Meta has conducted a new AI reorganization that eliminated roughly 600 positions across research, product, and infrastructure groups, according to reporting from Axios. In a memo attributed to Chief AI Officer Alexandr Wang, the company framed the reductions as a way to create smaller, faster-moving teams with fewer approvals and clearer scope, while encouraging affected employees to apply for other open roles within Meta. The move lands after a summer of aggressive hiring for Meta’s superintelligence push and follows weeks of tension over publication policies and research priorities.

Axios says the cuts touched the long-running FAIR (Fundamental AI Research) organization along with some product and infra pods, but spared Meta’s superintelligence group, TBD Lab, which Wang oversees directly. In parallel, Meta has continued recruiting from rivals, recently hiring OpenAI scientist Ananya Kumar and TML co-founder Andrew Tulloch, even as some legacy researchers—including AI pioneer Yann LeCun—voiced concerns about new review requirements for publications and code releases.

• Rationale: Management says smaller teams reduce coordination overhead and enable faster iteration on frontier model work.
• Scope: Impact reported across FAIR, product, and infra; TBD Lab remains intact to pursue Meta’s superintelligence agenda.
• Talent motion: Ongoing recruiting from external labs suggests Meta is rebalancing toward agentic systems and large-scale training programs.

Why it matters: The shake-up underscores a broader industry shift from open research to productized, security-reviewed frontier work. By protecting TBD Lab while trimming elsewhere, Meta signals that superintelligence efforts are a top strategic bet, even if that creates friction with parts of the traditional research culture. For researchers and partners, the reorg hints at tighter information controls, faster decision cycles, and a clearer hierarchy around who sets model roadmaps inside Meta.

Source: Axios — Meta superintelligence & AI reorg.