Research
Recursive Superintelligence
Lab led by Richard Socher that aims to reach Super Intelligence through recursive self-improvement; raised $650 million at a $4.65 billion valuation.
Last verified
In its own words
Our main focus is to build truly recursive, self-improving superintelligence at scale.Source, May 13, 2026Recursive Superintelligence, usually shortened to Recursive, is an AI lab with offices in London and San Francisco. It is led by Richard Socher, the former chief scientist of Salesforce and the founder of You.com, with a founding team drawn from Google DeepMind, OpenAI, Meta and universities. Its co-founders include Tim Rocktäschel, Yuandong Tian and Tim Shi.
The company emerged from stealth on May 13, 2026, with $650 million in funding at a $4.65 billion valuation. GV and Greycroft led the round, and NVIDIA and AMD Ventures took part. At the time Recursive had fewer than 30 employees and no public product.
Its approach is in its name. Recursive wants to automate AI research itself, so that a system can propose ideas, implement them, test them and improve on the result without people running the cycle, and then apply the same method to other sciences. Socher has argued that most AI-assisted coding and research today does not count as recursive self-improvement, because people still run that cycle. Our guide to recursive self-improvement explains the concept and its history.
What Recursive Superintelligence has said about Super Intelligence
Recursive uses the term in its research sense and puts it in the company name. At launch Socher described the goal as “to build truly recursive, self-improving superintelligence at scale.” The launch is logged on our timeline. The name is four months older than Executive Order 14434, so it is not the US federal label for AI.
That places Recursive with Safe Superintelligence and Ineffable Intelligence among the labs that name Super Intelligence as what they are building. Recursive has not claimed to have built such a system.
Key facts
- Launched publicly: May 13, 2026
- Leadership: Richard Socher (CEO)
- Funding: $650 million at a $4.65 billion valuation, led by GV and Greycroft
- Offices: London and San Francisco
For current details see recursive.com.
On the Rise of Super Intelligence timeline
Full timeline →Related research and guides
Frequently asked questions
What is Recursive Superintelligence?
Recursive Superintelligence, also known as Recursive, is an AI lab led by Richard Socher that aims to build AI systems able to improve themselves. Recursive Superintelligence emerged from stealth on May 13, 2026, with $650 million in funding. It has offices in London and San Francisco and has not released a public product.
Who founded Recursive Superintelligence?
Richard Socher, the former chief scientist of Salesforce and founder of You.com, is chief executive and co-founder of Recursive Superintelligence. The founding team includes researchers from Google DeepMind, OpenAI, Meta and universities, among them Tim Rocktäschel, Yuandong Tian and Tim Shi.
How much has Recursive Superintelligence raised?
Recursive Superintelligence raised $650 million at a $4.65 billion valuation, announced on May 13, 2026. GV and Greycroft led the round, with NVIDIA and AMD Ventures taking part. At the time Recursive Superintelligence had fewer than 30 employees.
What does recursive self-improvement mean at Recursive Superintelligence?
Recursive self-improvement means an AI system that improves its own design, so that each version helps build a better one. Richard Socher has said Recursive Superintelligence's main focus is "to build truly recursive, self-improving superintelligence at scale," starting with the automation of AI research itself.