David Silver has never been one for small bets. He created AlphaGo, the system that beat the world champion at Go in 2016 and changed what the world thought AI could do. He led the reinforcement learning team at Google DeepMind for over a decade. Now he's left to start his own company, and investors are tripping over each other to give him money.

Ineffable Intelligence, Silver's London-based startup, has raised $1.1 billion in a seed round at a $5.1 billion valuation. It is the largest seed round in European history and among the largest ever globally. The round was led by Sequoia Capital and Lightspeed Venture Partners, with strategic participation from Nvidia, Google, and the UK Sovereign AI Fund.

The company has not released a product. It has not published a paper. Its stated mission is to "make first contact with superintelligence."

The great brain drain

Silver's departure is part of a pattern that should alarm every major AI lab. In the past 12 months, senior researchers have been leaving Google DeepMind, OpenAI, and Meta AI in waves to launch their own ventures. AMI Labs (founded by former DeepMind VP Koray Kavukcuoglu) raised $1 billion. Recursive Superintelligence Ltd (co-founded by Ilya Sutskever after leaving OpenAI) raised $1 billion. Periodic Labs pulled in undisclosed nine-figure funding.

The message from the talent market is clear: the best AI researchers in the world believe they can move faster outside the big labs than inside them. That is a fundamental shift. For years, Google, OpenAI, and Meta had a near-monopoly on top-tier AI talent because only they had the compute. But with Nvidia, sovereign funds, and top-tier VCs now willing to bankroll independent efforts with billion-dollar seed rounds, the moat is gone.

Why Silver is different

Most AI startup founders sell a vision of making existing models faster, cheaper, or more specialized. Silver is not playing that game. His research background is in reinforcement learning, the branch of AI where systems learn by doing rather than by reading text. AlphaGo, AlphaZero, and AlphaFold all came from this lineage. Silver's thesis appears to be that the path to superintelligence runs through RL, not through scaling language models.

That is a contrarian bet at a time when the industry is obsessed with transformer scaling. It is also the kind of bet that only someone with Silver's track record could raise a billion dollars on.

What it means

The venture math here is extraordinary. A $5.1 billion valuation on a seed round means investors believe Ineffable Intelligence will be worth tens of billions within a few years, before it has proven anything. The last time the market bet this aggressively on a pre-product AI company, it was called OpenAI.

Google participating in the round is particularly notable. Silver spent his entire career at DeepMind building the RL capabilities that Google has commercialized across its products. Now Google is paying a premium to maintain a stake in what its own former researcher believes he can build better independently. That is either shrewd hedging or an admission that the talent is more valuable than the institution.

First reported by CNBC. Additional reporting from Financial Times and Bloomberg.