In 1956, a group of scientists gathered at Dartmouth to formally discuss "Can machines think." They optimistically thought they could solve this problem in just one summer.
Seventy years later, this question still remains unanswered. However, there is a company that was just founded four months ago, yet has secured a $500 million investment, valuing it at $4 billion, all because it claims to have found a way for AI to learn to conduct research and evolve on its own.
This company is called Recursive Superintelligence.
Google Ventures (GV) led the funding round, with NVIDIA participating. The positions of these two companies in the AI ecosystem need no elaboration. By both investing in a startup that has not even released a product yet, the underlying logic behind their decision is worth careful analysis.
01 "Removing Humans from the Loop"
First, let's talk about what Recursive Superintelligence is actually doing.
The company was founded by former Salesforce Chief Scientist Richard Socher, and its core team hails from Google DeepMind and OpenAI. This is not an unfamiliar combination—over the past two years, there has been a noticeable wave of engineers and researchers leaving top labs to start their own ventures.
Socher is not the typical Silicon Valley founder who "got gold at a big company." Born in Germany in 1983, he studied under AI pioneer Andrew Ng and NLP authority Christopher Manning at Stanford University, completed his doctoral dissertation in 2014, and won the Stanford Computer Science Department's Best Doctoral Dissertation Award that year.
Richard Socher is one of the key figures who truly brought neural network approaches into the field of natural language processing. His early research on word vectors, context vectors, and cue engineering directly laid the technical foundation for today's BERT and GPT series models, with over 180,000 Google Scholar citations.
Upon completing his PhD, he founded the AI startup MetaMind, which was later acquired by Salesforce in a strategic acquisition two years later. Subsequently, he led Salesforce's AI strategy for several years as Chief Scientist and Executive Vice President, overseeing the development of enterprise AI products such as Einstein and GPT.
After leaving Salesforce, he founded the AI search engine You.com in 2020, which completed its Series C funding round in 2025, reaching a valuation of $1.5 billion. This time, he shifted his focus from search to a more foundational proposition.
Thinking Machines Lab, Safe Superintelligence, Ineffable Intelligence, Advanced Machine Intelligence Labs... Each one emerges with the label "former XX Large Model Core Team" and each tells a story of "next-generation AI."
However, Recursive's approach is more radical than most peers.
Its core proposition is "self-learning AI" — not to make AI smarter at answering questions, but to enable AI to autonomously complete the entire scientific research process: propose hypotheses, design experiments, evaluate results, iterate directions. In other words, it aims to completely remove human researchers from this cycle.
This is not a new direction, but Recursive places it within an extremely practical business logic. Currently, top AI researchers command annual salaries ranging from $15 million to $20 million. If a system can accomplish the same work at a lower cost and faster pace, the economic model of cutting-edge research will be fundamentally rewritten.
Investors evidently see this logic. The funding round was reportedly oversubscribed, with the final size potentially reaching $1 billion.
02 Google and NVIDIA Simultaneously Bet
GV leads, with NVIDIA participating. This investor combination itself is a signal.
Google's logic is not difficult to understand. DeepMind has long been the most important explorer in the "AI for Science" direction, with AlphaFold cracking the protein folding problem and AlphaGeometry surpassing human top competitors in math competitions.
However, DeepMind's approach is to use AI to solve specific scientific problems, while Recursive aims to do something more foundational—empowering AI systems to drive the scientific discovery process autonomously. This is both a competitive and a hedgeable relationship for Google.
More importantly, just earlier this month, Google and Intel announced a collaboration agreement for multiple generations of AI infrastructure. This indicates that Google's layout at the AI infrastructure level is gaining momentum. The investment in Recursive is a piece on this big chessboard—Google wants a stake in whoever is leading in state-of-the-art models.
NVIDIA's logic, on the other hand, is more direct. The core bottleneck of self-learning AI is not the algorithm but the computing power. If AI is to autonomously conduct experiments and iterate models, the scale of GPU clusters needed behind the scenes grows exponentially. NVIDIA's investment in Recursive is, to some extent, an investment in its own future orders.
With both companies making a move, they have also sent out a more subtle signal—the race may have reached a stage where "not investing means falling behind."
03 Is a $4 Billion Valuation Reasonable in Four Months?
It is estimated that when everyone saw the $4 billion figure for the first time, their initial reaction was "here we go again."
The AI startup valuation bubble has been a familiar topic in the past two years. A PDF, a demo, a few slides, along with some names from top labs, can leverage a few billion dollars—this is no longer a legend but a daily occurrence in Silicon Valley and London.
However, looking closely at Recursive's situation, there are several points that differentiate it from the typical "PPT unicorn."
First, the caliber of the founding team. Richard Socher has real academic achievements in the NLP field, not solely relying on the halo of a "former big tech company." The core team's experience at DeepMind and OpenAI also means they have truly encountered the pain points of cutting-edge research.
Second, the fact of oversubscribed funding. This implies that market demand far exceeds supply, with investors clamoring to get in rather than being persuaded to come in.
However, a $4 billion valuation for a four-month-old company with no public product is based on expectations, not reality. This is essentially paying for a direction rather than a product or revenue.
This kind of pricing logic is becoming increasingly common in the AI era, driven by investors' deep-seated fear of "missing out on the next OpenAI." Safe Superintelligence also received a sky-high valuation with almost no product back in the day, and Ilya Sutskever's name was the strongest asset.
Recursive is copying the same path. This is not a criticism, but an objective observation.
04 The Door Behind "Self-Study"
The name Recursive Superintelligence actually makes the company's ambition very clear.
“Recursive” means recursive. In computer science, recursion is a structure where a function calls itself, which is the core mechanism of many complex algorithms. In the context of AI research, the implication of "recursive superintelligence" is a system that can continuously optimize itself, a process of spiraling improvement.
This concept is not new. Its extreme version is the "intelligence explosion" — once a system surpasses a certain threshold, it can autonomously accelerate its own evolution, ultimately reaching a level of intelligence incomprehensible to humans. This has been one of the core concerns in the field of AI safety for a long time.
However, what Recursive is doing now is likely far from this level. A more realistic interpretation is that it is attempting to build a system that can autonomously drive a cycle of scientific exploration, with the goal of significantly reducing the human and time costs of AI research.
If it can really achieve this, the impact will not only stay within the AI community. It means that fields such as drug development, materials science, physics, and others may enter a stage where progress can rapidly advance without the involvement of human scientists.
Of course, this is still an "if."
From claim to implementation, the distance in the AI industry has never been linear.
05 The Logic of the Wave
Since the second half of 2025, there has been wave after wave of exodus from top labs to entrepreneurship. Thinking Machines Lab, Safe Superintelligence, Ineffable Intelligence... This list is still growing.
Recursive is the newest in this wave and currently the most highly valued.
The structural reason behind this is simple — the competition among OpenAI, Anthropic, Google DeepMind has made these leading labs more and more like big companies with KPIs, compliance, and politics.
Researchers who truly want to bet on the most radical directions feel that going out on their own offers more freedom.
Meanwhile, the logic of the capital market is reinforcing this trend. For top researchers backed by tech giants, the current window of opportunity for entrepreneurship may be the best in history—investors are more willing than ever to pay for the "vision."
At the core of this wave, the most pressing question is not "who will succeed," but rather "what does success look like?"
If Recursive ultimately proves the feasibility of self-learning AI, it will redefine the underlying paradigm of AI research. If it fails to do so, after burning through $500 million in ammunition, all that will be left is another concept that was overly hyped.
Both possibilities are equally real.
Four months, $4 billion valuation—this number is both exciting and cautionary. In the AI arms race that has developed to this day, even the act of "how to do research" has become a battlefield of competition.
After scientists debated an issue at Dartmouth for a summer, now someone intends to use AI to answer it—using AI to research AI, pushing towards superintelligence in a recursive manner.
Where this path leads, no one truly knows. But evidently, Google and NVIDIA have already decided that regardless of the destination, they cannot afford to be absent.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
