Featured Article : OpenAI’s UK Pay Packages Average £2.3 Million
OpenAI’s UK employees accounted for an average of £2.3 million each in wages and share-based compensation in 2025, showing how fiercely AI companies are competing for expertise while raising questions about what those packages include and whether the spending will pay off.
What Do The Accounts Show?
Accounts filed at Companies House on 13 October show that OpenAI UK recorded £87.4 million in wages and salaries and £167.8 million in share-based compensation across an average workforce of 110. Together, that works out at approximately £2.32 million per employee.
Including pensions and other benefits, total staff costs reached £258 million, compared with £60.8 million in 2024. Although average employment doubled from 55 to 110, spending increased more than fourfold, showing that the rising bill was about more than simply hiring additional people.
However, the accounts don’t show how that money was distributed. A small number of exceptionally large packages could pull the average upwards, so the headline doesn’t really show what most employees earned.
Is That Really An Annual Salary?
In fact, most of the £2.3 million actually comes from share-based compensation. Wages and salaries alone averaged approximately £794,000 per person, although that accounting category doesn’t necessarily mean contractual basic salary.
Share awards give employees a financial interest in the company, but their recorded cost isn’t necessarily money paid into someone’s bank account that year. For example, employees may need to remain with the business for a specified period before becoming entitled to their awards, while selling them depends on the arrangements available.
OpenAI’s accounts also show that share awards come with conditions and that some were converted into a different type of award when the company restructured in 2025. The £2.3 million average includes the value recorded for those awards, so it doesn’t really mean employees each received that amount in cash.
Why Is AI Expertise So Expensive?
It seems AI companies are now willing to offer substantial packages because a relatively small team can influence products used by millions of people. For example, a researcher who improves a model’s accuracy or an engineer who reduces the cost of running it could create considerable commercial value.
That potential has drawn several well-funded employers into competition for the same specialists. For example, OpenAI must now compete with Google DeepMind, Anthropic, Meta and others, and there was some particularly aggressive recruitment of leading researchers during 2025.
Substantial share awards are one way to help companies compete while giving employees a reason to stay and benefit from future growth. However, those rewards depend partly on expectations about the business. Paying heavily for expertise may improve its prospects, but does not guarantee that the resulting products will earn enough to justify the investment.
Why Does London Matter?
London offers access to experienced researchers, universities and an established technology sector, making it an important location for companies building advanced AI. It also provides a base for selling services and working with government, so recruitment extends beyond technical roles.
OpenAI’s planned permanent office at Regent Quarter in King’s Cross reflects that ambition. Expected to open in 2027, it has capacity for 544 people. The company said in April that it already employed around 200 people in London, a later headcount than the 110-person annual average used in its 2025 accounts.
Expansion brings skilled jobs and research activity to the UK, but it could also make recruitment harder for smaller employers. Universities and businesses may struggle to retain specialists when competing firms can offer rewards on this scale.
Does The UK Business Make Money?
The UK subsidiary’s growth is quite visible in its financial results, with turnover rising from £79.2 million to £293 million, and profit reaching £31.7 million before tax and £20.3 million afterwards.
Those figures don’t, however, show how much British customers spent on ChatGPT. The accounts explain that OpenAI UK charges its parent company for services through a “cost-plus intercompany arrangement”, meaning its charges reflect costs plus an agreed margin.
As the UK operation expands and spends more, the revenue it receives through that arrangement can increase too. Its results therefore show the growing scale of local activity, but cannot establish whether OpenAI’s worldwide business is profitable or whether its investment in staff will ultimately pay off.
What Does This Mean For Your Business?
For businesses recruiting technical staff, the practical lesson is to identify the skills they actually need. Most organisations adopting AI don’t need researchers building underlying models, but they may benefit more from people who understand their operations, can connect existing tools and know how to assess the results.
That also gives smaller employers a more realistic approach to recruitment. For example, rather than trying to match exceptional pay packages, they can offer clear responsibilities, development opportunities and manageable workloads. Building AI knowledge within an existing team may also reduce dependence on one expensive specialist.
For customers buying AI services, the size of a supplier’s wage bill is less important than the quality and cost of its products, and businesses should assess performance, support and pricing, while keeping a practical way to move elsewhere. OpenAI’s spending may demonstrate confidence in AI’s potential, but customers still need evidence that the services deliver value in everyday use.
