Meta: AI layoffs

Avatar
Lisa Ernst · 24.10.2025 · Technology · 5 min

Meta is currently cutting around 600 positions in its AI unit, while the company is simultaneously investing massively in AI infrastructure and continues to hire for key roles. The crucial question is therefore: Are these cuts a retreat – or a reorganization for the next AI push? Meta's plans for up to 60–65 billion USD CapEx in 2025 for AI hardware point more toward the latter.

Introduction

Meta is currently cutting around 600 jobs in its AI unit, while the company is simultaneously investing massively in AI infrastructure and continuing to hire for key roles ( Reuters, AP News). ). The crucial question is therefore: Are these cuts a retreat – or a reorganization for the next AI push? Meta's plans for up to 60–65 billion USD CapEx in 2025 for AI hardware suggest the latter ( Reuters).

Definitions

By layoffs we mean involuntary terminations or the reduction of announced positions. At Meta, they have affected several waves since 2022: 11,000 jobs in November 2022, a further 10,000 in 2023, followed by later, more targeted cuts ( Reuters, Reuters). ). AI stands for Artificial Intelligence – at Meta mainly models from the Llama family as well as the assistant “Meta AI”, which the company openly positions and integrates into products ( Meta AI Blog, Meta AI Blog). ). “Superintelligence Labs” is the internal umbrella organization for product-facing AI teams up to FAIR (Research) – many of the roles now affected sit there ( AP News).

Current status

On October 22, 2025, it became known: Meta is cutting around 600 positions in its AI unit; according to reports, this is part of a reorganization intended to shorten decision-making paths and focus the remaining teams. At the same time, the new “TBD Lab” remains untouched, and Meta continues to hire for strategic AI roles ( Reuters, AP News). ). Earlier in 2025 there were targeted cuts, including performance-related (“ca. 5 % of the low performers”) – with the intention of filling the roles again in 2025 ( Reuters). ). In parallel, Meta trimmed parts of Reality Labs (VR/AR) in 2025 ( Reuters). ). The broader frame: After the “Year of Efficiency” phase 2022/23 (11,000 + 10,000 jobs) there has been a permanent overhaul toward leaner, tech-focused teams ( Meta Newsroom, Reuters). ). Concurrently, Meta reports strong AI usage: More than 1 billion people use Meta AI monthly ( Meta Newsroom).

Abstract depiction of the impact of layoffs on employees.

Quelle: bbrief.co.za

Abstract depiction of the impact of layoffs on employees.

Analysis

Strategically, the approach fits: cut costs, remove duplications, but at the same time expand the AI leadership with massive investments and top recruiting ( Reuters, Washington Post). ). Open AI models (Llama 3/3.1) serve as a developer magnet and accelerate ecosystem effects – a means to drive user engagement and advertising innovations ( Meta AI Blog, Meta AI Blog). ). Globally, the IMF expects around 40 % of all jobs to have AI exposure; in advanced economies it's about 60 % – this increases the pressure to keep product teams lean and fast ( IWF Blog). ). At the same time, research shows that generative AI can significantly boost productivity especially for less experienced employees – 14 % more cases solved in a call center field test ( NBER).

Quelle: YouTube

The Connect keynote clip helps place Meta's AI product focus (Assistant, Glasses, Llama).

Fact-check

Fact: Meta cut around 600 AI roles on 22.10.2025; the TBD Lab is not affected; simultaneous additional AI hires planned ( Reuters, AP News). ). Also documented is the massive AI investment framework (60–65 billion USD CapEx in 2025) ( Reuters). ). Clearly documented are the large layoff waves 2022/23 (11,000 and 10,000) ( Reuters, Reuters).

Unclear: how many of the currently affected AI employees will switch internally, how quickly open roles will be filled, and how the net capacity in research vs. product will shift over 2025/26 ( AP News).

False/Misleading: “Meta is exiting AI” – the opposite is the case; CapEx, product roadmap and Llama releases point to a shift toward priorities, not away from AI ( Reuters, Meta AI Blog).

The Meta logo on a modern building symbolizes the company's presence in the technology sector.

Quelle: techlusive.in

The Meta logo on a modern building symbolizes the company's presence in the technology sector.

Reactions & Counterarguments

Media categorize the cuts as a reorg to speed and focus; at the same time they emphasize the aggressive AI race ( Washington Post, Reuters). ). The AP notes that TBD Lab remains untouched – a signal that Meta is continuing to ramp up there ( AP News). ). Critics point to earlier concerns about accountability in AI (e.g., restructuring of the “Responsible AI” roles 2023) ( The Verge). ). Proponents see in open models an innovation dividend for the entire ecosystem ( Meta AI Blog).

Impact & What it means for you

For employees: AI creates new, higher-value roles and makes some tasks obsolete; retraining and AI skills will become a career lever ( IWF Blog, OECD). ). For companies: Not “AI first” at any price, but clear prioritization, data quality, product integration and responsible use. For fact-checks on layoffs: check directly with Investor Relations/Newsroom and – if available – review in 8-K/10-Q (example Meta IR/Newsroom and SEC-EDGAR) ( Meta Newsroom, SEC EDGAR).

Quelle: YouTube

The opening of Connect 2025 puts the roadmap into context (AI assistant, devices, models).

Open questions

How will Meta allocate AI capacity in the long term between fundamental research (FAIR), product AI and TBD Lab? What net impact do internal transfers vs departures have on the release speed in 2026? How robust is the statement of "1+ billion monthly active users" for Meta AI across regions and products – and how will it be disclosed in the future ( Meta Newsroom)? What effects do global AI trends have on employment and wages in individual functional areas – beyond the IMF macro-exposure rates ( IWF Blog)?

Conclusion

The Meta layoffs in the AI space are not a withdrawal, but a hard prioritization: less breadth, more depth – backed by record investments in chips, data centers and models ( Reuters, Reuters). For you, that means: deliberately build AI skills, prioritize real product applications, take accountability seriously – and always check corporate statements against original sources, from the Newsroom to the SEC filing ( Meta Newsroom, SEC EDGAR).

Teilen Sie doch unseren Beitrag!