David Silver’s new lab lands a $1.1B seed to chase ‘superlearners’

Investors just put another billion-dollar marker on the part of AI research that wants to move past training on human text. On April 27, TechCrunch reported that Ineffable Intelligence, the new lab founded by former DeepMind researcher David Silver, raised $1.1 billion at a valuation of $5.1 billion. That is an enormous seed round even by 2026 standards, and the money is not going into another general-purpose assistant wrapper. Silver is betting on reinforcement learning and a “superlearner” that discovers knowledge from experience instead of relying on human data. The source is TechCrunch’s report, DeepMind’s David Silver just raised $1.1B to build an AI that learns without human data.

The pitch stands out because it rejects the dominant recipe behind today’s frontier labs. Large language models improve by absorbing vast amounts of human-written material and then layering post-training on top. Silver’s background points in a different direction. At DeepMind, he helped build systems such as AlphaZero that learned high-level play from experience rather than from human game records. Ineffable is trying to apply that instinct far beyond games, which is why the company calls its target a “superlearner” instead of just a bigger model.

The investor list explains why the round cleared so fast. According to TechCrunch, Sequoia Capital and Lightspeed Venture Partners led the financing, with participation from Index Ventures, Google, Nvidia, the British Business Bank and the U.K.’s Sovereign AI fund. That mix matters. This is not only venture money chasing hype; it is also infrastructure capital and industrial policy money backing a specific research thesis. London benefits too. The round adds another heavyweight to a local network already shaped by DeepMind alumni and helps keep the U.K. in the conversation when frontier AI talent spins out into new labs.

The harder question is whether the science can justify the valuation. Reinforcement learning has produced spectacular wins in constrained domains, but general-purpose intelligence outside those environments remains brutally hard. That is exactly why this fundraise matters. Markets are not just paying for current revenue anymore. They are prepaying for alternative paths beyond the current LLM paradigm. If Ineffable succeeds, it could reopen the architecture debate at the top of the AI stack. If it fails, it will still tell the market something useful: how much capital the industry is willing to burn in search of the next leap.