**Recursive raises $500M at $4B valuation with backing from Google and Nvidia**
The AMW Read
Novelty 2: New top-tier entrant with heavyweight founders, updates the foundation-model player map ($4B valuation is among highest for pre-product labs). Significance 2: Could reshape AI research economics and capital dynamics across frontier labs.
**Recursive raises $500M at $4B valuation with backing from Google and Nvidia**
Recursive Superintelligence, a four-month-old AI startup founded by former Salesforce chief scientist Richard Socher, has raised a $500 million round at a $4 billion valuation. GV (Google Ventures) led the round, with Nvidia participating. The company aims to build an AI system capable of autonomously performing scientific research — forming hypotheses, designing experiments, and iterating results — effectively removing human researchers from the loop.
This event signals the latest iteration of a recurring pattern in AI markets: the "fastest ARR ramp" model applied to early-stage research labs, where investor FOMO around next-generation capabilities drives sky-high valuations for pre-product entities. The involvement of both Google and Nvidia highlights a dual hedge — Google secures exposure to frontier self-learning research beyond DeepMind, while Nvidia ensures that compute-intensive autonomous research workloads will further fuel GPU demand. The round was reportedly oversubscribed, with a potential upsize to $1 billion.
For the AI industry, this validates that capital markets are now willing to finance foundational research direction itself, not just product-market fit. It also deepens the talent-poaching war between incumbents and well-funded startups. If Recursive succeeds, it could redefine the economics of AI research by replacing million-dollar human researchers with automated systems. If it fails, it will join a growing graveyard of overhyped AI research bets. Either outcome has implications for every segment reliant on frontier models.
