On Wednesday 20 May, Meta began notifying employees of layoffs affecting roughly 10% of its global workforce — approximately 8,000 people. Workers in North America were instructed to work from home, with notification emails scheduled for around 7am, after which system access would be immediately revoked.

The cuts are only part of the picture. Meta's Chief People Officer, Janelle Gale, confirmed in an internal memo that 7,000 employees would be transferred into four newly created AI-focused units: Applied AI Engineering (AAI), Agent Transformation Accelerator (ATA) XFN, Central Analytics, and a fourth unit, Enterprise Solutions, still to be detailed. In addition, Meta has closed 6,000 previously open roles and further reductions are expected later in the year — meaning the restructure will ultimately affect around 20% of the entire workforce.

The two AI units publicly announced by CTO Andrew Bosworth under Meta's "AI for Work" initiative are designed to develop autonomous AI agents capable of performing workplace tasks — effectively replacing the functions carried out by entire categories of human roles.

The restructure is not merely a headcount reduction — it is a fundamental reimagining of how Meta operates. The company is moving away from large, hierarchical teams toward small "pods" or cohorts with fewer management layers. According to Gale's memo, many of Meta's organisational leaders were explicitly directed to embed AI-native design principles into their new structures from the ground up.

CEO Mark Zuckerberg has publicly stated that 2026 will be "the year that AI dramatically changes the way we work," noting that projects previously requiring large teams can now be accomplished by a single, highly talented individual. This belief is being operationalised through the restructure — management layers that once coordinated those large teams are being eliminated as redundant.

The move also reflects Meta's growing reliance on AI tools to drive engineering productivity. CFO Susan Li noted during the company's Q1 2026 earnings call that AI tooling is increasingly central to Meta's output. The company has raised its 2026 capital expenditure guidance to between $125bn and $145bn — up from an earlier estimate of $115bn–$135bn — citing higher component costs and expanded data centre commitments tied to AI infrastructure.

The workforce transformation sits within a far larger strategic pivot. Meta committed approximately $60–65 billion in capital expenditure in 2025, primarily directed at AI compute and data centres. It launched Meta Superintelligence Labs (MSL) and brought in Alexandr Wang, founder of Scale AI — in a deal reportedly valued at $14.3 billion — to lead its push toward more advanced reasoning systems.

Alongside hardware, Meta has doubled down on open-source AI through its LLaMA model family. With LLaMA 4, the company is positioning itself not as a product competing directly with ChatGPT, but as AI infrastructure for the broader developer ecosystem — a strategy reminiscent of Android's rise in mobile. By making models freely available, Meta encourages adoption across start-ups, researchers, and enterprises, pulling the wider development community into its orbit.

Zuckerberg has framed the company's long-term ambition as building "Personal Superintelligence" — a personalised AI assistant for every user — with improvements to Facebook, Instagram, and WhatsApp as near-term expressions of that vision. Recent acquisitions of voice AI companies WaveForms AI and PlayAI signal that voice interfaces will be central to this roadmap.

Key Takeaways

• →Meta is laying off approximately 8,000 employees (10% of workforce) and transferring 7,000 into AI-focused roles — with the restructure ultimately affecting around 20% of the company.
• →Four new AI-native units — Applied AI Engineering, Agent Transformation Accelerator, Central Analytics, and Enterprise Solutions — are being stood up to replace traditional team structures with autonomous AI workflows.
• →Meta's 2026 capital expenditure has been raised to up to $145bn — double its 2025 spend — signalling an unprecedented commitment to AI infrastructure buildout.
• →The company's open-source LLaMA strategy positions Meta not as a ChatGPT competitor, but as AI infrastructure for developers and enterprises globally — mirroring Android's dominance of mobile.
• →Meta's restructure mirrors a broader Silicon Valley shift: Salesforce, Atlassian, and others are simultaneously cutting roles and reorienting operations around AI agents and automation.
• →Having laid off approximately 25,000 people in four years, Meta is betting its future on a leaner, AI-native operating model — with Zuckerberg declaring 2026 the year AI "dramatically changes" how the company works.