A team of fewer than 30 people. A valuation of $4.65 billion. And a bet that the fastest path to superintelligence is AI that rewrites itself.
Recursive Superintelligence, a London-based artificial intelligence startup incorporated just months ago, has emerged from stealth with a funding round of more than $650 million at a post-money valuation of $4.65 billion — one of the most striking early-stage raises in recent memory for a company with fewer than 30 employees.
The round was led by GV, Google’s venture capital arm, and US venture firm Greycroft, with participation from chip manufacturers Nvidia and AMD. The raise had been previously reported; Tuesday’s announcement confirms the figures and brings the company’s work into public view for the first time.
Also Read: Why Most AI Pilots Never Make It to Production
The Founding Thesis
Recursive Superintelligence’s central bet is that the fastest path to artificial general intelligence — and ultimately superintelligence — runs through AI systems capable of improving themselves by analyzing their own performance, without human intervention at each iteration. The company calls this recursive self-improvement, and it frames the approach as a structural departure from how frontier AI labs currently operate.
“The fastest path to superintelligence will be realized by AI that recursively improves itself, and does so via open-ended algorithms that drive endless innovation,” the company wrote in a public statement on its launch. “We will first focus on the science of AI itself — by creating AI that improves AI — but the playbook we create will soon allow us to revolutionize every scientific discipline.”
The Founders
The company was co-founded by Richard Socher, who serves as chief executive and previously held the role of chief scientist at Salesforce, and Tim Rocktäschel, a professor of artificial intelligence at University College London and a former scientist at Google DeepMind. Other members of the team have previously worked at Meta and OpenAI.
The combination of academic AI research credentials and frontier industry experience is deliberate — the company’s thesis depends as much on scientific originality as on engineering execution.
Also Read: Healthcare Is Scaling AI Without the Infrastructure to Manage It
A Crowded but Consequential Race
Recursive Superintelligence enters a field that has grown considerably more competitive in recent months. It joins a small but well-funded group of new AI research ventures pursuing novel paths to more capable AI systems, including AMI Labs, founded by former Meta chief scientist Yann LeCun, and Ineffable Intelligence, the $1.1 billion seed-funded lab launched by DeepMind alumnus David Silver.
What distinguishes Recursive Superintelligence’s positioning is its explicit focus on self-improving systems — a concept that has occupied the theoretical edges of AI research for decades but has rarely been the organizing mission of a well-capitalized commercial entity.
The company has offices in London and San Francisco. It has not disclosed a product timeline or detailed how it plans to ensure the safety of systems designed to modify their own architecture without human oversight — a question that, given the stated ambitions, will attract considerable scrutiny as the company’s work becomes more visible.
At $4.65 billion, the market is making a substantial early bet that the answer will be worth waiting for.
