Meta Platforms shares closed 2.33% higher at $627.45 on Monday, March 16, after a weekend Reuters report said the company is considering laying off up to 20% of its global workforce, a move that analysts quickly framed as a continuation of the cost discipline narrative that has driven Meta’s stock performance since 2023.
The company employed approximately 79,000 people as of December 31, 2025, meaning a 20% reduction would eliminate roughly 15,800 positions and mark the largest single round of layoffs in the company’s history, surpassing even the 21,000 cuts made across four months during the “year of efficiency” under Mark Zuckerberg’s leadership in 2022 and 2023.
Meta called the Reuters reports speculative in a statement, but the market’s reaction was unambiguous: investors read the prospect of a leaner cost structure not as a warning sign but as a value-creation mechanism, particularly given the extraordinary level of capital expenditure the company is committing to artificial intelligence infrastructure.
Meta forecast spending between $115 billion and $135 billion on AI in 2026, roughly double what it spent the year before, a capital intensity that makes the financial logic of simultaneous workforce reduction relatively straightforward to model.
Rosenblatt Securities analyst Barton Crockett put the arithmetic plainly: “A 20% staff cut could amount to about $6 billion in cost savings, or a 5% boost to adjusted core earnings.”
On the same day the layoff reports circulated, Meta confirmed a landmark five-year infrastructure agreement with Nebius Group worth up to $27 billion, structured as $12 billion in dedicated AI capacity with an additional $15 billion in supplementary compute access, making it one of the largest single infrastructure contracts the company has ever signed with a third-party provider.
Nebius, which describes itself as a “neocloud” provider building AI infrastructure that hyperscale cloud platforms cannot efficiently supply, will deploy Nvidia’s Vera Rubin platform as part of its delivery commitments, with capacity coming online across multiple locations beginning in early 2027.
The deal expands a relationship that already included a $3 billion agreement signed the prior year, and it follows Nvidia’s own $2 billion strategic investment in Nebius disclosed just days earlier, a sequence of moves that has transformed Nebius from a relatively unknown infrastructure company into one of the most prominently positioned names in the AI supply chain.
The context that makes the layoff discussion particularly charged is Meta’s AI model performance: the company has been developing a frontier model codenamed Avocado that, according to reporting from the New York Times, has failed to match the performance of Google’s Gemini 3 across reasoning, coding and writing benchmarks, leaving the company spending aggressively on infrastructure while simultaneously lagging on the model quality that the infrastructure is supposed to enable.
Whether the reported workforce reduction proceeds, and on what timeline, remains unclear given that Meta has not confirmed a date or final scale, but the market has already priced in the probability that some version of the cuts materialises, given the pattern of investor behaviour that has rewarded similar announcements from companies including Block, whose shares rose roughly 20% after CEO Jack Dorsey announced plans to eliminate 4,000 employees.