Meta cuts about 600 jobs in AI division as focus shifts to Superintelligence Labs

Meta cut roughly 600 roles from its Superintelligence Labs AI organisation, concentrated in the Fundamental AI Research (FAIR) group along with product-related AI and AI-infrastructure teams. The cuts were announced internally in a memo from chief AI officer Alexandr Wang, who had joined that June as part of Meta’s roughly $14.3 billion investment in Scale AI.

Wang’s memo argued for a smaller, faster-moving organisation, saying fewer people would mean “fewer conversations” required to make a decision and more scope and impact for those who remained. The cuts left Superintelligence Labs’ total headcount at just under 3,000. They spared TBD Lab, the small internal team — a few dozen researchers and engineers — tasked with building Meta’s next generation of frontier foundation models, underscoring that Meta was concentrating resources on a small, highly paid core rather than a broad research organisation. FAIR, once the standard-bearer of Meta’s open AI research under Yann LeCun, absorbed a substantial share of the reduction; the report that first surfaced the cuts noted uncertainty over how deeply they would cut into FAIR’s Tel Aviv research staff specifically.

The reduction followed a summer of aggressive AI hiring at Meta, including a wave of reported nine-figure compensation packages to recruit researchers from OpenAI, Google DeepMind and elsewhere, and came alongside Meta’s departure of former FAIR head Joelle Pineau earlier in the year (she subsequently became Cohere’s chief AI officer). It illustrated a pattern that recurred across the industry in 2025: labs cutting broad research staff even as they bid up pay for a small number of researchers seen as capable of frontier-level work.