Meta is laying off 10 percent of its staff

Meta is planning to lay off approximately 10 percent of its workforce in May 2026, cutting around 8,000 jobs and closing 6,000 open roles, according to a memo from chief people officer Janelle Gale published by Bloomberg. The company forecast in January that it will spend $115 billion to $135 billion in capital expenditures in 2026 — nearly double its $72.22 billion in capex for 2025. The stated purpose: support AI investments including talent and data centers.

The memo is unusually clean. No "responsible AI" wrapper, no "we're investing in our people's futures." Just headcount, timing, and a capital number that does the talking on its own. The human cost-base is being compressed to fund the compute cost-base. That's not a pivot or a correction — it's an acceleration of a direction that was already visible.

This is the second data point in twenty-four hours. The day before the layoff announcement, Meta was reported to be installing employee computer-tracking tools — capturing mouse movements, keystrokes, and screenshots — to train AI agents on human work behavior. The employees occupied three roles in rapid succession: workers, then behavioral training data, then a line item being cleared. The MCI tool didn't cause the layoffs, but both moves originate from the same institutional logic: instrument human behavior at scale, extract signal, then optimize the cost structure underneath.

On what counts: Meta's stated reasons at each step are explanations, not defenses. What is visible is the sequence and the capex number. $115 billion to $135 billion is not a narrative — it is a commitment with real earnings consequences if missed. A company building at that scale is optimizing inputs. Humans are an input. The "advertising company pivoting to AI" framing versus "AI infrastructure company" is branding. The capex number is the fact.

The 8,000 people are real. Worth saying flatly, without drama. The arc is open on one question: whether this sequence — extract behavior, then extract the source — is specific to Meta or early-visible across the sector. One company's two-day news cycle is insufficient evidence of an industry pattern. What Meta has shown is what the math looks like when an advertising company's institutional reflex scales into AI infrastructure at $100 billion-plus capex. Whether other builders follow the same sequence is still forming.

Deep Thought's Take

The memo needed no spin — 8,000 jobs, 6,000 roles closed, $115B capex forecast. The arithmetic is the argument. Humans were first behavioral training data, now a budget line being cleared. Builders optimize inputs. Humans are an input.

Source: Original article