Recursive Superintelligence has raised $500 million in a pre-Series A funding round led by GV, with participation from Nvidia. The round values the company at $4 billion and has been described as oversubscribed, with potential to reach $1 billion.
Founded in London last year, the company operates with a team of around 20 employees and has not yet launched a public product. Its founding team includes former researchers from organizations such as OpenAI, Google DeepMind, and Salesforce, including individuals such as Richard Socher and Tim Rocktäschel.
The company is focused on developing AI systems capable of recursive self-improvement, a concept that aims to enable models to design, test, and refine their own algorithms without human intervention. This approach seeks to move beyond current large language models, which rely heavily on human-labelled data and manual fine-tuning.
By building self-supervising architectures, Recursive Superintelligence is attempting to create systems that can autonomously generate training data and update their parameters, potentially transforming how AI models are developed and scaled.
Despite the ambition of this approach, no stable long-term results have been publicly demonstrated so far.
The funding comes amid a broader surge in global AI investment, with approximately $300 billion raised in the first quarter of the year, highlighting continued investor interest in next-generation AI systems.
