Meta drew up plans this year to cut some teams by as much as 60% as part of a sweeping push to build an “AI native” company, before Mark Zuckerberg pulled back at the last minute and scrapped a planned second wave of layoffs, according to internal documents and people familiar with the effort.
The plan, code-named Project OT for Organization Transformation, took shape in January at Zuckerberg’s Hawaii compound, where he and his top lieutenants gathered for their annual leadership retreat. The idea: shrink Meta’s human workforce and let AI handle much of the daily work currently done by employees, with smaller groups of staffers overseeing the technology.
Two waves, then a retreat
Executives mapped out the restructuring in two phases. The first wave hit on May 20, cutting 10% of Meta’s workforce. A second wave, planned for November, never happened. Zuckerberg called it off on the night of May 19, hours before the first round of layoffs went out, according to an internal document reviewed by Reuters.
Scenario planning for the second wave had explored cuts as steep as 60% across some teams, with a human-resources executive projecting a reduction on par with or larger than the roughly 25% workforce cut Meta carried out between late 2022 and early 2023.
Meta confirmed the existence of Project OT after Reuters approached the company, describing it as a year-long effort focused on cutting costs, redesigning team structures, and shifting employees into priority areas such as producing training data for its AI models. The company said the most aggressive scenarios did involve reducing some teams by up to 60%, but it disputed any suggestion that it intended to cut 60% of its entire workforce.
“As part of our company restructuring earlier this year, we asked some teams to conduct a scenario planning exercise looking at the potential impact of redeployments, open role closures and cuts,” Meta said in a statement, adding that leaders “didn’t move forward with every scenario from the exercise.”
Chasing the “AI native” model
The push traces back to ideas that took hold among Meta executives last year, inspired partly by trips to Asia where leaders including Chief Data Officer Alex Schultz and Head of Product Naomi Gleit studied how startups there had built entire org charts around AI tools.
Gleit told Reuters she spent considerable time in Meta’s Singapore office last year, and said the practices she saw there “inspired some of the teams in California and New York.” She described many of the resulting changes as bottom-up, reflecting shifts some employees had already begun making on their own.
One early pilot came from Ime Archibong, a longtime vice president of product management, who announced in July last year that his team was restructuring around small “tech pods” of two to three engineers and a designer, each equipped with AI tools. The pods ditched Meta’s traditional six-month planning cycles in favor of four-week sprints.
By October, an internal document called the “AI-Native Playbook” laid out a template for other teams to follow. Traditional titles for designers and engineers disappeared, replaced by a single generic role: “builder.” Middle-management layers were stripped out, with pods reporting instead to one high-level unit head, supported by what the document called “agent-assisted analysis.”
Meta declined to make Archibong available for comment or answer questions about the early pilots.
By June, at least 11 units across engineering and research had adopted the pod structure, according to internal announcements and people familiar with the arrangements. Traditional teams of 10 to 20 people with specialized roles gave way to pods of three to five “builders,” with a single direction lead and specialists shared across multiple pods.
Under the new model, unit heads known as “Org Leads” decided job performance ratings and promotions, each overseeing 30 to 50 people, with help from human-resources staff and what one internal post called “AI systems.” Pod leads managed day-to-day work but held no formal management authority. One staffer wrote on an internal message board that they weren’t getting manager training or access to standard management tools despite being asked to run a pod.
Meta told Reuters that teams “experimented in different ways with how to be more agile,” and that performance ratings and promotion decisions “were and are made by people, not AI.”
The backlash builds
Reuters first reported in March that Meta was weighing layoffs that could affect 20% or more of its staff, catching many employees off guard before executives had briefed them. A company spokesperson dismissed the report at the time as “speculative reporting about theoretical approaches.” Internally, senior leaders told managers to prepare teams for roles that would “evolve” because of AI, without confirming specific numbers.
In April, Reuters reported more detail: a first wave of roughly 10% cuts set for May 20, with further reductions planned for later in the year. Meta confirmed the 10% figure to staff, and Zuckerberg told employees the cuts stemmed from heavy capital spending.
Some engineers were reassigned during this period to a new Applied AI Engineering unit, where they built coding puzzles used to train Meta’s AI models on software engineering. Employees frequently described the work as tedious in internal posts. Meta said the unit’s output has since helped train an AI model the company released last month.
Reuters also reported that Meta had required tracking software on U.S. employees’ devices to record keystrokes and mouse movements, meant to teach AI agents how humans interact with computers. Many employees worried they were training their own replacements, and anger spread across Meta’s internal Workplace network. Some posted images of elephants as a nod to the layoffs nobody was formally discussing. Others clashed directly with Chief Technology Officer Andrew Bosworth, who defended the AI transformation in the comments. Meta declined to make Bosworth available for comment.
Employee sentiment, tracked in Meta’s half-year Pulse survey, fell from 74% favorable to 55% favorable during this stretch, and labor organizing efforts picked up momentum.
Signs the technology wasn’t delivering
Internal data complicated the case for pressing ahead. AI tools drove a 220% year-over-year jump in code changes to Meta’s internal platforms, Bosworth wrote in an early June post, but changes that actually reached users grew just 36%. Infrastructure teams flagged “reliability warning signs” tied to the AI coding surge as early as March, and an April post warned that unchecked AI agents were carrying out “large-scale, disruptive actions that humans are unlikely to execute.”
Major technical and security incidents, including service disruptions and possible data leaks, rose 40% year over year, with time spent responding to them up 70%, according to internal posts. In early June, hackers exploited Meta’s AI-powered customer support bot to access high-profile Instagram accounts, including one belonging to the dormant Obama White House page.
Meta declined to comment on the internal data describing the disruptions.
Zuckerberg pulls back
Facing mounting pressure from multiple directions, Zuckerberg met again with top lieutenants on the night of May 19 and scrapped the planned November layoffs. Meta proceeded with the 10% cut the next day, and Zuckerberg posted a memo telling remaining staff he did not expect further company-wide layoffs that year and wanted to offer more stability.
Executives followed with a round of morale-boosting steps: pausing the mouse-tracking program, letting some employees in the Applied AI Engineering unit return to their old teams, and sending Chief Financial Officer Susan Li to improve office perks, including snack quality and spending on travel and social events. Meta declined to make Li available for comment.
In early July, Zuckerberg told employees at a town hall that AI agent technology hadn’t “accelerated” as fast as he’d expected, though he predicted improvement within three to six months.
A new pitch: “betting on people”
Since then, Zuckerberg has shifted his public messaging, launching an ad campaign built around the phrase “betting on people” and publishing a 6,500-word essay on his vision for AI’s future. In an internal post in June, he told employees, “We are the only major company focusing on empowering people and putting the power of this new technology in billions of people’s hands across all our products.”
Even so, Zuckerberg’s public language has stuck closely to the terms “company-wide” and “this year,” leaving some employees to speculate about team-specific cuts, performance-based dismissals, or a delay of broader reductions into next year. Meta faces its own cash pressure as it plans to spend at least $130 billion on AI chips and infrastructure this year, a figure analysts expect will consume its operating cash for 2026, according to LSEG estimates.
In his essay, titled “The Future is for Everyone,” Zuckerberg predicted an eventual “abundance of jobs” even as individual companies shrink. “Company sizes may shrink, just as they did in the transition from industrial giants to tech companies,” he wrote. “But this doesn’t mean fewer jobs overall. It implies a larger number of companies with fewer people each.”
Reuters was unable to determine what ultimately prompted Zuckerberg to change course in May, or what Meta’s current plans are for reshaping its workforce going forward. The company declined to make Zuckerberg available for comment.
