Meta cuts staff stock awards for a second year as AI spending ramps up

Meta has reduced its annual distribution of stock options for most employees for the second year in a row, as the company ramps up spending on top AI talent and infrastructure.

The Financial Times has reported that equity-based awards were reduced by roughly 5% for most of Meta’s staff, affecting tens of thousands of employees. The exact change, however, might differ based on each role.

The company already cut stock awards by about 10% last year, which the FT says shocked its staff.

It also laid off approximately 10% of employees from Meta’s Reality Labs group, which had about 15,000 workers. The decision was driven by Meta’s virtual reality projects underperforming, prompting it to shift focus to AI-powered wearable technology.

Generally, Meta’s employees receive annual equity refreshers, as well as base salaries and annual bonuses. These refreshers are adjusted each year based on industry trends.

The company is also reportedly changing its performance review system this year, with top performers receiving bonuses, which could mean that despite the cuts, the compensation budget has gone up.

Meta is facing fierce competition from major tech rivals to improve its AI standing. CEO Mark Zuckerberg, as the FT puts it, “has embarked on a staggering AI spending spree,” aiming to attract key AI talent from rival companies and develop sophisticated models.

The conversation on this topic is live. Join in the discussion.

In late June, OpenAI lost three prominent AI researchers to Meta, with Sam Altman saying that Meta was offering its new employees a massive signing bonus of $100 million. In July, Ruoming Pang, Apple’s distinguished software engineer, also left the company to join Meta’s superintelligence group.

In August 2025, Zuckerberg outlined his vision “to bring personal superintelligence to everyone,” saying that Meta has begun to see glimpses of AI systems improving themselves in a slow yet “undeniable” improvement, which could mean that “developing superintelligence is now in sight.” Such ambitions would require significant investment in AI infrastructure, such as data centers and advanced chips.