Ineffable Intelligence raises $1.1B

A brand-new AI lab just raised the kind of money most startups never see at any stage, let alone seed. That alone would be news. But the bigger thing is what the money is for: not another chatbot, not another wrapper, and not even the usual “bigger model, more data” playbook. David Silver — one of the key researchers behind AlphaGo and AlphaZero — says his new company, Ineffable Intelligence, wants to build a “super learner” that gets smarter by acting in environments, running experiments, and learning from experience. That is a very specific bet — and investors just put $1.1 billion behind it. (cnbc.com) ### Why is David Silver a big deal? Silver is not just “ex-DeepMind” in the loose résumé sense. He was central to the reinforcement-learning work that made AlphaGo, AlphaZero, and later systems like AlphaProof feel like a different branch of AI progress from large language models. His research reputation matters here because t(cnbc.com)c operating history. (cnbc.com) ### What exactly happened? On April 27, 2026, Ineffable Intelligence came out of stealth and announced a $1.1 billion seed financing at a $5.1 billion post-money valuation. Sequoia Capital and Lightspeed Venture Partners led the round. Other backers named across coverage include Nvidia, Google, Index Ventures, and UK-linked ca(cnbc.com)thesis. (cooley.com) ### What is a “super learner”? Basically, Silver wants AI that does not depend mainly on human-generated text and labels. The idea is to build systems that can learn by interacting with simulated environments, using self-play, search, and reinforcement learning to discover useful strategies on their own. If lang(cooley.com)ng things and seeing what works. (cnbc.com) ### Why move away from human data? Because human data may be the bottleneck. The current generation of AI got huge gains from scraping text, code, images, and feedback from people. But that pool is finite, messy, and already heavily mined. Silver’s pitch is that truly general systems may need to generate their own training sig(cnbc.com)a simulator can teach more than another pass over the internet. (cnbc.com) ### Why are investors so excited? The short version is that frontier AI investors now believe elite research talent is itself an asset class. If a small group of researchers can open a new scaling path, the payoff could be enormous. That helps explain why a company only months old could command a $5.1 billion valuation and wha(cnbc.com)rce technical direction, not near-term business fundamentals. (cnbc.com) ### What makes this different from the LLM race? Large language models mostly learn from static corpora and then get tuned with feedback. Silver’s line of work is stronger when the system can act, get rewards, and improve through repeated interaction. That worked spectacularly in games because games have clean rules and clear (cnbc.com)simulated environments can be built at enough scale to train systems that transfer beyond toy worlds. (cnbc.com) ### Why does the UK angle matter? Ineffable is London-based, and the financing has been framed as a major European milestone, with UK-backed participation also showing up in coverage. That matters because most frontier-model narratives still orbit San Francisco. This round suggests top-tier compute, capital, and talent can sti(cnbc.com) room. (eu-startups.com) ### Bottom line? This is a giant, very early wager that the next leap in AI will come from systems that learn by doing, not just by reading. Maybe that works. Maybe it runs into the old problem that real-world learning is expensive, slow, and hard to simulate. But the size of the round already tells you something important: investors think the post-chatbot race has started. (cnbc.com)