DeepMind Alumni Line Up VC Cash for Non-LLM AI Startups

Nando de Freitas is in talks to raise at least $100 million to launch Revolution Labs, a startup built around diffusion models as an alternative to the transformer architecture underpinning today's large language models, Bloomberg reports. Accel is in talks to lead the round.

De Freitas, who spent about a decade at DeepMind before a stint at Microsoft, is one of a widening cohort of Google DeepMind alumni raising serious money for post-LLM bets. David Silver's Ineffable Intelligence has raised roughly $1 billion. Jack Parker-Holder's Emulate is in discussions to raise $700 million. Veteran DeepMind scientist Thore Graepel, who left this summer, is seeking substantial funds for a reasoning-model startup called Metis Reasoning, which we covered separately this week. Bloomberg reports that a group of fifteen Google DeepMind employees and alumni met for a breakfast in central London to compare notes on raising money.

Google and DeepMind have now produced more than 40 founders of so-called 'neolabs,' per the reporting, more than OpenAI, Anthropic and Meta.

Janusz Marecki, a former DeepMind lead scientist now an AI partner at Ahren Innovation Capital, put the exit motive plainly. 'There's a top-down pressure to drop everything you're working on and focus on large language models. Then you just leave, there's no point. These are smart people, smart scientists,' he said. He also flagged the pricing. 'Valuations used to be 10 times revenue. Now it's 200 times, 300 times.'

Demis Hassabis, who in August handed day-to-day control of the lab to technology chief Koray Kavukcuoglu and moved to Alphabet's chief scientist role, pushed back on the framing. 'There's a lot of talent movement between all the leading labs and we win our fair share of the top talent. We have by far the biggest and broadest research bench of any of the labs out there,' he said.

The reporting does not name Revolution Labs' other backers beyond Accel, or say when de Freitas expects to close.