Meta’s AI workforce overhaul collapses under the weight of its own ambition

Mark Zuckerberg had a plan to remake Meta from the inside out. Thousands of human jobs would be handed to AI. Entire teams would shrink by as much as 60%. Then, hours before the cuts were supposed to happen, he paused it.

What Project OT actually was

Meta launched what it called Project OT, short for Organization Transformation, in early 2026. The goal was straightforward on paper: use AI to absorb the daily workload of thousands of employees, then reduce headcount accordingly through layoffs, hiring freezes, performance exits, and role reassignments.

In May 2026, Meta followed through on the first wave. The company laid off roughly 8,000 employees, about 10% of its global workforce. It simultaneously closed 6,000 open positions and moved around 7,000 existing workers into AI-focused roles.

The deeper restructuring, however, a November wave that was supposed to cut even further, was canceled on May 19, 2026, one day before it was set to begin.

Inside Meta, the reaction to Project OT was not quiet acceptance. More than 1,500 employees signed a petition against the company’s plan to use keystroke and mouse-tracking software to train its AI systems, objecting to the idea that their own daily work patterns would become training data for the machines replacing them.

The numbers that made Zuckerberg pause

Meta’s internal AI tools generated a 220% year-on-year increase in code changes across the company. That same period produced only a 36% improvement in actually shipped products.

Technical incidents rose 40% as AI systems took on more internal responsibilities. The time engineers spent resolving those incidents, what the company internally called firefighting time, climbed 70%.

Employee sentiment scores tell their own story. Favorability ratings at Meta dropped from 74% to 55% following the restructuring announcements.

Zuckerberg acknowledged that progress with agentic AI, the kind capable of handling complex, multi-step work independently, had not met the company’s expectations. He suggested improvements were likely within three to six months.

A $130 billion bet that has to work

Zuckerberg projected more than $130 billion in AI infrastructure investment for 2026 alone.

Zuckerberg indicated after the May cuts that further large-scale layoffs in 2026 are unlikely, though he left room for team-specific reductions tied to performance. That is a more cautious posture than Project OT’s original 60% reduction targets implied.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.