK filed full accounts for the year ended 31 December 2025 on 13 October 2026. — The Companies House filing history lists the filing date and accounting period. source
- confirmed OpenAI UK reported £258m in total staff costs and an average of 110 employees in 2025. — The figures are reported from the filed accounts by Bizztor. source
- confirmed £167.8m of the staff-cost total was share-based compensation and £87.4m was wages and salaries. — The detailed cost breakdown is reproduced from the accounts by Bizztor. source
- confirmed OpenAI UK’s research and development spending rose to £117.8m from £48.2m. — The comparison appears in the published account summary. source
- confirmed OpenAI UK reported 2025 turnover of £293m and pre-tax profit of £31.7m. — Bizztor reports these figures from the filed accounts and explains the cost-plus structure. source
- confirmed The £2.3m figure is an average staff-cost calculation, not a typical salary. — It follows arithmetically from £258m divided by average headcount of 110, and the source cautions that the average does not show distribution. source
Editor's note
Companies House confirms the new filing date and the reporting period. The detailed figures are reported from the filed accounts by Bizztor; the average is a staff-cost calculation, not evidence that every employee received £2.3m.Sources
- OPENAI UK LTD filing history — Companies House
- OpenAI Paid 110 UK Staff £258m in 2025, Mostly in Share Options — Bizztor
More on this in other media
- iNews — „openai artificial intelligence”
- BBC News — „openai artificial intelligence”
- The Guardian — „openai artificial intelligence”