NextFin News - Yann LeCun, the Turing Prize-winning pioneer who spent a decade shaping Meta’s artificial intelligence strategy, has secured $1.03 billion in funding for his new venture, AMI Labs, marking a definitive shift in the industry’s pursuit of "world models." The capital injection, announced on March 9, 2026, values the Paris-based startup at $3.5 billion pre-money and signals a massive institutional bet that the current era of Large Language Models (LLMs) has hit a wall. Co-led by Cathay Innovation, Greycroft, and Bezos Expeditions, the round also drew support from a roster of tech royalty including Eric Schmidt and Mark Cuban, underscoring the urgency with which investors are seeking an alternative to the hallucination-prone architectures of the mid-2020s.
The sheer scale of the round—raised just months after LeCun’s departure from Meta—reflects a growing consensus that predicting the next word in a sentence is insufficient for achieving true machine intelligence. While OpenAI and Google have doubled down on scaling transformer-based models, AMI Labs is betting on Joint Embedding Predictive Architecture (JEPA). This approach seeks to mimic how humans and animals learn: by observing the physical world and building internal representations of cause and effect, rather than merely ingesting the internet’s worth of text. CEO Alexandre LeBrun, who previously led Facebook’s Wit.ai, noted that while LLMs are "stuck in a text box," world models are designed to understand the high-dimensional reality of video, sensor data, and physical interaction.
This funding milestone places AMI Labs in a direct, high-stakes rivalry with Fei-Fei Li’s World Labs, which closed its own $1 billion round only last month. The emergence of two billion-dollar "world model" startups within thirty days suggests a structural pivot in the venture capital landscape. Investors are no longer satisfied with incremental improvements to chatbots; they are now financing the infrastructure for "spatial intelligence" and autonomous reasoning. For AMI Labs, the immediate priority is not a consumer product but fundamental research and high-stakes applications in healthcare and robotics. The company has already tapped Nabla, a digital health startup, as its first partner to test whether JEPA-based models can eliminate the life-threatening inaccuracies common in generative AI.
The geographical footprint of the deal is equally telling. Headquartered in Paris with satellite hubs in New York, Montreal, and Singapore, AMI Labs is positioning itself as a global counterweight to the Silicon Valley monoculture. By securing backing from Singapore’s Temasek and Sea, the firm is ensuring it has the geopolitical and financial runway to compete for the world’s most expensive commodity: top-tier research talent. LeCun has long argued that the next breakthroughs will come from academia and specialized labs rather than the industrial giants, and with over $1 billion in the bank, he now has the resources to prove that a contrarian architecture can outpace the brute-force scaling of his former employers.
The economic reality of this bet is daunting. Unlike the "wrapper" startups of 2023 that generated revenue within weeks, AMI Labs is a deep-tech play with a multi-year horizon. The capital will be consumed primarily by massive compute requirements and a payroll that includes Meta’s former VP for Europe, Laurent Solly, and chief science officer Saining Xie. As the industry moves away from the "generative" buzzword toward "world models," the success of AMI Labs will likely determine whether the next decade of AI belongs to those who can predict the next token, or those who can finally understand the world in which those tokens exist.
Explore more exclusive insights at nextfin.ai.
