Reuters Deep Dive: Zuckerberg's "Full AI Transformation" Hits a Snag, Code Crisis Forces Emergency Halt of Layoff Plans

Original Author: Claude, TechFlow

TechFlow Intro: A Reuters investigative report published on August 26 revealed that Meta internally simulated an "AI-native" restructuring in early 2026: some teams could see staffing cuts of up to 60%, with AI agents taking over daily work, while smaller "high-density talent" groups supervise the virtual employees. On the night of May 19, Zuckerberg called off the second wave originally scheduled for November. The next day, the company still laid off approximately 10% of its staff. Internal data shows code changes increased 220% year-over-year, but features delivered to users only increased 36%; major technical and security incidents rose 40% year-over-year, and employee favorability dropped from 74% to 55%.

Reuters published an investigative report on August 26: Zuckerberg had defined an organizational overhaul using AI to replace a significant portion of daily human labor, codenamed Project OT (Organization Transformation). In the most aggressive scenario, many team sizes could be reduced to 40% of their original. The plan was set to roll out in two waves, in May and November. On the night of May 19, he halted the planning for the second wave.

Reuters reviewed dozens of internal documents, posts, and recordings, and spoke with more than 20 informed sources. Meta acknowledged the year-long project existed, but characterized it as "scenario planning," emphasizing it did not mean laying off 60% of the entire company and that "not every scenario was pursued."

Hawaii Closed-Door Meeting Sets the Tone; Some Teams Modeled Cuts of Up to 60%

In January 2026, Zuckerberg brought his core leadership team to his Hawaii estate for the annual leadership offsite. Reuters, citing informed sources, reported that the direction set at the meeting was to make Meta an "AI-native" company. An internal planning document stated that AI would take over a significant portion of daily work for thousands of employees, with virtual workers overseen internally by smaller, more "talent-dense" human teams.

In the scenario planning, multiple teams could see staffing cuts of up to 60%. Some people would be transferred to new units, while others would be let go. One HR executive predicted in another internal document that the scale of this round of cuts could reach or even exceed the company's roughly 25% layoff from about three years prior. The restructuring plan was set for two waves: the first cut in May, and another in November, along with closing open positions and removing those deemed low-performing.

These scale details had not been previously reported publicly. Meta's written response to Reuters was clear: the most aggressive scenarios did include reducing some teams by up to 60%, with the path involving both layoffs and transfers; the company never intended to cut 60% of the entire group; several key business units were out of scope. The statement read: "As part of the company restructuring earlier this year, we asked some teams to do scenario planning to assess the potential impact of transfers, closing open roles, and layoffs." "This ultimately led to thousands of employees moving to several newly formed teams to work on priority projects, which has been publicly reported. We did not pursue every scenario in the planning — and never assumed all of them would be implemented."

Reuters could not confirm a single trigger for Zuckerberg's change of heart that night, and the company declined to arrange an interview with him.

Team Sizes Shrink from a Dozen to Three to Five; "Builders" Replace Old Roles

Since ChatGPT launched in late 2022, Silicon Valley management has been fascinated with "agents": they shouldn't just answer questions, but should be able to place orders, book flights, and write applications themselves. Internally, Meta codified this philosophy as "AI-native" — a Project OT document stated: "AI-ready tools and agents work together, workflows are automated, and new projects default to AI-first." Externally, the company also planned to sell business agents capable of booking and closing deals to other enterprises, having already launched an enterprise-facing version in June.

Product head Naomi Gleit spent a "considerable amount of time" in the Singapore office last year. She told Reuters in June that the practices there "inspired some teams in California and New York," with many changes being "bottom-up." Chief Data Officer Alex Schultz also visited Asia last year; according to three informed sources, the group appreciated how local startups were redrawing their organizational structures around AI.

An earlier pilot took place in July 2025. Ime Archibong, a long-time vice president of product management, announced in an internal post that his team would first build five "small technical squads," each with two or three engineers and a designer, equipped with AI tools, using four-week sprints to replace the original roughly six-month product cycle. He used a basketball analogy: "In basketball, embracing the fast break allows you to get more shots up, and better shots. That's our expectation for AI tools: they let us explore more ideas at lower cost and with higher fidelity." His next sentence: "This isn't just more fun — in the AI-first era, we believe this is the winning strategy."

In October 2025, people on his team posted the "AI-Native Playbook," a guide for other groups on "how to make the transition": the traditional roles of product designers and engineers disappear, squad members are uniformly renamed "Builders"; middle management is dismantled; squads report directly to a senior unit lead; daily priorities are delegated to "agent-assisted analysis."

Earlier this year, Zuckerberg pushed Project OT into implementation, requiring management to change structures. By June, at least 11 units (including engineering and research) had already switched to the squad model. One internal notice stated that performance ratings and promotions would now follow a "village method": decisions are made by unit leads managing 30 to 50 people, assisted by HR and unnamed "AI systems"; squad leads only handle day-to-day matters without formal management authority. One employee assigned to lead a squad wrote on an internal message board: "I didn't go through management training, and I don't have access to ratings or management tools." Meta emphasized to Reuters that ratings and promotions "were and are human decisions, not AI."

Traditional product groups typically have 10 to 20 people: seven to fourteen engineers, plus one each for product, design, data science, UX research, and data engineering. The AI-native squads in the design mockups only have 3 to 5 people. HR is simultaneously looking for "irreplaceable talent" and "10x producers" — theoretically, one person plus AI can match ten people, and the money saved from layoffs is used to give them raises.

Code Changes Up 220%, But Features Reaching Users Only Up 36%

The centerpiece of the plan was autonomous agents. Internal data failed to support this pillar.

Chief Technology Officer Andrew Bosworth wrote in an internal post in early June: for the internal software platforms and infrastructure employees use daily, code changes increased 220% year-over-year; changes that actually became new features or upgrades delivered to Meta's users increased only 36% year-over-year. The infrastructure team had already warned in March that the surge of AI-generated code was creating "reliability alerts." Another internal post in April put it more bluntly: unconstrained agents were executing "large-scale, destructive operations that humans would be unlikely to perform."

The results were documented in the same batch of internal posts. Major technical and security incidents (service outages, potential data leaks, etc.) rose 40% year-over-year, and employee time spent "fighting fires" increased 70% year-over-year. In early June, attackers also used Meta's AI customer service bot to compromise a number of high-profile Instagram accounts, which Reuters named, including the deactivated Obama White House account. Meta declined to comment on these internal incident data.

At an internal town hall in early July, Zuckerberg admitted the timeline was miscalculated. In a recording heard by Reuters, he said, "for at least the past four months, the trajectory of agent development hasn't accelerated the way we expected," and that the bet on the company's new structure "hasn't paid off yet either." He also said he hoped to see more obvious returns in the next three to six months. At the same event, he admitted the restructuring was "not clean," and that during planning in January and February, senior leaders' concern was that "we weren't adapting fast enough."

Investors were asking about a different issue. Meta's capital expenditure guidance for this year has been revised to $130 billion to $145 billion, primarily for AI chips and infrastructure. The Reuters investigative report wrote "at least $130 billion," and cited LSEG estimates that this spending is expected to consume the company's 2026 operating cash flow. The second-quarter earnings report subsequently confirmed the pressure: free cash flow declined 91% year-over-year.

Employees Saw Monitoring and Transfers as Bodysnatcher Training; Favorability Dropped from 74% to 55%

By the time the May 20 cuts landed, employees no longer believed the "transformation" was just about changing tools. Reuters first reported on March 13 that the company was discussing layoffs of 20% or more, and then in April reported the first wave of roughly 10% on May 20 and the possibility of further cuts later in the year. Informed sources told Reuters that the reports disrupted management's original communication timeline.

Alongside the layoffs, a large number of engineers were pulled into new units like Applied AI Engineering to work on training data. Earlier Reuters reports noted that the May global layoffs affected about 10%, roughly 8,000 people, with another approximately 7,000 transferred to new teams related to AI workflows. At the end of Q2, the company had approximately 75,500 employees. Some employees forced to transfer internally called themselves "conscripts." Devices in the US were also required to install keyboard-mouse tracking to capture keystrokes and mouse movements to train agents that could mimic human computer operation. Employees posted angry messages and sarcastic jokes on the internal Workplace platform; some replied to executive posts with images of elephants, hinting that the layoffs were the elephant in the room that couldn't be discussed. Earlier reports stated that over 1,600 people had signed a petition against the monitoring.

In the semi-annual Pulse survey, employee favorability dropped from 74% to 55%. Labor organizing efforts were also intensifying. On the night of May 19, the planning for the second wave was halted. The next day, the first wave proceeded as scheduled. Zuckerberg subsequently told those who remained that he "doesn't anticipate further company-wide layoffs this year," hoping to provide a bit more "stability." The company later paused the keyboard-mouse tracking, allowed some of those pulled to return to their original teams, and increased budgets for travel, team-building, and snacks. Employees in internal communications fixated on the two qualifiers he repeatedly used — "company-wide" and "this year" — speculating that headcount would continue to decline through team cuts or performance-based removals, or that the next round would be pushed to next year.

In June, Zuckerberg attempted to shift the external narrative back to "betting on people" in an internal post: "We are the only major company focused on empowering people and putting this new technology into the hands of billions of users, rather than primarily using it to automate work."

The Reuters investigative report places this statement after the plan had already been scaled back. Currently, Meta's official stance on this investigation remains that "not all of the scenarios in the planning were implemented."