Earlier this week, social giant Meta Platforms made headlines after a memo to its employees said it would install software on employees’ computers to train AI tools.
Internally, it was understandably controversial, as there’s no way to “opt out” of task tracking. Externally, it’s likely to become a study for other companies thinking about doing something similar. But although the move likely runs aground of privacy regulations or stipulations in employment law, the undertones are obvious.
After months of rumors and speculation, the memo was seen as a potential admission that Meta would be training AI replacements for thousands of its employees. Now, the move appears to be a starting gun for layoffs.
Related: Mark Zuckerberg sends shocking message to Meta employees
Meta announces May layoffs
On Thursday, Meta told employees it would lay off 10% of its workforce, or about 8,000 jobs. Decisions will land on May 20, the latest in a row of downsizings at the company. In addition, Meta says it will not fill 6,000 open roles.
Meta Chief People Officer Janelle Gale says that the move is part of a “continued effort to run the company more efficiently.”
What costs more: Hardware or humans?
Despite purporting that the layoffs are a matter of efficiency, Meta Platforms has struggled to keep its headcount down. The company’s headcount nearly doubled from the end of 2019 to 2022, rising from 44,942 to 86,482. After sizable layoffs in 2023, Meta’s headcount returned to growth in 2024 and 2025.
That underscores one of the other reasons for the layoffs: Meta’s investments in AI, which Gale acknowledges in the memo by saying that they will “allow us to offset the other investments we’re making.” That is largely expected to refer to the company’s planned $600 billion investment in data centers.
That spend-a-thon is already breaking the bank for Meta, which burned through the majority of its cash on hand in 2025, tapped the private credit circuit to raise $29 billion for data centers, and is raising red flags with its own auditor over how its investments fit on the balance sheet.
Reality Labs
It’s not the first time this year that Meta has shifted resources from one business to its AI ambitions. In January, Meta cut over 1,000 jobs from its beleaguered Reality Labs division as it pivoted from one unprofitable business to another. That was about 10% of the VR and AR-focused group, which has run up a $100 billion loss since spinning out as a separate business unit during the pandemic.
Meta’s AI ventures are already surpassing that total, even as the company lags competitors like Google, Anthropic, and OpenAI. That’s especially ironic because Meta’s LLaMa AI models were, at one point, industry-leading. But after falling behind, the company spent $14 billion to acquire a minority stake in Scale AI, a data-labeling company.
How can Meta’s move pay off?
Meta is late to the party, but their spending indicates that they expect this is a game of resource accumulation. They might not have a consumer product like their competitors; they don’t even have a frontier model ready for prime time. But with enough spending, they suspect that they can play ball in the congested AI ecosystem.
However, it remains unclear if Meta’s latest “efficiency” layoffs are because of AI or in spite of it. At this stage, Meta concedes that it’s about “efficiency” (read: saddling fewer employees with more work, or, alternatively, focusing on business areas of interest and doing a whole lot else). However, it also admits that it’s about affording its AI investments without significantly affecting its financial performance.
The problem with the aforementioned explanations is that both are terrible for morale. Employees are now on notice that they’ll make themselves redundant.
If this investment pays off for Meta, it’s problematic for the state of white-collar work. When companies downsize, historically, there is a loss of institutional knowledge and resources. But with a new AI hawk to retain that knowledge (and even do the job), neither of those things may be a worry. It means that these layoffs might just be the starting gun.
There is, of course, the possibility that it doesn’t work out — at least not over the near term. Many AI labs are struggling to rein in costs, and where they have, there has been an observable decline in model quality. This is not especially bullish for the intermediate-term prospect of AI models as “replacements” for human labor; at best, they still look like expensive tools.
If that’s the case, we might expect Meta’s headcount to begin growing again next year …
