Recursive Superintelligence raises $500M at $4B valuation

In short: A new London AI startup called Recursive Superintelligence raised $500 million and reached a reported $4 billion valuation, with plans to publicly launch in mid-May 2026.

What happened

Recursive Superintelligence is an AI company founded in London in December 2025. It was started by well known AI researchers, including Richard Socher (formerly Chief Scientist at Salesforce), Tim Rocktäschel (a University College London professor and former Google DeepMind scientist), and three former OpenAI researchers, Josh Tobin, Jeff Clune, and Tim Shi.

The company raised $500 million in a funding round led by Google Ventures (GV), with participation from NVIDIA. The New York Times reported that the round was oversubscribed, meaning more investors wanted in than the company chose to accept. The funding round valued the company at about $4 billion, just four months after it was founded.

Recursive Superintelligence says its goal is to automate AI development. That includes picking research ideas, running many experiments, judging results, choosing training data, and training and refining models. You can think of it like trying to build a lab assistant that not only runs tests, but also decides which tests to run next.

The startup reportedly has around 20 employees. Its team includes people who previously worked at places like Google DeepMind, OpenAI, Meta, and Google.

Why it matters

If companies succeed at automating more of AI research, progress in AI could speed up and require fewer human researchers to do the routine work. That could affect how quickly new AI systems arrive in products and services people use, and it could also raise new questions about oversight and safety when machines help steer the research.

Source: NYTimes