Meta cuts hundreds of staff in Reality Labs and sales while ramping up AI spending

On March 25, 2026, Meta began a new wave of layoffs: the cuts affected Reality Labs, Facebook, recruiting, sales, and global operations teams. The company continues to redistribute money and personnel in favor of AI, while sharply increasing spending on data centers, accelerators, and its own model ecosystem.

Scale of Layoffs

According to sources, this involves hundreds of employees. For Meta with 78,865 employees at the end of 2025, this does not look like a one-time mass collapse, but in the context of recent months the picture is different: the company is cutting staff in waves and barely pausing. The current restructuring is presented as a routine adjustment of teams to business goals, but the list of affected divisions shows that this is no longer about local correction, but about a reassembly of priorities.

In January, Meta already cut around 1,500 people from Reality Labs — roughly a tenth of the division — and closed three VR studios: Twisted Pixel, Sanzaru Games, and Armature Studio. Before that, the company conducted performance-based layoffs affecting approximately 3,600 employees. And in mid-March, information surfaced publicly about possible plans to prepare cuts of up to 20% of staff, although Meta itself called such estimates speculative.

Therefore, the March wave looks not like a separate episode, but as a continuation of a long restructuring.

Where the Money Goes

The reason for this wave is quite transparent: Meta is sharply raising the stakes on AI infrastructure. For 2026, the company budgeted capital spending in the range of $115 billion to $135 billion versus $72 billion the year before. At the same time, overall business expenses could reach $162–169 billion. This is almost a twofold jump in investments, and the market appears willing to tolerate it if simultaneously Meta becomes more compact, faster, and tighter on costs. The main areas of spending are already clear:

• construction and expansion of data centers
• purchase of NVIDIA GPUs for model training and inference
• development of proprietary chips
• support for the Llama ecosystem and Meta Superintelligence Labs

Reality Labs remains a painful point. The division showed an operating loss of $19.2 billion for 2025, and cumulative losses since its creation have approached $90 billion. Meta is not abandoning hardware ambitions entirely, but shifting focus: instead of expensive bet on VR headsets, the company is increasingly looking at smart glasses and wearable devices with AI, where it's easier to connect hardware with the current boom in assistants and personal models. In other words, the money is not leaving hardware altogether, but moving from old VR logic into new AI scenarios.

Market Logic

For investors, the signal is straightforward: if Meta is spending tens of billions on AI, it must show that these investments will pay off through increased efficiency. Against this backdrop, cutting support functions and slowly growing divisions looks like an attempt to free up budget for more expensive infrastructure. Analysts are already warning that the company's free cash flow could drop sharply, but the market, judging by the stock reaction in March, is more concerned not with the size of spending, but with the speed at which Meta turns it into product and margin.

This logic is visible not only at Meta. In the first quarter of 2026, technology companies worldwide eliminated more than 45,000 jobs, and in at least one in five cases the driver was precisely AI. Meta, for its part, claims that AI tools are already noticeably boosting engineer output: the company reported a rise in output per developer of roughly 30% since the start of 2025, and even higher for the most active users of internal AI systems.

If these figures hold up, the model of "fewer people, more computing" will take hold across the industry.

What This Means

Meta is increasingly moving away from the metaverse story as the main growth driver and betting on infrastructure, models, and AI-powered devices. For the market this is yet another marker: in 2026, large tech companies compete not by headcount but by access to computing, chips, and speed of AI implementation in real products. For employees inside such companies, this means a tighter link between AI budget and job security.