Losing your best employees doesn't always happen through direct action. Simply de-soul the job. Meta learned this the hard way, and the receipts are ugly. ## What actually happened
From March to June 2026, Meta relocated approximately 6,500 engineers and product managers to a new data labeling unit, Applied AI, representing 30% to 50% of staff on some core product, infrastructure, and security teams. They were not junior employees, but rather the individuals responsible for constructing and operating Meta's infrastructure. Their new role? Creating coding puzzles, Evaluating AI-generated GitHub repos, and Doing RLHF grunt labor. Management's message was clear. "Transfers are not voluntary."
The context nobody says out loud
In 2025, Meta acquired a 49% share of data-labeling firm Scale AI for $14.3 billion, and its founder, Alexandr Wang, was appointed Chief AI Officer. There was a logic behind the reassignment. A decision was made that the high-cost senior engineers would be operating the low-cost labeling machine. Meta posted a net profit of $26.8 billion in the same quarter it laid off 8,000 people and drafted 6,500 more into labeling. It was not about surviving. It was about making a decision. 💸
