Paris-based Advanced Machine Intelligence (AMI) Labs, founded by Turing Award winner Yann LeCun after his departure from Meta, has raised $1.03 billion in a funding round valuing the company at $3.5 billion pre-money. The startup is focused on developing “world models,” a distinct approach to artificial intelligence that emphasizes learning from a representation of reality, rather than solely relying on vast language datasets. This funding signals a growing interest in alternatives to the current generative AI boom, and positions AMI Labs as a key player in the emerging field.
Although generative AI currently dominates headlines, AMI Labs CEO Alexandre LeBrun predicts a shift in focus. “My prediction is that ‘world models’ will be the next buzzword,” LeBrun told TechCrunch. “In six months, every company will call itself a world model to raise funding.” LeBrun acknowledges the potential for hype, but believes AMI Labs’ fundamental difference – its commitment to understanding the real world – will set it apart. The company’s initial focus will be on applications in healthcare, partnering with digital health startup Nabla, where LeBrun also serves as chairman.
The limitations of large language models (LLMs), particularly the risk of “hallucinations” – generating incorrect or misleading information – were a key factor in LeBrun’s and LeCun’s decision to pursue world models. These models aim to build a more robust and reliable understanding of the world, crucial in high-stakes fields like medicine. AMI Labs is building its approach on the Joint Embedding Predictive Architecture (JEPA), a framework proposed by LeCun in 2022. Learn more about JEPA on Yann LeCun’s website.
A Growing Field Attracts Major Investment
AMI Labs isn’t alone in pursuing world models, but it’s quickly becoming a leader in the space. SpAItial, a European startup, recently raised a $13 million seed round, considered unusually large for the region. Fei-Fei Li’s World Labs secured a substantial $1 billion in funding last month. AMI Labs’ $890 million raise (equivalent to approximately €890 million at current exchange rates) surpasses initial expectations of €500 million reported last December.
The company’s strong team is a major draw for investors. In addition to LeCun’s role as chairman and LeBrun’s entrepreneurial experience, AMI Labs boasts Meta’s VP for Europe, Laurent Solly, as COO, and leading researchers including Saining Xie as chief science officer, Pascale Fung as chief research & innovation officer, and Michael Rabbat as VP of world models. This combination of academic expertise and industry leadership positions AMI Labs for ambitious growth.
Strategic Investment and Global Expansion
The investment round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, with participation from a diverse group of investors including Tim and Rosemary Berners-Lee, Jim Breyer, Mark Cuban, Mark Leslie, Xavier Niel, and Eric Schmidt. Strategic investors include NVIDIA, Samsung, Sea, Temasek, Toyota Ventures, and several prominent French firms like Association Familiale Mulliez and Groupe Industriel Marcel Dassault.
AMI Labs plans to allocate the funding to two key areas: compute power and talent acquisition. The company is prioritizing building a high-quality team across four locations: Paris (headquarters), New York (where LeCun teaches at New York University), Montreal (where Rabbat is based), and Singapore, to tap into Asian talent and potential clients. While immediate revenue generation isn’t the focus, AMI Labs intends to collaborate with customers early on to refine its models with real-world data. Nabla will be the first partner to access these early models, with further collaborations expected.
Open Source Commitment and Future Research
Staying true to LeCun’s principles, AMI Labs will prioritize open research and code. “We will also create a lot of code open source,” said LeBrun, who previously worked at Meta’s AI research laboratory, FAIR. “We think things move faster when they’re open, and it’s in our best interest to build a community and a research ecosystem around us.” This commitment to transparency and collaboration aims to accelerate progress in the field of world models.
The development of world models is a long-term undertaking, and AMI Labs acknowledges that commercial applications are still years away. However, the substantial funding and strong team position the company to make significant strides in this emerging area of AI. The company’s focus on understanding the underlying dynamics of the real world, combined with its commitment to open research, could pave the way for a new generation of AI systems with greater reliability and real-world impact.
What comes next for AMI Labs will be a period of intensive research and development, focused on building and refining its world models. The company’s progress will be closely watched by the AI community as it seeks to demonstrate the potential of this alternative approach to artificial intelligence. Share your thoughts on the future of world models in the comments below.
