Yann LeCun, a leading figure in artificial intelligence and former head of AI research at Meta Platforms, is embarking on a groundbreaking venture that aims to shift the focus of AI development away from the current chatbot craze. With a fresh startup named Advanced Machine Intelligence (AMI), LeCun has successfully amassed approximately $1.03 billion in funding to pursue a radically different vision for the future of AI technology.
Over the last several years, the technology sector has been captivated by large language models (LLMs), the sophisticated engines behind chatbots capable of generating human-like text and answering complex queries. These models have dominated headlines and investment portfolios alike, promising to revolutionize communication and information access. However, LeCun has consistently voiced skepticism about the true capabilities of these systems, emphasizing that while they excel at predicting the next word in a sentence based on vast data patterns, they lack genuine understanding of the world around them.
In contrast to the prevailing trend, AMI is dedicated to creating what researchers refer to as “world models.” These AI frameworks aim to grasp the underlying mechanics of the physical environment, including spatial awareness, cause-and-effect relationships, and the progression of events over time. This approach is designed to enable machines to reason more deeply, plan effectively, and make decisions with greater reliability—qualities that current chatbot models struggle to achieve.
LeCun’s innovative research methodology, known as Joint Embedding Predictive Architecture (JEPA), underpins this vision. JEPA focuses on training AI to recognize and learn from patterns and interactions in the real world, rather than merely processing and regurgitating text. This technology holds significant promise for sectors where errors can have serious consequences, such as manufacturing, robotics, healthcare, and biomedical research. In these fields, the risk of AI providing plausible but incorrect information—an issue often seen in chatbot outputs—could lead to costly or even dangerous outcomes.
After dedicating over ten years to building Meta’s AI research division, LeCun decided to step away in late 2025. During his tenure, the industry witnessed an intense competition to develop increasingly larger and more powerful chatbots. Yet, LeCun felt this direction was not aligned with his vision for truly intelligent machines. He believed that by operating independently, he could advance his ideas more swiftly, efficiently, and effectively, which motivated the founding of AMI.
Leading the startup as CEO is Alexandre LeBrun, who has emphasized that AMI is not aiming for rapid product launches typical of many AI startups eager to capture market share. Instead, the company is committed to investing time and resources into foundational technology development, even if it requires years before tangible products emerge. This patient, research-driven approach reflects a long-term commitment to reshaping the AI landscape.
AMI’s mission aligns with a broader shift in AI research, as several groups and companies have recently begun exploring AI systems that comprehend the physical world rather than focusing solely on language processing. For instance, Fei-Fei Li, a renowned computer scientist, has raised a similar amount—around $1 billion—for World Labs, a startup dedicated to advancing “spatial intelligence,” which enables AI to interpret and generate three-dimensional environments.
Major technology corporations are also recognizing the potential of world models. Nvidia, for example, has highlighted their importance in training robots and autonomous vehicles by simulating real-world scenarios, which can enhance safety and performance. This growing interest signals a possible paradigm shift in how AI is developed and applied across industries.
The impressive funding round for AMI attracted a diverse group of high-profile investors, including tech giants like Nvidia and Samsung, automotive innovator Toyota Ventures, and investment firm Temasek Holdings. Additionally, prominent individual backers such as entrepreneur Mark Cuban and former Google CEO Eric Schmidt have thrown their support behind the startup. Headquartered in Paris, AMI plans to expand its presence with offices in New York, Montreal, and Singapore, positioning itself as a global player in the AI innovation space.
Despite the substantial financial backing, AMI currently has no commercial products available and does not anticipate immediate market entry. The company’s strategy hinges on the belief that the current chatbot-focused trajectory may not ultimately yield truly intelligent machines. If LeCun’s vision proves accurate, the next transformative wave of artificial intelligence will emerge from systems that possess a genuine understanding of the world itself, transcending mere text generation.
The technology community and industry watchers alike will be closely monitoring AMI’s progress, eager to see whether this ambitious gamble will revolutionize AI or become an expensive experiment. Either way, Yann LeCun’s new venture represents a bold challenge to the status quo, potentially redefining the future direction of artificial intelligence research and application.
