AMI Labs, co-founded by Turing Award winner Yann LeCun after departing Meta, has secured $1.03 billion at a $3.5 billion pre-money valuation. The company focuses on world models, AI that learns from reality instead of just language.
While this field currently has fewer players than generative AI, it might soon become more popular. “I predict ‘world models’ will become the next buzzword,” stated AMI Labs CEO Alexandre LeBrun. “In six months, every company will label itself a world model to attract funding.”
LeBrun emphasizes that AMI Labs is distinct due to its mission to comprehend the real world, with potential applications in healthcare. AMI Labs’ first partner will be Nabla, a digital health startup, and LeBrun is its chairman.
LeBrun and LeCun both recognized the limitations of large language models (LLMs) such as hallucination issues, which can be hazardous. However, developing an alternative based on LEcun’s JEPA (Joint Embedding Predictive Architecture) may take time.
“AMI Labs is an ambitious endeavor, starting with foundational research. Unlike typical applied AI startups that launch products quickly for revenue, world models might take years to shift from theory to application,” LeBrun explained.
Despite the long development timeline, world model developers are securing substantial funding. SpAItial obtained a $13 million seed round, significant for a European startup, and Fei-Fei Li’s World Labs received $1 billion recently. Now, AMI Labs has also garnered more funds than rumored.
Initially seeking €500 million last December, AMI Labs raised approximately €890 million. This success is due to its team, including LeCun as chairman and LeBrun’s entrepreneurial experience. The company also has Meta’s VP for Europe, Laurent Solly, as COO, and researchers Saining Xie, Pascale Fung, and Michael Rabbat.
High interest allowed AMI Labs to select investors based on aligned expectations. The round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, with further backing from several funds, industry backers, and individuals like Tim and Rosemary Berners-Lee, Jim Breyer, Mark Cuban, Mark Leslie, Xavier Niel, and Eric Schmidt.
This funding provides AMI Labs with significant resources for two main expenses: compute and talent. LeBrun aims for quality over quantity while building AMI Labs’ team in Paris, New York, Montreal, and Singapore to recruit AI talent and connect with potential clients in Asia.
Though AMI Labs isn’t planning to generate revenue soon, it will engage with prospective customers early on. “Developing world models requires real-world testing,” LeBrun remarked.
AMI Labs will eventually collaborate with partners, with Nabla being the first disclosed. This partnership hints at the industrial interest in the investment round.
Alongside lead investors and angels, AMI Labs is supported by NVIDIA, Samsung, Sea, Temasek, Toyota Ventures, and French entities like Association Familiale Mulliez, Groupe Industriel Marcel Dassault, and Publicis Groupe. Other contributors include Aglaé Lab, Alpha Intelligence Capital, Artémis, Bpifrance Digital Venture, New Legacy Ventures, SBVA, and ZEBOX Ventures.
Though commercial applications may take time, AMI Labs is committed to LeCun’s philosophy by publishing papers and making code open source. “Open research accelerates progress,” LeBrun emphasized, highlighting the founders’ belief in building a collaborative community and research ecosystem.