DeepMind's David Silver Raises Record $1.1B Seed for 'Ineffable Intelligence' — Betting RL Beats LLMs to Superintelligence
AlphaGo architect David Silver came out of stealth with a $1.1B seed at $5.1B valuation — Europe's largest ever — to build AI that learns from experience instead of human data.
David Silver, the DeepMind researcher behind AlphaGo, AlphaZero, and AlphaFold, broke stealth on April 27 with the largest seed round in European startup history: $1.1 billion at a $5.1 billion post-money valuation. His new London-based lab, Ineffable Intelligence, has no product, no revenue, and a contrarian thesis — that the path to superintelligence runs through reinforcement learning, not larger language models.
A Cap Table That Reads Like a Truce
The round was co-led by Sequoia and Lightspeed, with Nvidia, DST Global, Index Ventures, Google itself, and the U.K.'s Sovereign AI Fund all writing checks. It's an unusually broad coalition for a seed — the kind typically reserved for scaling rounds — and a signal that the most credible names in AI infrastructure are willing to place a billion-dollar option on a non-LLM future. Silver founded the company in November 2025 while still teaching at University College London, where he holds a professorship.
Why the "No Human Data" Pitch Lands
Silver's bet is that LLMs have hit diminishing returns because they are constrained by the data humans have already produced. His paper "The Era of Experience," co-authored at DeepMind, argued that the next leap requires agents that learn from their own trial-and-error — exactly what AlphaZero did to surpass human chess and Go without studying any human games. Ineffable will pursue that thesis at frontier scale, with compute commitments that suggest training runs measured in hundreds of thousands of GPUs rather than thousands.
What to Watch
A $1.1B seed is also a $1.1B liability if RL-from-experience plateaus the way LLMs are plateauing. Silver has the technical credibility to attempt the bet, but the track record of "post-LLM" labs is shorter than the marketing suggests. The first real signal will be benchmark performance from whatever Ineffable ships in 2026 — and whether Nvidia's check-in this round translates into preferential GPU allocation when supply is tight.
Sources
https://decrypt.co/365853/google-deepmind-veteran-raises-1-1-billion-ai-isnt-trained-human-data
How we report: This article cites primary sources, regulatory filings, and on-chain data where available. BlockAI News uses AI tools to assist with research and first-draft generation; every article is reviewed and edited by a human editor before publication. Read our full How We Report page, Editorial Policy, AI Use Policy, and Corrections Policy.
