Meta has begun laying off employees as it focuses more of its cash on building out datacenters, training its own large language models, and recruiting talent for AI.

A person familiar with the cuts told The Register they would number about 700.

According to The Information, the job losses will fall hardest on Meta’s Reality Labs, its social media division, and recruitment.

“After 6 years at Meta, my role was impacted by the recent reduction in force today,” wrote a woman who worked as a senior recruiter with Meta until this morning in a LinkedIn post. “This one is especially tough. After returning as a short-term employee in 2024, I was grateful to receive a full-time offer again last year and I’m incredibly proud of what I was able to accomplish during that time. The gratitude I feel far outweighs the disappointment.”

In a statement to The Register, Meta said this reduction in force is about streamlining the business to work more effectively with AI as laid out by Meta CEO Mark Zuckerberg during earnings reports in January.

“Teams across Meta regularly restructure or implement changes to ensure they’re in the best position to achieve their goals. Where possible, we are finding other opportunities for employees whose positions may be impacted,” a spokesperson wrote.

In a post-earnings note on January 28, Zuckerberg said this was the year Meta would begin “flattening teams.”

“We're elevating individual contributors, and flattening teams. We're starting to see projects that used to require big teams now be accomplished by a single very talented person,” he wrote. “I want to make sure as many of these very talented people as possible choose Meta as the place they can make the greatest impact – to deliver personalized products to billions of people around the world. And if we do this, then I think we'll get a lot more done and it's going to be a lot more fun.”

Reuters reported recently that Meta plans to lay off 20 percent of its workforce – some 15,000 employees – but the layoffs that have reportedly begun this week are on a smaller scale thus far.

Meta said it had 78,800 employees as of the end of January, a number that had grown in recent years as it sought to build a bench of AI talent that could build a platform capable of competing with frontier model providers such as Anthropic and OpenAI.

If Meta were to follow through with a 20 percent cut, it would mean the elimination of about 15,000 jobs and bring Meta’s headcount to its lowest point since 2021, when it had about 58,600 full-time employees.

Spending on AI

Meta has dramatically increased spending in recent years to keep up in the AI arms race as it focuses on building its own AI infrastructure and datacenter properties to match competitors Anthropic, Google, and OpenAI. Expenses rose 24% during 2025 to $118 billion, and the company has said it plans to spend between $162 billion and $167 billion this year (although it expects operating income to increase, meaning revenue will grow faster than expenses). Of that, capital expenditures - including datacenter buildouts to power its AI efforts - will amount to between $115 billion and $135 billion.

The company is also designing its own custom chips for GenAI workloads which it plans to build over the next two years. The first of its inhouse MTIA chips was released in 2023.

“(The) MTIA 300 will be used for ranking and recommendations training, and is already in production. MTIA 400, 450 and 500 will be capable of handling all workloads, but we will primarily use these chips to support GenAI inference production in the near future and into 2027,” the company said.

The release of Meta’s next reasoning model, code-named Avocado, has reportedly been delayed, after delivering underwhelming results during internal tests, according to the New York Times. That news comes even as Meta offered dramatic nine figure pay packages to lure AI researchers from competitors last year with OpenAI defectors reportedly commanding $100 million sign-on bonuses.

Zuckerberg also invested $14 billion in Scale AI and tapped its co-founder Alexander Wang to lead Meta’s AI efforts. Wang reportedly clashed with Meta’s former chief AI scientist Yann LeCun, who called Wang young and inexperienced after he quit. LeCun was also known as the godfather of AI. LeCun accused Zuckerberg of pushing aside the former AI team after the company’s disappointing Llama 4 model release.

Recently, Facebook CFO Susan Li talked with analysts at Morgan Stanley at the company’s Technology, Media & Telecom conference about the uncertain path for a return on Meta’s AI investments.

• Meta's latest AI improves its terrible content moderation, just a little
• Meta – yep, Facebook Meta – is now a defense contractor
• AI nonsense finds new home as Meta acquires Moltbook
• Meta, international cops use handcuffs and AI to stop scammers

“That's not like, okay, in 2026, the ROI is this in 2027, the ROI is this and so on, which pains me, to be clear,” Li said. “I really wish that, that were the world we live in, but it's not. And we have to be willing to sort of make temporal bets, and that's a big part of what we have to do in an intelligent and thoughtful way.”

She said Meta can accurately gauge what the costs will be for personnel and infrastructure to run the platform’s existing apps and experiences. It can also calculate how much it will cost to build new AI capabilities, including the employees and compute costs.

But there is a blindspot when it comes to guessing how much inference power will be needed if the AI products that the company produces need to be scaled quickly to meet user demand.

“The teams that are working on basically AI training today, they have the most immediately sort of clearly defined buttoned-up road map for how much capacity, let's say, they think they need to train models for the next 12, 24 months,” Li said. “That's kind of like a demand road map from the teams that they have more certainty into. The part I think that is the most challenging for us to have certainty around is inference needs because that's both - you have to predict meaningfully into the future because of the lead time and getting capacity.” ®

More about

• Meta

More like these