Recursive Superintelligence

In an increasingly crowded AI landscape, Recursive Superintelligence chose to stay in stealth until it had something concrete to show. When it emerged in May 2026, it did so with $650 million in funding and a $4.65 billion valuation — one of the most spectacular debuts in European venture history. The startup was founded in London in 2025 by a team drawn from the world's most important AI organisations: former leaders and researchers from Meta AI, Google DeepMind, OpenAI, and Salesforce AI. The mission is to build AI systems capable of recursive self-improvement — models that can improve themselves autonomously, iterating on their own architectures and capabilities without continuous human supervision. The round was backed by GV (Google's venture fund), Greycroft, Nvidia, and AMD — the participation of the two leading AI chip manufacturers is not coincidental: Recursive Superintelligence works directly with hardware as a critical layer for making self-improvement viable at scale. The project fits into a moment where competition to build the next generation of AI labs has gone global. Recursive represents the most ambitious European bet in this space, with the thesis that recursive self-improvement is not just a theoretical horizon but an engineering problem solvable with the right talent, the right computational resources, and the right architectural approach.