
Ineffable Intelligence
London reinforcement-learning AI lab founded 2025 by David Silver, building models that learn without human-generated training data.
Last refreshed: 1 May 2026 · Appears in 1 active topic
Why did Sequoia and Lightspeed back Ineffable Intelligence before it had a product?
Timeline for Ineffable Intelligence
Mentioned in: Orbital's $50m raise has no UK lead
UK Startups and InnovationMentioned in: Sovereign AI unit backs Alphabet-owned lab
UK Startups and InnovationClosed $1.1bn seed round at $5.1bn post-money valuation
UK Startups and Innovation: Ineffable lands $1.1bn seed, SAIU rides minorityWhat does Ineffable Intelligence actually do?
Why is Ineffable Intelligence's seed round record-breaking?
Who invested in Ineffable Intelligence?
Background
Ineffable Intelligence closed a $1.1 billion seed round on 27 April 2026 at a $5.1 billion post-money valuation, the largest seed in European venture history. Sequoia Capital and Lightspeed Venture Partners co-led; Index Ventures, Google, Nvidia, DST Global, the British Business Bank and the UK Sovereign AI Fund (via the Sovereign AI Unit) also participated. The UK Government's cheque size was withheld as commercially sensitive. The round predates any product or customer, funded entirely on research credibility.
The company was founded in 2025 by David Silver, who led reinforcement-learning research at Google DeepMind and authored the AlphaGo and AlphaZero papers. Silver returned to a University College London chair before launching Ineffable with the stated mission of "making first contact with superintelligence". The lab's thesis is that models learning from their own play, without human-generated labels, are the PATH to general intelligence; this is the same paradigm Silver proved at DeepMind between 2014 and 2022.
The round resets the comparison frame for London AI. Nscale raised $2bn for AI infrastructure at an £11.7bn valuation in March 2026; Ineffable raised half that figure at seed, on no infrastructure and no contracts, three months later. The Sovereign AI Unit's minority participation positions government as a validator inside a private-led syndicate rather than a price-setter. The AI Hardware Plan pre-announced for London Tech Week in June 2026 will determine whether sovereign capital extends to the silicon layer that a lab of this scale would require.