Recursive Superintelligence Secures $500M Funding at $4B Valuation
The AMW Read
The article introduces a new top-tier entrant targeting the automation of the foundation model development cycle, supported by a $500M round (cross.§D) and elite talent from OpenAI/DeepMind (cross.§C).
Recursive Superintelligence Secures $500M Funding at $4B Valuation
London-based startup Recursive Superintelligence has raised at least $500 million in an oversubscribed funding round led by Google's venture arm GV, with participation from Nvidia. The four-month-old company, founded by former Salesforce chief scientist Richard Socher alongside researchers from DeepMind and OpenAI, is valued at $4 billion pre-money. The capital injection could potentially reach $1 billion depending on the finalization of the oversubscription. The company currently employs approximately 20 staff members and is focused on developing self-improving AI systems designed to automate the complete frontier AI development pipeline, including data selection, training, and research direction.
This massive capital infusion for a company with only 20 employees highlights the extreme premium being placed on elite research talent and the specific pursuit of AI automation. By targeting the automation of the entire development cycle—from evaluation to post-training—Recursive Superintelligence is positioning itself to address the scaling bottlenecks currently faced by human-centric research labs. The involvement of Nvidia suggests a strategic interest in software innovators that drive high-intensity compute demand, while GV's leadership indicates that major players view autonomous development tools as a critical, rather than purely competitive, layer of the evolving AI stack.
The shift toward self-improving systems represents a pivot from scaling existing architectures to attempting to automate the architectural evolution itself. If the team can deliver on their vision of a system that manages its own research direction and training, they could fundamentally alter the unit economics of frontier model development by reducing human dependency. However, the primary challenge remains whether these recursive loops can maintain technical progress without the grounding influence of human-led evaluation, especially as they prepare for a planned public launch in mid-May 2026.
#AI #VentureCapital #Nvidia #GoogleGV #RecursiveSuperintelligence #MachineLearning
