After leaving Meta, Yann LeCun raised $7 billion in two months.

Just as the feud between Lin Junyang and Alibaba settled down, there came financing news about Yann LeCun, another top - tier tech expert who left a major AI company.

This Turing Award winner and former Chief Scientist at Meta started to establish his own AI startup after announcing his departure at the end of last year. By January this year, his world - model startup, AMI Labs, was officially founded at its headquarters in Paris.

Two months later, AMI Labs officially announced a financing of $1.03 billion (approximately 7.087 billion RMB). It was co - led by five institutions and angel investors, including Cathay Innovation and Bezos Expeditions (the family investment company of Amazon founder Jeff Bezos). Strategic investors include well - known companies such as NVIDIA, Toyota Ventures, and Temasek. Other investors include former Google CEO Eric Schmidt and Samsung.

It is reported that the pre - investment valuation of AMI Labs for this financing was $3.5 billion (approximately 24.082 billion RMB), and the post - investment valuation has not been disclosed yet. Before this, Yann LeCun planned to conduct the first - round financing at a valuation of 3 billion euros (approximately 24.117 billion RMB) with a target of raising 500 million euros. The result clearly exceeded expectations.

After the popularity of large language models (LLMs) represented by ChatGPT, world models are almost regarded as another technological disruption in the AI wave. Not only Yann LeCun, but also Fei - Fei Li and others have started their entrepreneurial financing.

Top - tier Tech Expert Assembles a Luxury Co - founding Team

In the field of AI, Yann LeCun is undoubtedly a top - tier tech expert.

As an important pioneer of deep learning and one of the main proposers of convolutional neural networks, since he founded the Meta Fundamental AI Research (FAIR) in 2013, he has been the spiritual totem of AI R & D at this social media giant.

However, in November 2025, this scholar who had long served as Meta's Chief AI Scientist began to plan to leave the company within a few months to establish a new AI startup.

This news sparked extensive discussions in the industry. By December, Yann LeCun officially confirmed the information of the new company, AMI Labs, on LinkedIn. He serves as the Executive Chairman of the company, and the CEO, Alexandre LeBrun, was invited from the French health - tech company Nabla. The company is headquartered in Paris, which is also Yann LeCun's birthplace.

AMI Labs has a very clear positioning: it bets on "world models" and does not focus on LLMs.

According to the company's official website, AMI Labs is developing a model system that can learn abstract representations from real - world sensor data. This system aims to ignore unpredictable details and make predictions in the representation space, thus forming an intelligent system that can understand the real world, has long - term memory, can perform reasoning and planning, and remains safe and controllable. Its potential application scenarios include industrial process control, automation systems, wearable devices, robots, and medical health, which have extremely high requirements for reliability and safety.

In terms of team configuration, the co - founder lineup of the new company is quite luxurious. Xiesaining, a former Meta AI research scientist and an assistant professor at New York University, serves as the Chief Science Officer; Pascale Fung, a professor in the Department of Electronic and Computer Engineering and the Department of Computer Science and Engineering at the Hong Kong University of Science and Technology, serves as the Chief Research and Innovation Officer; Laurent Solly, the former vice - president of Meta in Europe, serves as the Chief Operating Officer of AMI; and Michael Rabbat, the former director of Meta AI research, serves as the Vice - President of Global Models.

Currently, AMI Labs has set up offices in Paris, New York, Montreal, and Singapore, and is continuously recruiting core positions such as AI scientists and engineers on its official website.

Alexandre LeBrun also revealed that the company will open - source a large amount of code in the future: "We believe that open - sourcing can significantly accelerate technological progress. Building a community and a research ecosystem is also in the long - term interests of the company."

Jensen Huang Predicts a $90 - Trillion Scale

For a long time, Yann LeCun has always adhered to a controversial judgment: "Large language models (LLMs) represented by ChatGPT will never lead to AGI; it's a dead end."

To simply understand the difference between the two, LLMs learn "how humans speak," while world models learn "how physical laws work."

Now, corresponding to this technological divergence, the popularity of the "world model" track is continuously rising. As Alexandre LeBrun said, financing activities around this direction are significantly increasing.

On February 19th, World Labs, a spatial intelligence company founded by Fei - Fei Li, announced the completion of a $1 - billion financing, with a post - investment valuation of approximately $5 billion. The investors include a $200 - million investment from Autodesk, as well as industrial capitals such as NVIDIA and AMD. Although the company has not publicly disclosed the specific valuation, the market generally expects its valuation to be close to $5 billion.

At the product level, World Labs released its first spatial intelligence product, Marble, at the end of last year. The company describes it as a foundational model that can generate a three - dimensional world based on image or text prompts. By processing visual data from the real environment, it gradually builds an understanding of the physical space and develops more advanced reasoning abilities on this basis.

With this round of financing, World Labs plans to further strengthen its application capabilities in fields such as robotics and scientific discovery. Fei - Fei Li has also said that future spatial intelligence models are expected to play an important role in scenarios such as augmented reality, virtual reality, and robotics.

As one of the important explorers in the "world model" direction, World Labs is not an isolated case. More and more AI companies are starting to invest in the R & D of this technological path. The core goal of the so - called "world model" is to enable AI to understand the three - dimensional physical world and perform navigation, prediction, and decision - making in it.

Google DeepMind is also advancing related research. Its Genie model can generate and simulate a three - dimensional interactive environment and is regarded as an important attempt towards an interactive world model.

The trends of industrial capital are also worthy of attention. The heavy investment of software company Autodesk in World Labs to some extent reflects the transformation anxiety of the software industry under the impact of AI. On the day the investment news was announced, the stock price of Autodesk (ADSK) rose by nearly 2%. However, since the beginning of the year, the company's stock price has fallen by more than 22%. At the same time, the company has also filed a lawsuit against Google, accusing its AI video - generation tool of infringement.

In fact, for AI companies to build world models, they need the ability to perceive, generate, and interact with the three - dimensional world, which has also brought back the attention of a few software companies with three - dimensional industrial modeling capabilities. For these companies, AI may not only be an impact but also a new growth opportunity.

Earlier this month, French industrial software giant Dassault Systèmes also announced a long - term strategic partnership with NVIDIA to jointly promote technologies related to "world models." NVIDIA CEO Jensen Huang once predicted that "physical AI" for the real world will become the next important frontier of artificial intelligence, and its potential market size may reach up to $90 trillion.

Capital Competes Fiercely for Tech Geniuses

Yann LeCun warned in an interview: "Many people have left, and many who haven't will also leave."

Tech idealists often make similar choices. After leaving, Yann LeCun founded AMI Labs, focusing on the "world models" he has been longing for. Tian Yuandong, the former research director of the Meta FAIR team, also rejected the offers from large companies and announced his entrepreneurship. The whereabouts of Lin Junyang from Alibaba are also worthy of expectation.

In the view of Sebastian Mallaby, the author of A History of Venture Capital, rebellion is the origin of venture capital. Venture capital liberates entrepreneurs, allowing them to freely leave large companies, rebel against their bosses, and establish new companies with different thinking.

The "Traitorous Eight" of Fairchild Semiconductor left and founded companies such as Intel and AMD, reshaping the global semiconductor industry landscape; Apple co - founder Steve Wozniak chose to leave after the company entered the commercial expansion period and continued to focus on his technological innovation. Countless tech talents choose to leave in the face of the business strategies of large companies, and their departures often drive a new round of innovation and transformation in the industry.

In the current AI wave, such stories are not uncommon. For example:

Jia Yangqing (former vice - president of technology at Alibaba) founded Lepton AI, which was acquired by NVIDIA for hundreds of millions of dollars in 2025;

Jing Kun (former CEO of Xiaodu) founded the AI agent platform Genspark and completed three rounds of financing within a year and a half, becoming a unicorn;

Zhou Chang (former core member of Alibaba Tongyi Qianwen) led more than a dozen team members to join ByteDance, which led to a non - competition dispute;

David Silver (former Chief Scientist at DeepMind) founded Ineffable Intelligence, and the seed - round financing reached $1 billion.

Competing for these leading tech figures from large companies has become a "must - do" for many investment institutions. They have often held key positions in core technology departments, have outstanding technical reputations, and have accumulated rich resources between the industry and capital.

Even before the projects of these tech stars are officially launched, they have been surrounded by sensitive VCs. As for who can take the lead, it depends on their own abilities.

This article is from the WeChat official account "China Venture Capital". Author: Wei Xianghui. Republished by 36Kr with authorization.