Yann LeCun’s AMI Labs raises $1.03 billion to build world models

AMI laboratoriesthe new venture co-founded by Turing Prize winner Yann LeCun after he left Meta has raised $1.03 billion at a pre-money valuation of $3.5 billion. AMI is working on world models, or AI that learns from reality, not just from language.

This category has fewer players than generative AI, but perhaps not for long. “My prediction is that ‘world models’ will be the next buzzword,” Alexandre LeBrun, CEO of AMI Labs, told TechCrunch. “In six months, every company will call itself a global model for attracting financing.”

LeBrun said this with a smile because he thinks AMI Labs is fundamentally different: its goal is to understand the real world. This could have applications in healthcare, where AMI Labs’ first partner will be Nabla, the digital healthcare startup he now chairs.

As CEO of Nabla, LeBrun had come to the same conclusion as LeCun about the limitations of large language models (LLMs) where hallucinations could have life-threatening consequences. But he also knows that it will take a while before the startup offers a viable alternative based on that JEPAthe Joint Embedding Predictive Architecture proposed by LeCun in 2022.

“AMI Labs is a very ambitious project because it starts with basic research. It is not the typical applied AI startup that can release a product in three months, generate sales in six months, and generate $10 million in revenue. [annual recurring revenue] in twelve months,” says LeBrun. In contrast, it can take years for world models to move from theory to commercial applications.

Despite this time horizon, companies developing global models have received big checks. SpAItial has raised a $13 million seed round – unusually large for a European startup; while Fei-Fei Li’s World Labs raked in a whopping $1 billion last month alone. Now AMI Labs is joining the club with more funding than initially thought.

The French AI laboratory was said to be looking for just €500 million last December, but ended up raising around €890 million, likely thanks to his team. In addition to LeCun’s involvement as chairman and LeBrun’s track record as an entrepreneur, it also boasts Meta’s VP of Europe Laurent Solly as COO, and high-profile researchers Saining Xie as chief science officer, Pascale Fung as chief research & innovation officer and Michael Rabbat as VP of world models.

According to LeBrun, the high interest gave the startup the opportunity to take its pick from investors, both in terms of alignment of expectations and background. The round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital and Bezos Expeditions, with participation from several other funds and industry backers, as well as individuals including Tim and Rosemary Berners-Lee, Jim Breyer, Mark Cuban, Mark Leslie, Xavier Niel and Eric Schmidt.

Beyond the added value, this funding will give AMI Labs a meaningful runway to fund its two major cost centers: compute and talent. LeBrun said he will put quality over quantity to build the AMI Labs team in four key locations: Paris, where its headquarters are located; New York, where LeCun teaches at NYU; in Montreal, where Rabbit is located; and in Singapore, both to recruit AI talent and to be close to future customers in Asia.

Although AMI Labs has no plans to generate revenue for the time being, it still plans to engage with potential customers early on. “We’re developing world models that try to understand the world, and you can’t do that locked up in a lab. At some point we have to put the model in a real situation with real data and real evaluations,” LeBrun said.

When the time comes, AMI Labs will turn to partners to explore implementations – and Nabla is the first announced partner to expect to have access to these early models, but certainly not the last. “This could explain the presence and strong interest of certain industry players and potential partners in the investment round,” LeBrun said.

In addition to its lead investors and angels, AMI Labs is backed by NVIDIA, Samsung, Sea, Temasek and Toyota Ventures, as well as French players Association Familiale Mulliez, Groupe Industriel Marcel Dassault and Publicis Groupe. Aglaé Lab, Alpha Intelligence Capital, Artémis, Bpifrance Digital Venture, New Legacy Ventures, SBVA and ZEBOX Ventures also participated.

It may take some time before these investments are translated into commercial applications. But staying true to LeCun’s beliefs, AMI Labs will publish articles as it goes.

“We’ll also open source a lot of code,” said LeBrun, who had also worked at Meta’s AI research lab FAIR. Although open research is “becoming increasingly rare,” the founders of this startup still believe in it. “We think things move faster when they are open, and it is in our interest to build a community and a research ecosystem around us.”