David Silver, the creator of DeepMind’s AlphaGo and AlphaZero, left the company in January 2026 after a sabbatical to lead Ineffable Intelligence. This stealth London startup aims to develop “endlessly learning superintelligence” using reinforcement learning and world models.
Rumour has it, Ineffable is about to close Europe’s biggest-ever seed round: $1 billion at a $4 billion pre-money valuation. Sequoia is leading, with Nvidia, Google, and Microsoft circling.
Names like Jensen Huang (Nvidia), Patricia Huang (Sequoia), and Sinai Garousi are already in the mix.
So, who exactly is David Silver, and why is everyone watching his next move?
From AlphaGo architect to Ineffable CEO
Silver led DeepMind’s legendary AlphaGo win over Lee Sedol in 2016, pioneering deep RL, Monte Carlo Tree Search, and self-play to crack a game with more positions than atoms in the universe. Now, he’s the only named director at Ineffable, according to Companies House.
Silver is actively recruiting ex-DeepMind talent for agentic RL systems, but no hires have been announced yet. By contrast, peers like Mistral and Sutskever’s SSI (ex-OpenAI all-stars) tout stacked teams from day one, spotlighting legitimate execution risks for Ineffable’s moonshot ambitions.
He’s still a UCL professor, but now he’s busy building a dream team of ex-DeepMind researchers. This time, focusing on agentic RL systems instead of scaling LLMs.
Chasing the “era of experience,” beyond LLMs
Ineffable is all about breaking past the limits of today’s LLMs. Instead, the team is betting on agents that keep learning through world models: internal simulations that let them evolve by trial and error.
This method targets robotics and real-world uses. It’s similar to AlphaStar. UK government sources view it as a boost for London’s AI scene, especially as some talent leaves San Francisco.
Europe’s record-breaking raise: $1B seed at $4B valuation?
According to the FT, a $1 billion seed round is in the works, with Sequoia’s Patricia Huang leading after Silver’s pitch. Jensen Huang (Nvidia) has already met the team, and both Google and Microsoft are showing serious interest.
Even though the product isn’t ready yet, the excitement rivals that around Sutskever’s SSI. There’s no set closing date, but the timeline is aggressive.
A DeepMind legacy that keeps growing
Over 15 years at DeepMind (consulting since 2010, full-time since 2013), Silver published 16+ papers, including “Mastering Go,” picked up the 2018 ACM Prize (with Demis Hassabis and Yoshua Bengio), became a Royal Society Fellow in 2023, and took home the first-ever Mensa Prize in 2024.
His work in reinforcement learning has already optimised the UK power grid (saving over £1M), made Google’s data centres 30% more efficient, and even helped ESA probes.
After his sabbatical, one thing was clear: Silver is all about solving the hardest problems with small, focused teams.
A talent magnet, and a sovereignty play for Europe
Ineffable supports UK and EU efforts to keep talent, helped by post-Brexit visas. It aligns with the EU Data Union and the Mistral ecosystem, where reinforcement learning agents enable model sharing among the US, China, and the EU.
Reddit and AI forums are already buzzing: “AlphaGo 2.0 for the real world?” If this round lands, Silver could become Europe’s answer to Ilya Sutskever. This time, with a focus on robotics and science simulations.
Silver’s next chapter? A bold mission to build superintelligence. If Ineffable delivers, it could even outshine DeepMind’s game-changing legacy.
