Meta’s AI research division, known as FAIR (Fundamental AI Research), is witnessing a significant leadership exodus, marked by the recent resignation of Vice President Jitendra Malik. Malik, who has played a pivotal role in AI developments at Meta, will transition to lead Amazon’s robotics research in San Francisco. His departure follows that of Chief AI Scientist Yann LeCun, further emphasizing a troubling trend within FAIR as the unit grapples with internal turmoil and resource constraints amid Meta’s renewed focus on generative AI products.
Malik announced his departure on January 4, revealing that he would officially join Amazon on January 5. In a post on X, he reflected on his time at Meta, where he contributed to advancements in video action recognition models and tactile sensors, as well as the recent SAM 3D breakthrough in computer vision. “It is time to move on,” he stated, signaling his readiness for new challenges.
The ongoing transition at FAIR underscores a broader identity crisis within the unit, which was once regarded as Meta’s crown jewel in AI development. Since CEO Mark Zuckerberg initiated a “year of intensity” strategy in January 2025, the lab has undergone multiple organizational restructurings, faced budget constraints, and witnessed more than half of the original authors of the Llama research paper depart shortly after its publication. Former employees have reported that Meta has increasingly deprioritized open-ended research in favor of product-driven initiatives under the GenAI organization, leaving FAIR with diminished computing resources compared to teams focused on generative AI.
In August, Meta consolidated its AI efforts by splitting the Superintelligence Labs into four teams, a move that created confusion over project ownership and team assignments. This restructuring was accompanied by 600 job cuts in October, with Chief AI Officer Alexandr Wang asserting that layoffs were essential for expediting decision-making processes. The continuous reshuffling has contributed to a perception of instability within the division.
Further complicating matters, LeCun’s departure highlighted internal tensions as he pursued an alternative approach focused on “world models,” diverging from the company’s prevailing emphasis on large language models (LLMs). In a December interview with the Financial Times, LeCun acknowledged that results for Meta’s Llama 4 were “fudged a little bit” and criticized the company’s strategy as being overly reliant on LLMs, which he argued was incompatible with aspirations for superintelligence.
Former research scientist Tian Yuandong, who left Meta in November, indicated that conflicts arose within the organization as computing resources became increasingly scarce in an industry driven by LLM advancements. Another ex-employee, Joena Zhang, expressed frustration, noting a lack of clarity regarding project objectives during the initial months of 2025 at what is now referred to as Meta Superintelligence Labs. Despite these challenges, Meta has asserted its commitment to FAIR, and recent employee sentiment surveys have indicated a slight uptick in optimism, reported at 80% by late 2025.
The exodus of key personnel and the internal strife at FAIR serve as a cautionary tale for organizations navigating the rapidly evolving AI landscape. As Meta pivots toward commercial AI products, the implications of these shifts extend beyond internal dynamics, potentially affecting its competitive positioning in the broader tech ecosystem. The decisions made in this transitional phase could shape not only the future direction of Meta’s AI initiatives but also the overall trajectory of AI research and development in the industry.
See also
Tuba Joins NVIDIA Inception to Transform AI-Driven Healthcare Benefits Management
Gold Surges 2% and Dollar Strengthens as AI-Driven Tech Stocks Soar Amid Venezuela Turmoil
Elon Musk Shares Essential Tips for Crafting AI Images with Grok Imagine
U.S.-China AI Dynamics: From Rivalry to Cooperative Governance for Global Stability
Good Tokens Expands AI and Blockchain Solutions to Enhance Global Humanitarian Impact
