AI giants flock to London for talent, but space runs short

TL;DR:

• US frontier-AI firms including Anthropic, OpenAI and Cursor are scaling up in London, drawn by a talent base rooted in DeepMind and the city’s universities.
• Recruiters call London the deepest pool of frontier-AI talent outside the US, but warn well-funded American firms are squeezing local startups on pay.
• A projected 10.4m sq ft shortage of prime office space to 2030, plus power and compute limits, could throttle the expansion.

The wave of US AI firms enlarging their London bases — which Resultsense covered as record office take-up and through Cursor’s choice of London — has a clear driver and a growing constraint. The draw is talent; the catch is whether the city can house and power the people arriving to use it.

A talent base a decade in the making

“It’s all about talent,” British Land’s Mike Wiseman told CNBC, pointing to an ecosystem built over many years. Recruiters trace much of it to DeepMind, founded in London in 2010 and bought by Google in 2014, which seeded a generation of researchers and engineers. Anthropic flagged that “exceptional pool” when it took space for 800 staff — roughly quadrupling its local headcount — in the Knowledge Quarter it now shares with OpenAI, Google DeepMind, Meta, Synthesia and Wayve.

The pull is reshaping the local market. Heidrick & Struggles partner Frederic Groussolles called London the deepest frontier-AI talent pool outside the US, helped by its proximity to venture and growth capital. But search firm Erevena warns the cash-and-equity packages on offer from American giants are making it harder for home-grown startups to hire — a quieter cost of the boom.

The binding limit may be physical. British Land projects a 10.4m sq ft shortfall in new or refurbished prime office space to 2030, as AI firms compete with finance for the same buildings. LocalGlobe’s Ziv Reichert put it bluntly: “Talent brought the labs to London, but keeping them here will depend on whether the UK builds the infrastructure around them” — power, housing, transport and compute included.

Looking forward

The talent magnet is working; the question is supply. If office, energy and compute capacity lag, the same firms now planting flags could find expansion capped — turning an infrastructure problem into a competitiveness one for the UK’s AI ambitions.