Meta plans layoffs to offset AI costs
Reports say Meta is planning deep staff cuts to offset the rising costs of its AI push and rebalance spending, a sign of mounting cost pressure across big AI programs reported. That dynamic is already influencing customer conversations about cost‑per‑inference and predictable TCO.
Reuters' March 13 report said no date has been set and that the final magnitude of any cuts remained undecided(newsbreak.com). Meta guided(investor.atmeta.com) $115–$135 billion in capital expenditures for 2026 in its Q4 2025 results, and its Form 10‑K lists 78,865 employees as of Dec. 31, 2025(sec.gov). A 20% reduction would remove roughly 15,800 roles based on Meta’s Dec. 31, 2025 headcount of 78,865(sec.gov); by comparison, Meta eliminated about 11,000 employees in November 2022 and roughly 10,000 more in early 2023 during its prior “Year of Efficiency”(cnbc.com). Enterprise procurement conversations have shifted toward token-level and "cost‑per‑inference" economics, with analysts arguing inference is reshaping cloud pricing and architecture decisions(forbes.com) and market analysis estimating inference workloads now consume a majority share of AI‑optimized cloud spend (above 55% in early 2026). (byteiota.com) Vendor benchmarks and field deployments report steep inference‑cost improvements: NVIDIA’s Blackwell runs have shown 4–10× reductions on certain workloads in recent case studies(venturebeat.com), while vendor TCO whitepapers model that sustained, high‑throughput inference workloads can favor on‑prem appliance purchases over pay‑as‑you‑go cloud instances. (lenovopress.lenovo.com) Markets reacted quickly: Meta shares rose about 3% after the reporting as investors priced potential margin relief, while a Meta spokesperson described the coverage as "speculative reporting about theoretical approaches." (cnbc.com)
