Legacy Teams Hit as Superintelligence Labs Grows
Meta has laid off around 600 employees from its artificial intelligence unit as part of a broader effort to streamline operations and accelerate its AI strategy. The cuts were confirmed Wednesday and were outlined in a memo by Chief AI Officer Alexandr Wang, who joined Meta in June following the company’s $14.3 billion investment in Scale AI.
The layoffs target workers across several AI infrastructure groups, including Meta’s long-established FAIR (Fundamental Artificial Intelligence Research) unit and other product-oriented teams. However, employees within TBD Labs — a new team staffed by recent elite AI recruits — were unaffected, signaling CEO Mark Zuckerberg’s confidence in newer talent over legacy teams.
Power Shift to Superintelligence Labs
According to insiders, the internal structure of Meta’s AI division had become bloated, with FAIR and product teams often competing for limited compute resources. Following the formation of Meta Superintelligence Labs, those inefficiencies became more pronounced.
The layoffs are part of an internal restructuring to solidify Alexandr Wang’s authority over Meta’s AI direction. After the creation of Superintelligence Labs, Meta’s AI workforce now stands just under 3,000 people. Affected employees were placed on “non-working notice period” until their termination date on November 21, with access revoked and no further duties required.
Compensation and Developer Discontent
Impacted workers are being offered 16 weeks of severance plus an additional two weeks for each full year of service, minus the notice period. The move comes amid internal tension following the lukewarm reception to Meta’s Llama 4 models released in April.
Zuckerberg, reportedly dissatisfied with the company’s AI performance, has doubled down on external hires and large-scale projects. His frustration has shaped a more aggressive approach to AI development and spending. The shift reflects Meta’s desire to stay competitive with rivals like OpenAI and Google in the AI arms race.
Hyperion and Escalating AI Spend
Meta continues to scale its AI ambitions through infrastructure and partnerships. On Tuesday, the company announced a $27 billion deal with Blue Owl Capital to build its Hyperion data center in rural Louisiana. Expected to rival Manhattan in size, Hyperion aims to provide the necessary computing power for Meta’s expanding AI capabilities.
During its Q2 earnings call, Meta projected total expenses of up to $118 billion for 2025, with even higher expenditures forecasted for 2026 due to AI-related costs. With strategic consolidation and massive infrastructure bets, Meta is clearly positioning itself as a long-term AI powerhouse, even if that means shedding hundreds of roles in the process.
