FT: the UK risks wasting its DeepMind dividend as alumni start-ups raise record rounds

What's the argument? The UK cannot afford to waste its DeepMind dividend, FT innovation editor John Thornhill argues in an opinion column published on 8 October. A shortage of domestic capital and the lure of the US threaten the benefits of Britain's biggest AI research success, and he says the next 12 to 18 months will be critical.

The alumni wave: Google DeepMind, founded in London in 2010 and acquired by Google in 2014, is seeing departures as Google shifts its AI focus to California; Zeki data cited in the column show more researchers have left DeepMind for Anthropic and OpenAI than the other way round. Its alumni are now building their own labs: Isomorphic Labs raised $2.1bn in May, David Silver's Ineffable Intelligence raised a record seed round of over $1bn, and Tim Rocktäschel co-founded Recursive Superintelligence, which has raised more than $500mn. Emulate, Revolution Labs and Metis Reasoning are other alumni ventures.

The capital question: Thornhill points to British Business Bank data showing UK VC pooled returns for 2020–24 outstripping US peers on some measures, yet domestic capital is "only trickling in". AI minister Kanishka Narayan said the government is working "extremely rapidly" on reforming non-compete clauses, while Phoenix Court's Saul Klein highlighted an expanded British Business Bank, pension consolidation and 17 fund managers pledging to invest in domestic private markets.

What could go wrong? The column warns that founders could still decamp to the US, with taxes and energy costs among the pressures.

Read more: Financial Times