A former top researcher at Google's DeepMind has launched one of the most expensive blank pages in AI history. David Silver, the University College London professor who led DeepMind's reinforcement learning team for more than a decade, announced Monday that his months-old startup Ineffable Intelligence has closed a $1.1 billion seed round at a $5.1 billion valuation. The company says it is the largest seed financing ever raised in Europe.
The round was co-led by Sequoia Capital and Lightspeed Venture Partners, with participation from Nvidia, DST Global, Index Ventures, Google, the British Business Bank and the UK's Sovereign AI Fund. The investor roster reads like a deliberate transatlantic hedge — American capital paired with British state backing — as London tries to anchor a domestic frontier lab against the gravitational pull of San Francisco.
A reinforcement-learning bet against the LLM consensus
Silver's pitch is a direct challenge to the language-model orthodoxy that has defined the last three years of AI scaling. Rather than train on human-generated text, Ineffable plans to build what it calls a "superlearner" that acquires skills through reinforcement learning and trial-and-error experience. It is the same philosophical bet that produced AlphaZero, the system Silver helped build at DeepMind that mastered chess, shogi and Go without human game records.
"Our mission is to make first contact with superintelligence," Silver said in a statement accompanying the announcement. The company's own framing is even more grandiose: its website claims that explaining all intelligence "would represent a scientific breakthrough of comparable magnitude to Darwin."
That thesis is contrarian for a reason. Reinforcement-learning agents are notoriously sample-inefficient and difficult to steer in open-ended environments, which is part of why frontier labs pivoted to language pre-training in the first place. Silver is wagering that with enough compute and the right learning architecture, agents that generate their own data can leapfrog systems trained on the finite supply of human text.
Europe's biggest seed, and a UK sovereignty signal
The deal is also a political artifact. The British Business Bank and the UK's Sovereign AI Fund putting capital alongside Sequoia and Nvidia is the clearest signal yet that Westminster wants a homegrown lab capable of competing with OpenAI, Anthropic and DeepMind itself. Several former DeepMind staffers are reportedly joining the executive team, suggesting Silver is building from the same talent pool that produced AlphaGo and AlphaFold.
Why it matters
A $1.1 billion seed round priced at $5.1 billion is, in any normal market, absurd — and in 2026's AI funding environment, almost routine. What makes Ineffable notable is not the dollar figure but the technical bet underneath it. If Silver's superlearner thesis holds, the next frontier model may not look anything like GPT-5 or Claude. If it doesn't, $1.1 billion will buy one of the most expensive negative results in research history. Either outcome will reshape how investors price the gap between language-model scaling and whatever comes next.
