The Human Cost of Meta's AI Race

Meta could be preparing to launch one of the biggest waves of layoffs in its history. According to sources obtained by Reuters, the group led by Mark Zuckerberg is considering job cuts that could affect up to 20% of its workforce, or approximately 15,000 to 16,000 employees out of the nearly 79,000 staff the company had at the end of 2025.

This decision is part of a major strategic transformation. Meta must absorb the colossal investments required for the development of artificial intelligence, particularly in computing infrastructure and data centers. At the same time, the rapid progress of AI tools allows for the automation of certain tasks previously performed by entire teams.

The Scale of the Shock

If these layoffs materialize, they would mark the most significant restructuring since the job cuts of 2022 and 2023. At the time, Meta eliminated about 11,000 jobs in November 2022, or 13% of its workforce, before announcing 10,000 new layoffs a few months later. Mark Zuckerberg had called this period the "year of efficiency," intended to streamline the group's internal structures after years of rapid growth.

In a sector disrupted by AI, these mass layoffs signal a shift from crisis management to a strategy of sustainable efficiency. Meta remains cautious in its communication, however. Questioned by Reuters, group spokesperson Andy Stone called the reports "speculative," referring to theoretical scenarios rather than final decisions.

Investing Billions, Saving on Humans

To remain competitive in the AI race, Meta must invest colossal sums while reducing certain expenses related to human resources. The group notably plans up to $600bn in data center investments by 2028 to support the development of increasingly powerful artificial intelligence models.

At the same time, Meta is increasing acquisitions in the sector. The company recently announced the purchase of Moltbook, a social platform designed for AI agents, and plans to invest at least $2bn in the Chinese AI startup Manus.

Meanwhile, Meta is doing everything to attract the technological elite. The group has offered contracts that can reach several hundred million dollars over four years to recruit the most talented specialists for its new team dedicated to "superintelligence."

A Silicon Valley in Structural Mutation

Meta is not the only company in this situation. In January, Amazon announced the elimination of approximately 16,000 positions. For its part, the fintech company Block, led by Jack Dorsey, reduced its workforce by nearly half, explaining that artificial intelligence tools now make it possible to do more with smaller teams.

These decisions reflect a structural evolution in Silicon Valley. Tech leaders are increasingly promoting a leaner and more flexible organizational model, in which automation and AI gradually replace certain human functions.

The Risky Bet on Superintelligence

The company has experienced several difficulties with its Llama 4 models, notably after criticism regarding the benchmarks used to measure their performance. Meta also abandoned the launch of the most ambitious version of the model, dubbed Behemoth, initially expected for the summer.

To regain momentum, the team dedicated to superintelligence is currently working on a new model named Avocado. But according to several sources, the performance of this system remains below expectations for now, increasing pressure on research teams.

Despite the promise of a leaner and more profitable structure, markets remain nervous, a sign that investors are as concerned about the human cost as they are about the technical difficulties of the Avocado and Llama models.