Meta Lays Off 600 From AI Division as Wang Pushes Leaner Teams

Meta cut 600 AI jobs on Oct 22 but spared TBD Labs' elite recruits. Affected workers get severance through Nov 21.

Meta fired 600 people from its AI division on October 22, sending termination notices that set November 21 as the last official day. Chief AI Officer Alexandr Wang broke the news through an internal memo, explaining the cuts would make teams smaller and decisions faster. Affected employees lost system access immediately but stay on payroll through November 21, creating an awkward month where they're paid to do nothing.​

The layoffs hit Meta's legacy AI teams - the Fundamental AI Research unit (FAIR), infrastructure groups, and various product teams. What didn't get touched was TBD Labs, the new elite division packed with expensive recruits hired just months ago from OpenAI, Google, and Apple. Meta spent hundreds of millions bringing in top AI talent earlier this year. Now the company is cutting longtime employees while protecting these new hires who command massive salaries. The message is clear - Zuckerberg values fresh AI expertise over people who've been there for years.​

This creates an uncomfortable dynamic inside Meta. Imagine working on AI research for five years, only to watch someone hired last month keep their job while yours disappears. The decision reveals how tech companies think about talent now. They don't just want good engineers. They want the specific people who worked at OpenAI or Google DeepMind, believing they carry special knowledge worth the premium price.​

Wang's memo said smaller teams enable faster decisions. He argued that fewer people means less coordination and more individual impact. The reasoning follows what tech companies have been saying for two years now - we hired too many people during the pandemic and need to trim down. Meta isn't alone here. Amazon, Google, and Microsoft all cut staff recently while simultaneously pouring billions into AI. The contradiction is striking.​

Here's what's interesting about the timing. Meta invested $14.3 billion in Scale AI back in June, Wang's own company. Wang then joined Meta as Chief AI Officer. Now he's cutting 600 people from the division he just took over. Some sources told CNBC that Meta's AI unit had become "bloated," with different teams competing for computing resources instead of working together. When Meta formed its Superintelligence Labs by merging existing groups, the overlap became obvious.​

Affected employees enter what Meta calls a "non-working notice period" until November 21. Their access gets revoked immediately even though they remain on payroll. They can apply for other positions within Meta during this month, but that's tough when you can't access systems or show recent work. The severance package includes 16 weeks of base pay plus two weeks for every year worked. It's generous compared to most companies, but doesn't ease the sudden shock.​

The situation is particularly difficult for H-1B visa holders. Indian tech workers on temporary work visas need to find new jobs quickly or face deportation. One laid-off Meta employee shared her situation on LinkedIn, immediately receiving multiple offers from startups. Sudarshan Kamath, who runs a San Francisco AI startup called Smallest AI, publicly invited Meta's fired workers to apply, offering $200,000 to $600,000 base salaries plus equity. His pitch was aggressive - he wants people who are "smart and hungry".​

This recruitment scramble shows how Meta's loss becomes smaller companies' gain. Startups can't normally poach Big Tech employees. But layoffs create windows where experienced engineers suddenly become available and motivated to move. The irony is thick - Meta declares its AI unit overstaffed while startups compete to hire the same people as fast as possible. This suggests the problem wasn't too many AI workers in the industry. The problem was too many at one company competing for the same resources.​

Meta also announced separate organizational changes this week. An internal memo said "many routine decisions can now be handled efficiently by technology," suggesting AI is replacing human roles beyond just development teams. The company is consolidating London operations and merging its privacy and regulatory compliance divisions. These changes point to AI automation replacing mid-level roles across different functions.​

Zuckerberg told employees earlier this year that Meta expects AI agents capable of performing midlevel engineering work by late 2025. The company already uses AI for hiring processes, conducting automated interviews and assessing coding skills. Amazon CEO Andy Jassy made similar predictions in June, telling staff that generative AI will eventually reduce headcount across various departments. Meta's layoffs accelerate this vision where AI doesn't just help humans work - it replaces them entirely.​

The cuts reveal contradictions in Meta's AI strategy. The company fires 600 AI workers while betting billions that AI represents the future. It protects expensive new hires while eliminating longer-tenured employees who built the foundation. It preaches about faster decision-making while maintaining a division approaching 3,000 people. After the cuts, Meta's Superintelligence Labs still employs just under 3,000 workers - hardly a skeleton crew.​

Meta's third-quarter earnings report comes next week. The company already warned that 2026 expenses will grow faster than 2025 because of AI initiatives. Earlier this week, Meta announced a $27 billion deal with Blue Owl Capital to fund its massive Hyperion data center in Louisiana. Zuckerberg said the facility would cover a significant portion of Manhattan's footprint. So while the company cuts people to save money, it simultaneously spends billions on infrastructure. The priorities are clear - invest in machines, reduce headcount.