Meta AI head Yann LeCun exits to launch a startup

LeCun plans to develop “world models” – AI systems that learn from videos and spatial data to understand the world more like humans do, the Financial Times reports. 

His decision comes after a period of internal reorganization at Meta’s AI division. In recent months, Meta restructured its efforts under a new Superintelligence Labs unit and changed reporting lines for senior AI leaders. LeCun, who had reported to Meta’s chief product officer, now reports to the head of the new unit. His exit follows the underwhelming performance of Llama 4 and Meta’s layoffs of about 600 employees from its AI research unit. 

Previously, LeCun criticized the idea that simply scaling large language models can lead to real reasoning. 

LeCun joined Meta (then Facebook) in 2013 to build the Facebook AI Research (FAIR) lab and later became chief AI scientist. He is widely known for pioneering convolutional neural networks and shared the 2018 ACM A.M. Turing Award with Geoffrey Hinton and Yoshua Bengio. 

Meta’s renewed push into AI has been tied closely to core products and advertising. In Q3 2025, Meta reported revenue of $51.24 billion, with advertising accounting for roughly 98% of income. The company pointed to AI tools for better targeting and engagement, noting ad impressions rose 14% year over year and the average price per ad increased 10%. To support AI workloads, Meta lifted 2025 capital-expenditure guidance to $70-$72 billion. 

In mid-2025, Meta ramped up its AI hiring by offering pay packages exceeding $2 million a year, with some top researchers receiving multi-million-dollar signing bonuses and total deals worth more than $100 million. The strategy helped Meta recruit senior AI talent from rivals, including Jason Wei and Hyung Won Chung, who were key contributors to core models and “chain-of-thought” reasoning at Google and OpenAI.