Meta’s Superintelligence Lab Hit by Wave of Exits Within Months of its Launch
Just two months after CEO Mark Zuckerberg opened Meta’s Superintelligence Lab, the new division is already facing serious growing pains. Now a wave of high-profile exits, among fresh recruits and long-tenured employees, is sparking questions about the company’s AI strategy and its ability to keep hold of its talent.
In recent weeks, at least eight staff members have left the unit. That includes three AI researchers who had just recently joined. Avi Verma and Ethan Knight, who were at Meta for less than a month, have already gone back to OpenAI. Rishabh Agarwal, hired in April to work with generative AI projects before joining the lab, also said he was leaving.
The departures are not limited to new hires. A few Meta veterans have also left. Bert Maher left to join Anthropic after 12 years at the company. Tony Liu, the manager who spent eight years managing PyTorch GPU systems, has departed to start his own AI-focused newsletter. Chi-Hao Wu, an AI researcher for five years, is now chief AI officer at startup Memories.ai.
Among the most prominent departures is Chaya Nayak, Meta’s director of generative AI product management. She is moving to OpenAI to work on special projects after more than 10 years with the company. Nayak was instrumental in creating Meta’s Llama models and had previously overseen projects around election integrity, disaster response, and data transparency. In a LinkedIn post, she called her time at Meta “a bold experiment that grew into the bedrock of my career.”
The departures underscore the intense competition for AI expertise. OpenAI also seems to be a big winner, with several former Meta employees returning to the company. Even dumping Meta’s huge pay packages—often in the nine-figure range—the company appears unable to keep hold of key talent.
Meta spokesperson Dave Arnold downplayed it, telling me, “In a competitive recruiting environment, some people will ultimately make the decision to stay in their current job rather than take a new one. That’s normal.” But the continual departures indicate more profound structural problems.
Reports suggest that Meta has struggled with bureaucracy and repeated restructuring in its AI divisions. Employees have been shuffled among different groups multiple times, making it difficult to bring stability to the teams.
The Superintelligence Lab was Zuckerberg’s attempt to push Meta into the AGI race. But early departures have led to questions about the pace at which the company is able to move. With competitors such as OpenAI and Anthropic luring talent from Meta, the big social media company has a long fight on its hands when it comes to retaining top researchers and advancing its AI ambitions.
