Meta’s Superintelligence Team Faces Exits Amid Talent Wars and Internal Turmoil

Meta’s ambitious push toward building “superintelligence” is facing early challenges, as at least eight employees—including researchers, engineers, and a senior product leader—have left the company within two months of the launch of Meta Superintelligence Labs (MSL). The departures come amid a broader restructuring and intense hiring effort aimed at catching up with rivals like OpenAI, Google, and Anthropic. The exodus includes several long-time veterans who played key roles in developing Meta’s foundational AI infrastructure. Bert Maher, who spent 12 years at Meta, left to join Anthropic. At Meta, Maher was instrumental in creating PyTorch, one of the most widely used frameworks for AI development, and Triton, a language and compiler that improves AI model efficiency. He later announced his move on X, expressing excitement about helping make Anthropic’s Claude AI faster. Tony Liu, who had been at Meta for over eight years and led teams focused on PyTorch GPU systems, also departed. In a LinkedIn post, Liu said he would launch a newsletter offering insights into scaling AI systems. He did not respond to requests for comment. Chi-Hao Wu, an AI and machine learning specialist who worked at Meta for more than five years, left to become chief AI officer at Memories.ai, a startup focused on AI-driven video and visual analysis. Wu told Business Insider that instability caused by frequent reorganizations contributed to his decision, noting that some team members felt the pace of change was overwhelming. Meta has restructured its AI division multiple times this year, including dissolving one team and creating four new ones, according to an internal memo. While Meta’s communications chief Andy Stone dismissed media coverage of the changes as “navel gazing,” internal tensions appear to be growing. Other notable exits include Aram Markosyan, a research scientist who worked on AI safety and Meta’s smart glasses, and Chaya Nayak, a nine-year veteran who led product management for Meta’s generative AI initiatives, including the Llama models. She has now joined OpenAI to work on special projects. Afroz Mohiuddin, a senior staff engineer who came to Meta from Google after more than 14 years, also left to join OpenAI’s technical team. Meanwhile, two recent hires—Avi Verma and Ethan Knight—returned to OpenAI after less than a month at MSL, with Verma reportedly leaving before his start date. Rishabh Agarwal, who joined Meta from Google DeepMind in April, left after just five months. He cited a desire to take on a different kind of risk and is now joining Periodic Labs, a new AI startup focused on discovering new physical materials using artificial intelligence. Meta said that some turnover is normal, especially during periods of rapid growth and recruitment. The company has offered high compensation packages to attract top talent, including some from OpenAI and Google DeepMind. However, the influx of new hires and shifting priorities appear to have strained morale among long-term employees, raising questions about the sustainability of Meta’s aggressive AI expansion.