OhAnthropic, one of the world’s largest AI companies, is recruiting machine learning research engineers in London, with salaries up to £630,000 per year. Seeing this on the American Society website certainly made me question my degree in modern languages.
Anthropic announced last week that it was moving to a new London office that can accommodate 800 people, four times the number it currently has in the British capital.
Similarly, its rival, OpenAI, is opening a permanent space in London for more than 500 people, more than double its current workforce, and has pledged to make London, its largest research center outside the United States.
These two will be based in King’s Cross (along with Google DeepMind) in what has become the technological heart of the city, aka the Knowledge Quarter.
The salary range offered for this particular Anthropic job is from £260,000 to £630,000 and includes a bonus but does not cover stock options – always a vital part of tech pay. Including them, the total package would be worth a lot more.
This role is for someone who can build and train the brain of generative AI. According to the job specification, strong candidates should be familiar with the architecture of large language models and reinforcement learning (in which an AI agent learns to make better decisions over time by receiving rewards or penalties in response to its actions).
There aren’t many people who can do that. And suddenly, everyone wants them. London, whose AI talent once revolved around one dominant player, Google DeepMind, is becoming a very crowded battleground.
A £630,000 salary could well be just the tip of the iceberg as AI companies turn their attention to the sports team economy. Meta, OpenAI, Google and xAI would all be spend millions for superstar AI talent. Google DeepMind apparently offers some top researchers compensation packages of up to $20 million per year.
THE expansion of leading AI companies in London is a big reward for the city, which remains the undisputed “AI capital” of Europe, with around 1,700 venture capital-backed AI start-ups worth around $125 billion based here and $7.1 billion raised in 2025, according to a March 2026 Prosus/Dealroom report.
This is testament to the incredible technological expertise that the UK is celebrated for and will attract even more as these two cutting-edge AI labs bring scientists from around the world to Britain.
The flip side is that competition for a small pool of elite talent at all levels is intensifying. Tech startup founders report that recruiting is becoming increasingly difficult. To tempt new entrants, they have to offer more and more money, and that often means substantial dilution of a founder’s ownership to offer them shares.
“The salaries are brutal,” said one startup founder, who complained that corporate America has been sucking up all the best talent: “I have to raise funding early just to hire a senior person. When you have to hire and you know OpenAI is next door, you have to be prepared to offer an insane package.”
A government study on AI skills released in January shows that inflated wages risk putting skilled hires out of reach for small businesses and entrenching the advantage for the better-funded ones. Researchers also like having the time and space to write academic papers, which does not necessarily meet the needs of the business and is a luxury that is difficult for start-ups to provide.
London is now a three-way battleground for talent, with local startups competing against U.S. companies expanding into Europe as well as large AI labs recruiting aggressively, according to Sandra Schwarzer, vice president of talent at Index Ventures. The UK is now the default location for US companies entering Europe, she says, moving away from Dublin. “Compensation converges with that of the United States for corporate account executives and research talent.” So how can small businesses compete and, most importantly, keep great employees?
One of the most imaginative initiatives comes from Jeremy Fraenkel, managing director of Fundamental, which bases its European team in Barcelona. His theory is that the city is a pretty nice place to live, but with very little AI competition.
Once settled, he hopes, people will remain loyal to the Spanish way of life and there will be less risk of them being poached by their rivals. Fundamental just came out of stealth mode with a $255 million Series A funding pot, so we still have a few years to go before we know if the tapas offensive has paid off.
Another UK AI company, Doubleword, is expanding its network in a different way by employing people other than typical AI recruits and training them.
“Many of our best engineers didn’t study AI in college,” says Meryem Arik, CEO and co-founder. “There are a lot of people from hard sciences like physics who are really, really good at math who are becoming AI researchers. We’re looking for that talent that we can convert.” The AI discipline is evolving so quickly, she explained, that the tactic of on-the-job training has worked well.
Synthesia, the AI video company, announced plans to increase its global workforce by 70%, investing more than $25 million in its new and existing offices in the United States and Europe this year. Yet the British AI unicorn says recruiting in the traditional way would hold back growth.
One of its recruitment tactics is to offer flexibility, relaxing office attendance requirements for some roles, so it can find people from outside London and even the UK.
Of course, it’s not just about money. A few lucky founders take advantage of their start-up status to hire engineers straight out of university. The promise of autonomy and fairness in an early-stage company may prove compelling as AI expands.
Even though it’s difficult to hire, London has at least one advantage over AI hotspot San Francisco: retention. People outside of Silicon Valley stay more loyal to their employer and don’t move on to the next job every few months.
Yet there is no doubt that this strange divided economy in London is becoming more and more pronounced. Headline-grabbing salaries in border labs, but a tougher environment for start-ups.