SAN FRANCISCO — Meta Platforms is internally planning workforce cuts that could eliminate up to 20% of its nearly 79,000 employees — roughly 16,000 jobs — as the company seeks to offset surging costs from its historic $600 billion commitment to artificial intelligence infrastructure, three people with direct knowledge of the matter told Reuters on March 13, 2026.
The scale of what’s being contemplated is striking. This would surpass the 22,000 jobs Meta cut between November 2022 and early 2023 — a stretch Zuckerberg branded the “Year of Efficiency” — making it the single largest workforce reduction in the company’s history. Top executives have already briefed senior leaders to begin drawing up reduction plans, according to sources cited by Reuters, though no date has been confirmed and final numbers remain in flux.
Who actually gets cut remains the overlooked detail in mainstream coverage. Documents reviewed by reporters indicate the reductions are expected to target non-AI roles — product, operations, and support functions — while Meta‘s newly formed Meta Superintelligence Labs (MSL) stays largely shielded. MSL currently employs just under 3,000 people and is, by all accounts, the division Zuckerberg is betting the company on. Meanwhile, Meta has been dangling pay packages worth hundreds of millions of dollars over four years to recruit individual AI researchers into that same unit, according to Reuters. The math is uncomfortable: lay off thousands of existing workers to fund extraordinary compensation for a few dozen elite hires.
Meta spokesperson Andy Stone pushed back on the reporting. “This is speculative reporting about theoretical approaches,” Stone said in a statement to Reuters — a denial that, notably, doesn’t actually deny the underlying planning.
Meta’s AI Spending Dwarfs Any Prior Bet
CFO Susan Li confirmed on Meta‘s most recent earnings call that 2026 capital expenditure will reach between $115 billion and $135 billion — nearly double the prior year — as the company races to build hyperscale data center campuses across the United States. The $600 billion plan through 2028, confirmed after a White House meeting with President Trump, includes a 2,250-acre Hyperion campus in Louisiana estimated at $10 billion to build and a Prometheus facility in Ohio expected online in 2026.
Meta has already purchased more than 1.3 million GPUs and secured $27 billion in Blue Owl Capital funding for data center construction, according to data reviewed by The AI Consulting Network. That’s a staggering capital commitment for a company whose own AI models have, by its internal assessments, repeatedly missed the mark.
Zuckerberg’s AI Division Faces Internal Panic
The urgency behind these layoffs isn’t just financial. Meta‘s AI efforts have been stumbling in ways the company hasn’t fully acknowledged publicly. A former Meta AI employee told Silicon Valley Insight that the open-source Llama 4 model’s clear lag behind China‘s DeepSeek “instilled a strong sense of crisis in Zuckerberg.” Llama 4, released in April 2025, drew only a tepid market response while rivals OpenAI, Anthropic, and Google continued shipping more capable models.
MSL‘s first original model, internally called Avocado, has already fallen behind internal benchmarks and been delayed. A second model, Mango, is also underperforming expectations. Behemoth, the largest planned Llama 4 variant, was quietly shelved altogether. One former employee described the internal dynamic bluntly: “middle management from product divisions took leadership roles over AI experts” — a misalignment of authority that several engineers cited as a core reason for Meta‘s AI struggles.
Chief AI Officer Alexandr Wang, who joined Meta after Zuckerberg acquired his startup Scale AI for $14.5 billion, has already executed one round of cuts — approximately 600 AI workers were let go from MSL in October 2025, affecting legacy research, product, and infrastructure units. His internal memo at the time stated that “by reducing the team size, the necessary discussions for decision-making will diminish.” The current round, if it materializes, would be orders of magnitude larger.
A Pattern Across Big Tech
Meta is not operating in isolation. The trade-off playing out in Menlo Park mirrors moves across the sector:
- Amazon cut approximately 16,000 jobs earlier in 2026
- Block reduced its workforce by nearly half, with CEO Jack Dorsey attributing cuts to AI productivity gains
- Atlassian announced plans to eliminate roughly 1,600 employees, or 10% of its staff
OpenAI CEO Sam Altman has publicly suggested some of these reductions are post-pandemic correction dressed up as AI strategy — what critics are calling “AI-washing.” Whether Meta‘s cuts reflect genuine AI-driven efficiency or a long-overdue rightsizing is a question several analysts contacted for this story declined to answer on record.
Meta also cut equity-based compensation for most employees by approximately 5% this year — the second consecutive annual reduction — while simultaneously offering historic pay to new AI recruits, according to the Financial Times. That combination has not gone unnoticed internally.
Meta has not confirmed a timeline for the reported cuts. The company’s next earnings report will be closely watched for any official acknowledgment. What remains unresolved is whether the cuts arrive before or after Meta‘s AI bets produce anything that can justify their scale.
