IMPLICATOR
Anthropic said Thursday it is taking London office space for up to 800 people, according to CNBC. Three days earlier, OpenAI said it had secured its first permanent London office, a 544-seat site due to open in 2027, according to Reuters coverage carried by RTE. Two leases, one message: frontier AI still needs a door badge.
Anthropic already has more than 200 people in the capital. OpenAI says London will become its largest research hub outside the U.S. That is an odd fact for companies selling automation. Rooms still matter.
The timing makes London look less like a satellite office and more like disputed ground. Anthropic chose the Knowledge Quarter, near Google DeepMind, Meta, Synthesia and Wayve. OpenAI picked Regent Quarter in King's Cross, spanning Jahn Court and the Brassworks Building. Different leases. Same neighborhood logic.
That is the tell.
Key Takeaways
- Anthropic is adding London office space for up to 800 people in the Knowledge Quarter.
- OpenAI secured a 544-seat permanent London office due to open in 2027.
- The office race now spans Europe and Asia, with India and Sydney next on the map.
- The U.K. can attract AI talent more easily than it can host power-hungry AI data centers.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
London is becoming a talent checkpoint
Anthropic framed the move as room to grow into a city that already matters to its research and commercial work. Pip White, the company's head of EMEA north, said the U.K. combines enterprises that understand AI safety with a deep AI talent pool. OpenAI's London site lead, Phoebe Thacker, used similar language, saying the U.K. has "an incredible depth of talent" and a strong record in AI.
Both companies sound confident. Both also sound defensive.
London gives them proximity to DeepMind alumni, U.K. policymakers, banks, universities and a regulatory class that wants to be courted. It also lets each lab tell customers it has people nearby, not just a sales deck and a data-processing agreement. You can sell frontier AI from San Francisco. But when bank executives, regulators and government departments are anxious, a local office changes the room.
For Anthropic, that matters because its Mythos model and Project Glasswing have pushed cybersecurity anxiety into boardrooms. For OpenAI, it matters because the company paused its Stargate U.K. data center plan after citing energy costs and regulatory uncertainty. One company is expanding after a Pentagon feud. The other is adding people after pulling back on infrastructure.
The overseas map is getting crowded
The London leases sit inside a wider office race. OpenAI's first stops were London, Dublin and Tokyo. After that came Paris, Brussels, Singapore, Munich and Seoul, plus a fast-building India plan.
Anthropic's list now sprawls across Europe and Asia-Pacific. London and Dublin are already old news for the company. After that, the trail jumps by region and by sales need.
The non-U.S. office map looks like this:
Follow the AI office race
Strategic AI news from San Francisco. No hype, no "AI will change everything" throat clearing. Just what moved, who won, and why it matters. Daily at 6am PST.
No spam. Unsubscribe anytime.
The table needs one caveat. Companies use office language loosely. OpenAI calls New Delhi an "existing presence" and says Mumbai and Bengaluru are planned for later in 2026, not already open. Anthropic said in November that Paris and Munich were plans, while also saying the company then had offices in 12 cities. For readers, the useful split is opened, announced and planned.
Offices solve what data centers do not
OpenAI's London move landed days after the company paused Stargate U.K., a data center project tied to Nvidia and Nscale. The Next Web reported that the pause reflected high British industrial electricity prices and unsettled AI copyright rules. The office lease does not fix either problem.
It does something else. It preserves OpenAI's claim on British AI talent and policy influence while the infrastructure math remains ugly. Hiring researchers, policy staff and enterprise teams in King's Cross costs far less than committing to thousands of Nvidia GPUs in a power-constrained market. In a country trying to sell itself as an AI hub, that is a careful compromise.
Anthropic has the cleaner story this week. Its office expansion follows a reported U.K. campaign to lure the company after its clash with the U.S. government over military use of Claude. It also follows Anthropic's European expansion, where the company said EMEA run-rate revenue had grown more than ninefold over the prior year and large business accounts in the region had grown more than tenfold.
Those numbers make an 800-person London space less surprising. Enterprise AI is walking into local procurement rooms now. Nervous ones.
The next race is presence, not press releases
OpenAI and Anthropic are not opening offices for symbolism alone. The practical target is closer access to engineers, enterprise buyers and governments that can make their products easier or harder to sell.
Anthropic said Sydney would become its fourth Asia-Pacific office, alongside Tokyo, Bengaluru and Seoul. OpenAI said India had more than 100 million weekly ChatGPT users in February and laid out plans for Mumbai and Bengaluru offices alongside its New Delhi presence. These moves follow demand. They also shape it.
The absence matters too. A company can announce a country strategy without owning much physical capacity there. It can sign a data-center memorandum without breaking ground. It can call a local presence an office, then hire slowly. The map above is therefore a hiring ledger, not a victory chart.
London now has the loudest entry. OpenAI has the permanent lease. Anthropic has the larger stated capacity. What neither has yet is proof that more desks in King's Cross and the Knowledge Quarter can turn British AI ambition into durable power.
For now, the proof is a floor plan.
Frequently Asked Questions
What did Anthropic announce in London?
Anthropic said it is expanding in London with office space for up to 800 people in the Knowledge Quarter. The company already has more than 200 people in the capital and describes London as one of its most important hubs outside the U.S.
What did OpenAI announce in London?
OpenAI secured its first permanent London office, an 88,500-square-foot Regent Quarter site with capacity for 544 team members. The office is expected to open in 2027 and supports OpenAI's plan to make London its largest research hub outside the U.S.
Why does London matter to both companies?
London puts OpenAI and Anthropic near DeepMind alumni, banks, policymakers, universities and enterprise buyers. That combination matters for hiring, regulation and customer trust, especially as banks and government agencies test frontier AI systems.
Where else has Anthropic opened offices outside the U.S.?
Anthropic has cited offices or announced offices in London, Dublin, Zurich, Tokyo, Bengaluru, Seoul, Paris, Munich and Sydney. Some source language describes announced or planned offices rather than already-open locations.
Where else has OpenAI opened offices outside the U.S.?
OpenAI has announced or opened offices in London, Dublin, Tokyo, Paris, Brussels, Singapore, Munich and Seoul. It also cites an existing New Delhi presence and plans offices in Mumbai and Bengaluru later in 2026.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Marcus Schuler
Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: [email protected]
