OpenAI’s UK arm paid an average of £2.32 million per employee in 2025, according to company accounts reported by The Guardian. That figure sounds like a football wage because, for much of the London office, it effectively was: 110 staff shared £87 million in salaries and another £168 million in shares.
The denominator matters. This was not an average cash salary of £2.32 million. It was total compensation, with equity doing most of the inflation. Even so, the split is still rich by any adult standard. The average package was about double the previous year’s £1.1 million, The Guardian reported, and about 58 times median full-time UK earnings using Office for National Statistics data.
What changed was not just generosity. OpenAI UK Ltd also recorded turnover of £293 million in 2025, more than triple the prior year, and made a £20.3 million profit, according to the accounts. So the London operation was not merely a cost centre collecting researchers like rare stamps. It was generating revenue as well, with the office housing research, sales, marketing and government affairs functions.
There is a limit to what the average proves. The Guardian reported a wide gap between the highest and lowest earners, with the largest packages reserved for top AI researchers. An average across 110 people can hide a small number of compensation outliers large enough to make a banker blush and a junior employee feel statistically useful.
Still, the broad point survives the accounting footnotes. OpenAI was willing to spend heavily in London during 2025, when competition for elite AI talent had become absurdly expensive. Shares worth £168 million in one year suggest the company was paying not just for labour, but for allegiance. In a market where the scarce asset is not office space or cloud capacity but a handful of people who can materially improve frontier models, stock becomes less a perk than a retention device with a postcode.
The London office’s location also says something about the map of this race. OpenAI is expanding in King’s Cross, close to Google DeepMind and the universities that feed the machine-learning pipeline. For all the talk about artificial intelligence remaking work, the accounts describe something older and simpler: a bidding war for humans.
Sophia Borg
Harvey Jr.
Gabriel Fenech