Meta posted $56 billion in quarterly revenue on May 14. The next week, it started cutting eight thousand people.
These two facts are not in contradiction. They are, in fact, the same sentence — a sentence about where a company of Meta's size believes the next five years of value will come from, and about how different a software engineering organization looks on the other side of that belief.
The May 20 cuts represent roughly ten percent of Meta's workforce of about seventy-nine thousand people. The company is also cancelling six thousand open requisitions it had planned to fill, bringing the effective headcount reduction closer to fourteen thousand positions. The spending freed up by this headcount reduction is being redirected into an AI infrastructure allocation of between $115 billion and $135 billion for 2026.
The person at the center of the restructure is Alexandr Wang, twenty-eight years old, formerly the CEO of Scale AI, now Meta's Chief AI Officer. Meta paid $14.3 billion for a forty-nine percent stake in Scale AI to bring Wang in. He runs Meta Superintelligence Labs, which in early May released Muse Spark, the division's first public model.
What Wang's division is building is not only models. It is a new organizational template. The restructure is creating three new job families: AI Builder, AI Pod Lead, and AI Org Head. Teams are being reorganized into small cross-functional pods oriented around AI product delivery rather than around traditional engineering functions. The engineers who remain after the May 20 cuts will largely be operating inside this pod structure.
The tension inside the restructure is visible from the outside. Wang is nominally Chief AI Officer, but the creation of a separate Applied AI Engineering unit under Maher Saba — reporting to CTO Andrew Bosworth, not to Wang — creates a parallel AI leadership track. The org chart reflects a bet that has not yet fully resolved into a single architecture.
What the organizational form makes visible is the assumption beneath the bet. The "AI Builder" title is not cosmetic. It is a claim about what software engineers will do when agentic AI handles more of the task-level execution: they will configure agents, define workflows, own the output — but they will not be writing the same volume of code they wrote before. The number of engineers required to operate a software product, under this model, shrinks. The kind of engineer required changes.
This is happening at a company with record revenue. Meta's Q1 earnings — $56.06 billion in revenue, $17.6 billion in net income — are not the financial profile of a company forced to cut. They are the profile of a company choosing to redirect. The capital that was in human salaries is being redirected into infrastructure because the company believes the infrastructure will compound faster.
The broader signal across the industry is consistent. Oracle cut thirty thousand workers. Snap cut a thousand. Cisco cut substantially. The running total of tech layoffs in 2026 is already approaching a hundred and forty thousand people. Record or near-record revenues appear alongside the cuts in nearly every case, with AI reallocation cited as the driver. Companies are not cutting because they are struggling. They are cutting because they believe they can produce more with less, and they are betting the savings on the infrastructure that will make that true.
For an engineer reading this outside the US tech hiring market — in Ulaanbaatar, in Nairobi, in Kraków — the pod structure is worth understanding not as a threat but as a format. The small cross-functional AI pod is a shape a three-person studio can take today, without Meta's budget or timeline. The question "am I doing the work that requires a human" is one a small team can answer right now, before the restructure reaches their context.
The bet Meta is making will either prove the model or expose the gap between what AI pods can actually ship and what the seventy-nine-thousand-person organization was producing — and that answer will show up in product quality before it appears on an earnings call.
The short of it.
On May 20, Meta began cutting eight thousand jobs — ten percent of its workforce — the same week it reported $56 billion in quarterly revenue. The cuts are part of a deliberate reallocation toward AI infrastructure budgeted at $115-135 billion for 2026 and toward a new organizational structure of "AI pods" with new job families centered on AI product delivery. The signal is consistent across Oracle, Snap, Cisco, and the nearly 140,000 tech workers laid off in 2026 so far: companies with record revenues are reducing engineering headcount because they have already bet the AI tools will do more of the work. The pod format is worth studying even if you are not at Meta — it describes what a small AI-capable team is supposed to look like.