Mark Zuckerberg told Meta employees on Thursday that the company's planned layoffs of roughly 8,000 staff are a direct consequence of its expanding AI infrastructure budget, the most explicit framing yet from a tech chief executive that compute spending is now displacing headcount.
The cuts, equivalent to about 10% of Meta's workforce, will begin on May 20. They were announced the same week the company raised its full-year 2026 capital expenditure guidance to between $125 billion and $145 billion, up from a prior range of $115 billion to $135 billion. Meta spent $72.2 billion on capital expenditure in 2025. The midpoint of the new 2026 range would nearly double that figure in a single year.
"We basically have two major cost centers in the company: compute infrastructure and people-oriented things," Zuckerberg told employees at a town hall, according to reporting from Reuters. With more capital flowing toward AI hardware, he said, less is available for headcount, and he declined to rule out further reductions later in the year. The town hall comments were initially surfaced by Forbes.
The timing undercuts any suggestion that Meta is cutting jobs out of financial necessity. First-quarter 2026 earnings, reported the day before the town hall, showed revenue of $56.31 billion, a 33% year-on-year increase, and net income of $26.8 billion. Q1 capital expenditure alone reached $19.84 billion. Chief financial officer Susan Li told investors during the earnings call that she could not predict the company's optimal long-term workforce size given the pace of AI capability improvements.
Compute crowding out labour, explicitly
Zuckerberg's framing is unusually candid. Most large-cap tech employers cutting headcount in 2026 have attributed reductions to "operational efficiency", "rebalancing", or AI-driven productivity gains. Zuckerberg pointed directly, instead, at infrastructure cost.
OpenAI chief executive Sam Altman told CNBC at the India AI Impact Summit in February that some firms engage in what he called "AI washing", attributing layoffs to the technology when the underlying drivers lie elsewhere. Zuckerberg's explanation is more specific than most CEOs offer. Its specificity, however, raises a different question.
If the binding constraint is infrastructure spend rather than AI-driven productivity, the trade is not obvious. Bryan Catanzaro, Nvidia's vice-president of applied deep learning, said earlier this week that compute already costs more per unit of output than the employees on his team. A 2024 MIT study found AI automation was economically viable in only 23% of vision-related roles. If AI infrastructure is currently more expensive than the labour it would replace, the case for trading one for the other is weaker than the layoff announcements imply.
The 2026 layoff context
The cuts are the largest at Meta since the 2022 and 2023 rounds that shed roughly 21,000 positions combined. Across the broader US tech industry, more than 80,000 workers have been laid off in 2026 so far, with separate large reductions reported at Oracle (around 10,000) and at Microsoft. The common thread is capital expenditure: combined capex for Microsoft, Amazon, Google, and Meta is on track to hit roughly $725 billion in 2026, up about 77% from the prior year.
Some Meta employees have used the company's internal message boards to push back, particularly given that the headcount reduction is happening alongside a separate initiative to monitor staff productivity through keyboard and mouse activity tracking. Executives have told employees the monitoring tooling is unrelated to the layoff decision. The internal reception has been sceptical.
What to read into it
For larger companies trying to forecast their own AI infrastructure budget pressure, the Meta announcement is the clearest signal yet that capital expenditure is starting to compete directly with payroll inside the largest balance sheets in the industry. The trajectory is not a one-off. Susan Li's refusal to commit to a long-term headcount target reflects exactly the uncertainty that smaller companies should be planning around.
The practical implication is that the line between "growth-stage spending" and "structural reallocation of inputs" has moved. Zuckerberg's explicit choice to fund GPUs over engineers is what the industry has been edging toward for two years. Now it is on the record. The same margin pressure that drives providers to quietly throttle compute on the customer side of the API is, on the producer side, starting to land on payroll.
