Paris, France – Advanced Machine Intelligence (AMI), a nascent artificial intelligence startup co-founded by AI luminary Yann LeCun, has announced a monumental funding round, raising over $1 billion to propel its ambitious vision of developing sophisticated AI "world models." The significant investment, which propels the company to a valuation of $3.5 billion, signals a powerful challenge to the prevailing paradigm in AI development, one largely dominated by large language models (LLMs). LeCun, formerly Meta’s chief AI scientist, asserts that true human-level intelligence will not emerge from simply scaling up LLMs, but rather from AI systems that possess a deep, grounded understanding of the physical world.
The funding round was co-led by prominent venture capital firms including Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions. A constellation of notable individual investors also participated, underscoring the broad confidence in AMI’s mission. Among them are entrepreneur and investor Mark Cuban, former Google CEO Eric Schmidt, and French billionaire Xavier Niel, a key figure in the European telecommunications sector. This substantial capital injection positions AMI to aggressively pursue its research and development goals, aiming to create a new generation of AI systems capable of nuanced understanding, persistent memory, robust reasoning, and safe, controllable operation.
LeCun’s departure from Meta in November 2025 marked the end of an era for the social media giant’s fundamental AI research efforts, led by his pioneering work at the Fundamental AI Research (FAIR) lab. However, it simultaneously heralded the dawn of AMI, his first commercial venture. The startup, whose name playfully echoes the French word for "friend," is designed with a global footprint from inception, establishing offices in key AI hubs including Paris, Montreal, Singapore, and New York. LeCun, who will continue his professorial role at New York University, plans to divide his time between academia and leading AMI, a testament to his dual commitment to foundational research and applied AI innovation.
The foundational principle behind AMI’s strategy is LeCun’s conviction that human intelligence is fundamentally rooted in our interaction with and understanding of the physical world, rather than being solely derived from linguistic patterns. This contrasts sharply with the current trajectory of many leading AI labs, such as OpenAI, Anthropic, and even LeCun’s former employer, Meta, which are heavily invested in the belief that by amplifying the scale and complexity of LLMs, they can eventually achieve artificial general intelligence (AGI) or even superintelligence. LeCun has been a vocal critic of this approach, famously stating in an interview with WIRED, "The idea that you’re going to extend the capabilities of LLMs [large language models] to the point that they’re going to have human-level intelligence is complete nonsense."
Challenging the LLM Dominance: The World Model Paradigm
LeCun’s skepticism towards LLMs is not an outright dismissal of their utility. He acknowledges their impressive capabilities, particularly in areas like code generation, and anticipates their continued usefulness across a wide spectrum of applications that benefit from such functionalities. However, he posits that these advancements, while significant, represent a technological leap within a specific domain and do not inherently lead to the kind of broad, adaptable intelligence characteristic of humans. "It’s true that [LLMs] are becoming really good at generating code, and it’s true that they are probably going to become even more useful in a wide area of applications where code generation can help," LeCun stated. "That’s a lot of applications, but it’s not going to lead to human-level intelligence at all."
The core of AMI’s technological ambition lies in developing "AI world models." These are AI systems designed to build internal, predictive representations of the physical world, enabling them to understand causality, infer consequences, and plan actions within a simulated or real environment. This approach is rooted in LeCun’s long-standing research at Meta, where he spearheaded the development of concepts like the Joint-Embedding Predictive Architecture (JEPA). JEPA, as explored in prior research, aims to imbue AI with a form of physical intuition by enabling it to predict future states of its environment.
LeCun’s decision to establish AMI as an independent entity stems from a strategic reevaluation of where his world model research could have the most impact. He observed a shift in Meta’s strategic priorities, which increasingly focused on catching up with the LLM frenzy, a direction that diverged from his core interest in world models. "There was a reorientation of Meta’s strategy where it had to basically catch up with the industry on LLMs and kind of do the same thing that other LLM companies are doing, which is not my interest," LeCun explained. He further elaborated that he presented his vision to Mark Zuckerberg, who, despite being supportive of world model research, was receptive to LeCun’s assessment that he could advance the technology more effectively and efficiently outside the corporate structure. "I told him I can do this faster, cheaper, and better outside of Meta. I can share the cost of development with other companies… His answer was, OK, we can work together." This collaborative understanding between LeCun and Meta’s leadership, including Zuckerberg, facilitated his transition and the launch of AMI.
Strategic Applications and Industry Partnerships
AMI’s business model is envisioned as a B2B enterprise, focusing on providing its advanced world model technology to industries that generate vast amounts of data and stand to benefit from enhanced AI capabilities. Sectors such as manufacturing, biomedical research, and robotics are prime targets. For instance, AMI could develop a highly accurate world model of an aircraft engine. This model could then be utilized by manufacturers for a range of critical applications, including optimizing fuel efficiency, minimizing emissions, and ensuring long-term reliability through predictive maintenance and anomaly detection. The ability of these AI systems to understand complex physical interactions and predict outcomes is seen as a significant leap beyond current analytical tools.
The cofounding team of AMI represents a formidable assembly of AI talent, many of whom have a history of working closely with LeCun. Beyond LeCun himself, the core group includes Michael Rabbat, former director of research science at Meta; Laurent Solly, former vice president of Europe for Meta; and Pascale Fung, former senior director of AI research at Meta. These individuals bring deep experience in AI research, development, and large-scale organizational management. Adding to this core leadership are Alexandre LeBrun, former CEO of the AI healthcare startup Nabla, who will lead AMI as its CEO, and Saining Xie, a former researcher at Google DeepMind, who will serve as the startup’s chief science officer. This blend of established AI researchers and experienced startup leaders creates a robust foundation for AMI’s ambitious undertaking.
The Broader Implications for the AI Landscape
The substantial funding secured by AMI and LeCun’s outspoken critique of the LLM paradigm have significant implications for the future direction of AI research and development. It represents a clear divergence in philosophical and technical approaches, potentially catalyzing a broader re-evaluation within the industry. If AMI’s world models prove successful in delivering on their promise of more grounded, reasoning AI, it could shift the focus away from purely linguistic models towards systems that exhibit a deeper understanding of causality and the physical world.
This development could also spur increased investment in alternative AI architectures and research methodologies. The success of LLMs has led to a concentration of resources and talent within a specific research track. AMI’s emergence, backed by significant capital, signals a viable alternative path, potentially fostering greater diversity in AI innovation. Furthermore, the emphasis on controllable and safe AI systems, a core tenet of AMI’s mission, addresses growing concerns about the ethical implications and potential risks associated with advanced AI.
The global presence planned for AMI also highlights the increasing internationalization of cutting-edge AI development. With offices strategically located across continents, the startup aims to tap into diverse talent pools and foster global collaboration. This international approach is crucial for building AI systems that can operate effectively and responsibly in a globalized world.
While the exact timeline for AMI’s product development and market entry remains to be detailed, the sheer scale of the funding and the caliber of its leadership team suggest a deliberate and well-resourced effort. The coming years will be critical in observing whether LeCun’s vision of AI world models can indeed unlock a new era of artificial intelligence, one that is more robust, more intuitive, and ultimately, more aligned with the complexities of the real world. The AI community, investors, and industry leaders will be closely watching AMI’s progress as it seeks to redefine the boundaries of machine intelligence.