← Journal0143 min read

The workforce receipt

Meta is cutting 8,000 people starting May 20. The same week, the company guides to $115 billion in AI infrastructure spending for 2026. The budget went somewhere.


The announcement came in late April and landed with the precision of a fiscal quarter. Meta would cut approximately 8,000 employees — roughly 10 percent of its 78,000-person workforce — beginning May 20, with more reductions planned for the second half of 2026. Mark Zuckerberg framed it as an AI efficiency push: teams were being reorganized into AI-focused pods, and the company needed fewer people in functions those pods would replace. The same week, Meta's capex guidance for 2026 stood at $115 to $135 billion. The budget did not disappear. It moved.

The tech industry has shed more than 128,000 workers in 2026 as of mid-May, an average of 856 jobs per day. Oracle is cutting between 10,000 and 30,000 employees. Cloudflare, Coinbase, and Upwork have added to the count. The four companies — Google, Amazon, Microsoft, and Meta — planning $725 billion in AI capital expenditure for 2026 are the same companies, or their close neighbors, generating most of the cuts. The pattern across all four is identical: flatten the cost structure, reallocate toward infrastructure, call it a transformation.

What makes the Meta restructuring notable is the specific machinery Zuckerberg has assembled to replace what he is cutting. Alexandr Wang, formerly CEO of Scale AI, now runs Superintelligence Labs as Chief AI Officer. The organization is built around the assumption that the next phase of product development is AI-authored — that models write the features, run the tests, manage the deployment loop, and that the team size required per unit of software shipped will continue to compress. The 8,000 cuts are not a single event. They are a bet on a ratio.

This is a different claim from the one that circulated three years ago, when companies cut during the post-pandemic correction and called it a right-sizing. In 2023, the companies cutting did not publish $725 billion capex guidance simultaneously. They were recovering their balance sheets. What is happening in 2026 is not recovery. It is deliberate reallocation at scale, and the destination is not ambiguous.

The question this raises for operators and builders is not whether AI is replacing jobs — the headline has been written and will be repeated. The question is what kind of work survives the restructuring, and in what concentrations.

The Meta cuts are concentrated in product management, mid-tier engineering, and support functions. The roles being preserved — and in some cases expanded — are the ones closest to model training, evaluation, and deployment infrastructure. The distinction is between people who shipped software and people who can evaluate whether software that a model shipped is any good. The first category is compressing. The second is, for now, scarce enough to matter.

For a small studio — for a team of six in Ulaanbaatar or twelve in Nairobi or twenty in Warsaw — the restructuring at large companies is not a direct threat. The threat is subtler: that the productivity floor rises fast enough to change what a client expects from a vendor. If a hyperscaler can ship a feature in two hours with three engineers and a model, the conversation about why your studio needs three weeks and five engineers changes in character. That conversation is arriving before most small shops have a ready answer.

The workforce receipt is being posted in quarterly filings and press releases, denominated in people. What it actually represents is a reallocation of organizational belief — from the idea that the bottleneck is human capacity to the idea that the bottleneck is compute and the humans who can direct it. Understanding what is actually being bet on is more useful than being alarmed by the headline number.

The short of it.

Meta begins cutting 8,000 employees on May 20 while guiding to $115-135 billion in AI infrastructure spending for 2026. Across the industry, 128,000 tech workers have been laid off this year while the same companies plan $725 billion in AI capex. The roles surviving are those closest to model evaluation and deployment, not feature shipping. For smaller operators, the risk isn't direct displacement — it's a rising productivity floor that changes what clients expect from everyone.

Working with us: hire MonArch

Founder-led studio. Two engagements at a time. Discovery first, software if needed.