David Silver-founded Ineffable Intelligence raises $1.1B at $5.1B valuation for AI self-improvement models
The AMW Read
Introduces a new top-tier entrant (Silver-founded, $1.1B) in the RSI subcategory, which could invalidate the assumption that model-scale is the only path to capability gains; $1.1B at $5.1B valuation triggers cross.§D; the open-source OpenMLE release from a competing team (Frontis-MA1) provides a co
David Silver-founded Ineffable Intelligence raises $1.1B at $5.1B valuation for AI self-improvement models
Ineffable Intelligence, founded by AlphaGo co-creator David Silver, has raised $1.1 billion in a funding round that values the company at $5.1 billion. The startup is building what it calls a "super learner" — an AI system that uses reinforcement learning and self-discovered knowledge to improve its own capabilities, rather than simply scaling up chatbot-style models. The round is part of a broader wave of capital flowing into recursive self-improvement (RSI) startups, including Recursive Superintelligence ($650M), Ricursive Intelligence ($300M), and Sakana AI ($200M), totaling over $2.5 billion in disclosed funding across the nascent category.
Why it matters: This event signals that the market is beginning to price AI companies on a new variable — the rate at which a system can autonomously accelerate its own capability improvement. If recursive self-improvement transitions from a research concept to a production reality, it could rewrite the capital-cycle dynamics that have historically rewarded model-size scaling. The capital is betting that the fastest-ARR-ramp pattern of the next decade belongs not to any single foundation model, but to the company that can build a closed loop of AI-conducted experimentation, feedback, and system-level improvement.
The core technical challenge here is the shift from linear agent loops to graph-based evolutionary search, as demonstrated by the Frontis-MA1 and OpenMLE open-source framework also discussed in the article. In that paradigm, each experimental outcome becomes a node in a program genealogy, not just a token in a conversation history. The investor thesis is that the ability to run thousands of parallel, verifiable ML engineering experiments — and feed successful trajectories back into the model — creates a structural moat distinct from traditional model-scale or data-scale advantages. However, the RSI field remains early-stage: Anthropic and OpenAI have both noted that full recursive self-improvement has not yet arrived, and the distinction between AI4AI, meta-evolution, and genuine cross-generation RSI remains unresolved. This funding wave should be read as a directional bet, not a proven technology.
#IneffableIntelligence #DavidSilver #RecursiveSelfImprovement #AI4AI #MegaRound #CapitalCycle #FoundationModels
