Recursive Superintelligence Raises $650M at $4B Valuation to Build Self-Improving AI

A new AI startup called Recursive Superintelligence has raised more than $650 million from backers including GV (formerly Google Ventures), Greycroft, Nvidia, and AMD, landing a valuation north of $4 billion — despite having fewer than 30 employees and being only six months old. The company is chasing one of AI's most audacious goals: building systems capable of improving themselves with little or no human input, a concept known as recursive self-improvement.

The eight co-founders bring serious pedigree. Richard Socher — who leads the company while also serving as CEO of AI search startup You.com — previously ran AI research at Salesforce. His co-founders include Josh Tobin, Jeff Clune, and Tim Shi from OpenAI, and Yuandong Tian from Meta. The company has also recruited Peter Norvig, who spent 25 years as Google's director of research and co-authored the foundational AI textbook 'Artificial Intelligence: A Modern Approach,' which has been a university standard for 30 years.

The core thesis is straightforward but staggering in its implications: AI systems are fundamentally software, and modern AI has become capable enough to write sophisticated code. If those two facts hold simultaneously, the theory goes, AI could in principle write better versions of itself — triggering an accelerating feedback loop. Many founders specialize in 'open-endedness,' a discipline focused on building AI that can pursue researcher-defined goals autonomously over periods of days, months, or years.

Recursive Superintelligence is not alone in this race. A rival startup called Ricursive Intelligence is also valued at $4 billion and pursuing near-identical goals. OpenAI has announced plans to build an 'automated AI researcher' and CEO Sam Altman has said the company aims to have a system capable of doing the work of a less experienced human researcher by fall 2026. Anthropic is also working in the same direction.

Socher himself acknowledges the technology is years away from maturity and that humans still generate the core ideas driving AI forward. The long-term vision extends well beyond AI itself — the company hopes to eventually apply recursive self-improvement to drug discovery and other biological research domains.