Self-learning AI company Recursive received US$500 million in financing: GV and NVIDIA are betting that the era of independent AI research is coming

Core conclusion

An AI company that was founded only four months ago and has no public products has received US$500 million in financing, and its valuation has soared to US$4 billion. This is not a bubble, it is investors betting in the direction of “AI doing scientific research independently”.

Key Points

• Event Time: May 6, 2026
• Company: Recursive Superintelligence
• Financing amount: US$500 million (reportedly oversubscribed and may eventually reach US$1 billion)
• Valuation: $4 billion
• Investors: Google Ventures (GV) led the investment, NVIDIA also participated
• Founder: Richard Socher (Google Scholar citations over 180,000 times, former chief scientist of Salesforce)
• Established: only 4 months (January 2026)

Who is Recursive Superintelligence?

This company is not just another AI chatbot. Its core proposition is "self-learning AI" - allowing AI to autonomously complete the entire scientific research process: proposing hypotheses, designing experiments, evaluating results, and iterating directions. The ultimate goal is to remove human researchers entirely from this cycle.

The founder is Richard Socher, who was born in Germany in 1983. He has a PhD from Stanford and studied under the guidance of AI pioneer Andrew Ng and NLP authority Christopher Manning. His doctoral thesis won the Stanford Computer Science Department's Best Paper Award that year, and he was one of the key figures in bringing neural network methods into natural language processing. Early research on word vectors and context vectors directly laid the technical foundation for the BERT and GPT series models.

Socher's entrepreneurial resume is extremely impressive: he founded MetaMind when he graduated with a Ph.D., which was acquired by Salesforce two years later. He has since served as chief scientist of Salesforce, leading enterprise-level AI product lines such as Einstein GPT. Founded the AI ​​search engine You.com in 2020 and completed Series C financing in 2025 with a valuation of US$1.5 billion.

US$500 million financing: What is the valuation logic?

A $4 billion valuation for a 4-month-old company with zero products is priced based not on revenue but on expectations.

Why did GV and Nvidia place bets at the same time?

Although Google and NVIDIA are both AI giants, they have completely different perspectives:

• Google (GV): DeepMind has been the most important explorer in the direction of "AI for Science" for many years. Investing in Recursive is both a strategic hedge and a layout. Earlier this month, Google just announced a multi-generation AI infrastructure cooperation agreement with Intel, and investment in AI scientific research is accelerating.

• NVIDIA: The core bottleneck of self-learning AI is computing power. AI autonomous running experiments require exponential growth in the size of the GPU cluster. Investing in Recursive is investing in your own future orders - if self-learning AI takes off, Nvidia's chip demand will increase by another order of magnitude.

What does "self-learning AI" mean?

The concept of Recursive Self-learning AI sounds like science fiction, but the logic chain is very clear:

1. Top AI researchers earn $15 million to $20 million annually
2. Talent bottleneck is the core limiting factor in cutting-edge research
3. If the system can complete the same scientific research work at a lower cost and faster speed, the economic model will be completely rewritten
4. The extreme version is "intelligence explosion" - once the system exceeds the critical point, it can autonomously accelerate its own evolution.

If Recursive succeeds, the fields of drug research and development, materials science, and physics may usher in a stage where "rapid advancement can be achieved without the participation of human scientists." Of course, this is also one of the long-term core concerns in the field of AI security.

Industry Trend: The trend of big names leaving to start businesses is intensifying

Since the second half of 2025, there has been one wave after another of people leaving top laboratories to start businesses:

• Safe Superintelligence (Ilya Sutskever, OpenAI co-founder)
• Thinking Machines Lab
• Ineffable Intelligence
• Advanced Machine Intelligence Labs

The common characteristics of these companies: top academic background + very early stage financing + sky-high valuation + no product yet. The investment logic has changed from "looking at the product" to "looking at people and betting"**.

Extended information

• YC holds 0.6% stake in OpenAI worth $5 billion: Controversy over conflict of interest disclosures returns
• Anthropic releases Claude Codex autonomous coding agent and announces $200,000 funding fund

Related reading

• Want to systematically understand how AI Agent is implemented? See: AI Agent-Driven Content Automation: n8n MCP Building Guide from Scratch
• Real case: See how this developer used AI Agent to build an automated system with a monthly income of US$5,000 from scratch: Indie Developer: n8n + OpenClaw Automation Workflow Earning $5,000/mo