Meta's Chief AI Scientist LeCun is reported to be leaving to start a business, with disagreements over the "superintelligent" vision with Zuckerberg

Meta is undergoing another significant personnel change in the process of reshaping its AI strategy.

According to the latest report from the Financial Times, Meta's Chief AI Scientist, one of the three giants of deep learning and Turing Award winner Yann LeCun plans to leave the social media giant in the coming months to establish his own startup.

The report states that, according to insiders, LeCun has informed colleagues that he will be leaving Meta in the next few months and is currently in early negotiations for funding his new venture. The departure of this French-American scientist, regarded as one of the pioneers of modern AI, highlights fundamental differences with Mark Zuckerberg regarding the path of AI development.

LeCun advocates for the development of a new generation of AI systems called "world models," believing that the large models that Zuckerberg is heavily investing in, while "useful," will never be able to reason and plan like humans, increasingly conflicting with Meta's current strategy centered around LLMs. This departure comes at a time when Meta is under pressure in the AI competition, with its Llama 4 model performing poorly compared to OpenAI and Anthropic.

Intensifying Strategic Discrepancies: Long-term Research vs. Rapid Productization

Since 2013, LeCun has led Meta's Fundamental AI Research Lab (Fair), focusing on foundational AI research. He advocates for the development of a new generation of AI systems he calls "world models," which understand the physical world through video and spatial data, aiming to achieve machines with human-level intelligence.

He believes that the current AI approach centered around large language models, while "useful," will never be able to reason and plan like humans, a view that is clearly at odds with Zuckerberg's strategy of emphasizing rapid productization and "super-intelligent" teams.

In response to the poor performance of the Llama 4 model and the lukewarm market response to Meta's AI chatbots, Zuckerberg decided to accelerate AI product iterations and reduce investments in long-term foundational research.

To this end, this summer, Zuckerberg spent $14.3 billion to hire 28-year-old Alexandr Wang to lead Meta's new "super-intelligent" team and acquired a 49% stake in Wang's data labeling startup Scale AI. This restructuring means that LeCun, who previously reported to Chief Product Officer Chris Cox, now needs to report to Alexandr Wang.

Zuckerberg also personally assembled a dedicated team called TBD Lab to advance the next generation of large language model development, poaching talent from competitors like OpenAI and Google with a $100 million compensation package.

Reports cite two insiders stating that LeCun's newly established company will focus on advancing his work on world models and has begun raising funds for the new project.

Personnel Turmoil vs. Cost Pressures

LeCun's departure is not an isolated incident. Since 2024, Meta's executive team has been in continuous turmoil, with AI Research Vice President Joelle Pineau having left and approximately 600 people laid off from the AI research department At the same time, Zuckerberg is attracting a new generation of AI leaders with a salary of hundreds of millions of dollars, including Scale AI founder Alexandr Wang and OpenAI ChatGPT co-creator Shengjia Zhao, who has now become the Chief Scientist of Meta's "Super Intelligence Lab."

This series of personnel changes reflects Meta's dramatic shift in AI strategy. Zuckerberg is facing increasing pressure from Wall Street to prove that his billions of dollars in investments to become an "AI leader" will yield returns and drive revenue growth