Recursive Superintelligence came out of stealth on May 13 with $650M raised at a $4.65B valuation. Under 30 employees. Nvidia, AMD Ventures, GV, and Greycroft all wrote checks. That’s roughly $22M per head before they’ve shipped a thing.
What they’re actually building
Not a product yet — a San Francisco research lab with a thesis. The pitch: a model that audits its own weaknesses and rewrites itself. Ideation, implementation, validation, all automated, no human in the loop. Think AlphaEvolve applied to the model itself, not just to code. First product is targeted for mid-2026. No API, no SDK, no demo. If you want to talk to it today, you can’t.
Why the check size kind of makes sense
Richard Socher was Salesforce’s chief scientist. Co-founder Tim Rocktäschel came out of DeepMind and UCL. Peter Norvig is on the team. Cresta co-founder Tim Shi too. If you believe self-improving AI is the path to superintelligence, this is the team you back — and Nvidia plus AMD writing the same check signals the chip side wants exposure to the next foundation-model contender before OpenAI swallows the category.
The risk: recursive self-improvement has been a research direction for twenty years. Nobody has made it compound. $650M buys a lot of compute to find out.
You Might Also Like
- Meridian Just Came out of Stealth and Finance Bros are Paying Attention
- Google Deepmind Aletheia Just Solved Math Problems Nobody Could Heres why That Matters
- Ricursive Intelligence Just Raised 335m in 4 Months and i Think Nvidia Should be Paying Attention
- Nvidia Personaplex Just Made Every Voice ai Pipeline Feel Obsolete
- Nvidia Dreamdojo Teaching Robots to Think by Watching 44000 Hours of us
Top AI Product 
Leave a comment