Meta’s bold move to create a dedicated division for artificial superintelligence has hit early turbulence, as several senior AI researchers and engineers have exited the company. The departures come just months after CEO Mark Zuckerberg unveiled the Meta Superintelligence Lab (MSL), a high-stakes initiative aimed at developing “personal superintelligence” for global users.
Talent Losses Despite Heavy Investment
Meta has poured billions into the MSL program, offering lucrative compensation packages to attract top minds from OpenAI, Google DeepMind, and other AI firms. However, despite these incentives, attrition is already visible.
According to reports, at least eight notable employees have recently left Meta, including veterans who played key roles in shaping its AI ecosystem.
Key Departures from Meta
- Bert Maher (12 years at Meta): Instrumental in building PyTorch and Triton. Joining Anthropic.
- Tony Liu (8 years): Managed PyTorch GPU systems teams.
- Chi-Hao Wu (5 years): AI and ML specialist, now Chief AI Officer at Memories.ai.
- Aram Markosyan (4+ years): Focused on AI safety and fairness, including smart glasses.
- Chaya Nayak (9 years): Former Director of Product Management for generative AI and Llama models, joining OpenAI.
- Afroz Mohiuddin (1 year): Ex-Google veteran, moving to OpenAI’s technical staff.
- Avi Verma & Ethan Knight: Left MSL after less than a month, returning to OpenAI.
- Rishabh Agarwal: Former DeepMind researcher, now with Periodic Labs.
- Ruben Mayer: Ex-Scale AI executive, departed Meta after just two months.
Internal Strains at Meta
Reports suggest that the lavish pay packages for MSL recruits created tensions among long-time Meta employees who felt sidelined. While Meta downplayed the exits, calling them “normal attrition,” the departures raise questions about the stability of its superintelligence bet.
In June, Meta announced a $14.3 billion investment in data-labelling firm Scale AI, with former Scale CEO Alexandr Wang brought in to co-lead MSL. The early exits of both former Scale executives and newly recruited AI talent underscore the challenges Meta faces in building its next-generation AI infrastructure.
The Road Ahead
Despite setbacks, Zuckerberg remains committed to the vision of building personal superintelligence for billions of users. Yet, with rivals like OpenAI, Anthropic, and emerging startups continuing to lure away top researchers, Meta’s ability to hold its talent pool will be critical in determining whether its superintelligence ambitions can gain momentum or falter.
