Industry speculates him launching a New AI Startup Focused on Next-Generation “World Model” Intelligence.
Yann LeCun, Meta’s Chief AI Scientist and one of the most influential figures in the global artificial intelligence community, is reportedly preparing to exit the company to establish a new AI startup, according to a report by the Financial Times. The move comes during a period of major restructuring within Meta’s AI strategy, led by CEO Mark Zuckerberg, who recently formed the Meta Superintelligence Labs to accelerate development of large-scale AI systems.
LeCun, a Turing Award winner and a professor at New York University, has been at the forefront of AI research for decades. He has led Meta’s Fundamental AI Research (FAIR) Lab since 2013, playing a key role in advancing deep learning, neural networks, and machine learning breakthroughs. However, recent organizational changes at Meta have shifted internal priorities away from long-term foundational research, creating strategic differences.
Reports suggest that LeCun has informed colleagues of his plans to depart in the coming months and is already in early talks to raise funding for his new AI venture. His vision focuses on developing next-generation AI systems capable of human-level reasoning, known as “world models.” Unlike traditional Large Language Models (LLMs) used by systems such as ChatGPT and Google Gemini, world models aim to learn from real-world video, spatial data, and sensory interaction, enabling machines to understand physical environments rather than only text.
Meanwhile, Meta has been consolidating leadership in its AI division. Earlier this year, Zuckerberg appointed Alexandr Wang, former CEO of Scale AI, to lead Meta Superintelligence Labs. This restructure resulted in LeCun shifting his reporting line within the company — a change seen by analysts as reflective of the evolving direction of Meta’s AI roadmap.
LeCun has openly questioned the industry’s focus on scaling LLMs, arguing that true artificial general intelligence (AGI) will require breakthroughs in world model architecture, a vision he now appears ready to pursue independently. His upcoming startup is expected to compete in the rapidly expanding global AI innovation landscape, where OpenAI, Google DeepMind, Anthropic, xAI, and Meta are all racing to define the future of intelligent systems.
