Recursive Emerges from Stealth with $650M for Self-Improving AI
AI startup Recursive has stepped out of stealth, announcing $650 million in funding to create systems that improve themselves. The company positions this recursive self-improvement as the direct route to superintelligence. Its announcement includes this statement: "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."
Recursive plans to begin by developing AI that enhances other AI. From there, the team intends to apply the method to additional areas of science. Co-founder Tim Rocktäschel points to Stanisław Lem's idea of an "information barrier." This barrier occurs when knowledge expands so rapidly that humans cannot keep pace or absorb it effectively. The startup aims to overcome this by automating the entire scientific process, beginning with AI development.
Funding Round Details
The latest funding brings in $650 million, placing the company at a $4.65 billion valuation. GV, or Google Ventures, and Greycroft led the round. AMD Ventures and Nvidia also joined as investors. Back in April, the Financial Times noted a raise exceeding $500 million.
GV, Alphabet's investment arm, focuses on early-stage tech companies, including many in AI. Greycroft has a history of backing software and internet firms. AMD Ventures supports semiconductor and computing advancements, while Nvidia invests in AI hardware and software innovators.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Leadership and Team Background
Richard Socher, former leader at Salesforce's AI division, heads the company. Tim Rocktäschel, previously at Google DeepMind, serves as co-founder. The team includes researchers who worked at OpenAI, Meta, and Uber AI. Socher gained recognition for his natural language processing work at Stanford and later drove AI products at Salesforce. Rocktäschel specializes in machine reasoning and contributed to major projects at DeepMind.
OpenAI pioneered large language models like GPT. Meta advances AI through labs focused on efficiency and openness. Uber AI explored machine learning for autonomous systems before winding down. This mix brings deep expertise in AI training, scaling, and applications.
Vision for Breaking Barriers
Recursive has not released specific technical outcomes yet. The focus remains on building tools that automate discovery. By targeting self-improvement first, the company seeks continuous progress without human limits. This approach draws from concepts like Lem's barrier, where data floods beyond comprehension.
The startup's emergence comes amid growing interest in advanced AI capabilities. Investors see potential in methods that enable ongoing enhancement. Recursive joins others pushing boundaries in automation and intelligence.