SUCKER-HILL, THE FILTHY BEAST.
Mark Zuckerberg had a bold plan to replace Meta staff with AI. Here’s how it imploded.
To make its workforce “AI native,” Meta explored slashing the size of many teams across the company by as much as 60% in two waves, internal documents show. Staff revolted - and hours before a big layoff in May, the CEO called off planning for further cuts.
In January, Meta CEO Mark Zuckerberg and his top lieutenants gathered for their annual leadership retreat at his Hawaii compound. There they hatched a radical plan to reimagine work at the social-media giant in the age of artificial intelligence.
Code-named Project OT – short for Organization Transformation – the plan envisioned an “AI native” future for the owner of Facebook and Instagram. AI would take over much of the daily work performed by thousands of human employees. Virtual workers would be overseen inside Meta by smaller, “talent-dense” cadres of human staffers, according to one internal planning document reviewed by Reuters and three people familiar with the project.
In scenario-planning exercises, two of these people said, executives explored slashing the size of many teams across Meta by as much as 60%. Some employees would be offered roles in new units, while others would be laid off as part of a culling that one human-resources executive projected would be as big as or bigger than the company’s cuts of around 25% three years ago, according to another internal document.
The restructuring would be carried out in two “waves,” beginning with a first purge in May and followed by another shake-up in November, internal planning documents seen by Reuters showed. Layoffs would be supplemented with the closing of open positions and pushing people out who Meta believed were poor performers. These and other details of the plan, including the scale of the restructuring, haven’t been previously reported.
AI would take over much of the daily work performed by thousands of human employees. Virtual workers would be overseen inside Meta by smaller, “talent-dense” cadres of human staffers.
But on the night of May 19, just hours before the first layoff wave, Zuckerberg blinked. Meta laid off 10% of its employees the next day, but it called off planning for the November cuts, according to one internal document reviewed by Reuters.
By then, Meta employees were in open revolt, convinced that the company’s AI transformation initiatives were partly aimed at replacing them. Internal data was also suggesting that autonomous AI “agent” technology at the heart of the strategy was failing to deliver hoped-for productivity gains. Some investors were questioning what Meta had to show for its gargantuan spending on AI.
This article reveals for the first time the rapid pace of the cuts Meta was considering, the thinking behind the plans and how they quickly unraveled. Based on scores of internal documents, posts and recordings reviewed by Reuters, as well as conversations with more than 20 people with knowledge of Meta’s inner workings, the reporting shows how the social-media giant attempted to position itself at the forefront of an AI-driven workplace overhaul, only to stumble in the execution.
In response to this story, Meta confirmed the existence of Project OT, which it described as a year-long project focused on cost cutting, redesigning team structures and shifting staff into new priority areas, such as producing training data for its AI models.
The company acknowledged the plan was to be carried out in two waves and that the most drastic scenarios involved reducing the size of some of Meta’s teams by up to 60%. But it declined to specify which units were involved and said several major ones weren’t part of it. Meta said the scenarios included both layoffs and redeployments, and that the company at no point intended to lay off 60% of its entire workforce. Company leaders canceled plans for the second wave before determining how many people overall would lose their jobs, it said.
“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. “This ultimately resulted in moving thousands of employees to do priority work on several newly-established teams, as has been publicly reported. Ultimately, we didn’t move forward with every scenario from the exercise – and it was never assumed we would.”
Reuters was unable to determine what exactly prompted Zuckerberg to shift course or what the social-media giant’s current plans are for continuing to reshape its workforce. Meta declined to make Zuckerberg available for comment.
THE ‘AI-NATIVE PLAYBOOK’
Ever since the release of OpenAI’s ChatGPT in late 2022, Silicon Valley leaders have been imagining the revolutionary possibilities that generative AI might unleash for the future of work. They have been especially intrigued by the promise of “agentic” capabilities. Agents, a nascent technology, are being designed to do more than just spit out answers to queries like typical AI chatbots. Instead, they are supposed to take autonomous actions such as shopping, booking travel and creating new apps.
At Meta, executives last year threw themselves into AI-inspired management philosophies. They were captivated by ideas emerging out of the startup world about how companies shouldn’t just incorporate AI into their existing processes, but instead go fully “AI native.” As one Project OT document envisioned it: “AI-ready tools and agents interact, workflows are automated, new builds are AI-first.” The strategy also included having Meta sell AI agents, which would perform tasks like scheduling appointments and closing sales, to other companies.
Meta executives, including Chief Data Officer Alex Schultz and Head of Product Naomi Gleit, visited Asia last year and admired how startups there had built their organizational charts around AI, according to three people familiar with the trips. Meta executives also commissioned their own research into how AI startups were organized and set up pilot projects to determine what being “AI native” would mean at the company, according to one source familiar with the research and internal documents describing the pilots.
Asked in June about the genesis of Meta’s AI workforce transformation, Gleit told Reuters she had spent “quite a bit of time” in Meta’s Singapore office last year and said the practices there had “inspired some of the teams in California and New York.” She said many of the ideas were “bottom-up,” reflecting how some Meta employees were already starting to implement changes themselves.
Ime Archibong, a long-time vice president of product management, announced one of Meta’s earliest pilot projects in July last year. In an internal post, he described how his team was taking its first step to modernize product development, describing the transition in sports terms.
“In basketball, embracing the fast break lets you take more shots – and better ones. We expect the same with AI tools: they allow us to explore more ideas with less cost and higher fidelity,” Archibong wrote. He predicted that prototypes created quickly with AI would more closely resemble finished products than before. “This isn’t just more fun – in an AI-first era, we believe it’s the winning strategy.” Zuckerberg, too, alluded to the potential for AI to make work “a lot more fun” in an earnings call six months later.
In practical terms, Archibong’s pilot involved setting up five “small tech pods,” each consisting of two to three engineers and a designer, all equipped with AI tools. The groups would dispense with established processes for releasing new products, such as fixed six-month planning cycles. Instead, they would aim to build prototypes in four-week “sprints,” he wrote in his post.
In October, a member of Archibong’s team previewed a broader roll-out of the approach. In an internal post titled “AI-Native Playbook,” the person laid out a guide for other teams to “make the leap.” Traditional roles for product designers and engineers would vanish, and tech pod members would get a new generic title: “builder.” Layers of middle management would be eliminated. Instead, pods would report up to a single, high-level unit head. “Agent-assisted analysis” would help set day-to-day priorities.
Meta declined to make Archibong available for comment or to answer questions about its early “AI native” pilots.
Early this year, Zuckerberg set Project OT in motion and directed executives to proceed with the changes to management structures. By June, at least 11 units, including engineering and research teams, had implemented small pods, according to internal announcements and five people familiar with the arrangements.
Teams & Roles
Traditional Product Development
10-20 people
Roles are specialized with clear expectations for each role.
7-14 ENG
1 PM
1 DES
1 DS
1 UXR
1 DE
Modern (AI-Native) Product Development
3-5 people
Roles are fluid and oriented around doing what’s needed to build and 1 lead to help provide direction. Specialists are pooled.
3-4 Builders (typ. eng)
1 Direction Lead (can be any XFN)
Shared Specialists XFN (DES/UXR/DS/DE/ML) pooled across pods
Org Structure
Traditional Product Development
Hierarchical, multi-layered
Organizations have a few large pillars that each have a few large teams, and so on, so forth.
Modern (AI-Native) Product Development
Flatter, more nimble pods
Organizations are managing many small pods that cover similar or more surface area.
Description of changes to team structures from an internal Meta post titled “AI-Native Playbook.” ENG refers to engineers; PM is product manager; DES means designer; DS is data scientist; UXR is user experience researcher; DE is data engineer; XFN is cross-functional; ML is machine learning.
With the changes, Meta introduced a “village approach” to managing people, according to an internal announcement from one unit. Job performance ratings and promotions would be decided by high-level unit heads, or “Org Leads,” supported by human-resources personnel and unspecified “AI systems,” the post said. Each unit head would oversee between 30 and 50 people, while “Pod Leads” would steer small pods day-to-day, but have no formal management authority, the sources and announcements said.
One Meta staffer assigned to manage a pod expressed confusion about the new setup on an internal company message board, according to a copy of the post. “I’m not going through manager training, and I’m not getting access to ratings & manager tools,” the person wrote.
Meta told Reuters that teams “experimented in different ways with how to be more agile.” The company declined to explain the post’s reference to “AI systems,” but said, “Performance rating and promotion decisions were and are made by people, not AI.”
Around the same time, Meta also launched a new human resources tool to identify “Irreplaceable Talent,” according to an internal document seen by Reuters. The document referred to a hypothetical “10X Performer” – a nod to the decades-old Silicon Valley archetype of an engineer who can do the work of 10 people. The term is back in vogue now, as many AI devotees believe the technology will enable extremely talented individuals to perform the work of entire teams. According to two people familiar with the matter, Meta intended to spend some of its savings from layoffs on giant pay packages to attract and retain such talent, especially AI engineering rock stars.
EMPLOYEE REBELLION
On March 13, before many leaders at the vice-president level had even been briefed on Project OT, Reuters reported that Meta was planning layoffs that could affect 20% or more of its workforce. It wouldn’t have been the first culling on that scale: Between late 2022 and early 2023, Meta slashed around 25% of its staff.
Still, the report caused alarm. Many rank-and-file employees were rattled. And they weren’t yet supposed to know about the cuts, annoying Zuckerberg and leaving executives unprepared for the backlash, according to a person familiar with the matter. A Meta spokesperson at the time called the story “speculative reporting about theoretical approaches.”
Internally, high-level executives hunkered down. They opted not to discuss the article with the rank-and-file and quietly dismissed it to senior managers, instructing them to tell their teams they should expect roles to “evolve” as a result of AI, according to a talking-points document seen by Reuters.
In April, Reuters published more details: Meta was set to cut about 10% of its workforce in a first wave of layoffs on May 20 and was aiming to shed more staff in the second half of the year. Meta soon confirmed the 10% reduction plan to staff, and Zuckerberg later told employees the cuts were due to heavy capital expenditures.
Executives had started moving ahead with other aspects of the plan. Some engineers, for example, were reassigned to a new Applied AI Engineering unit tasked with creating software-engineering puzzles to be used as training data to improve the coding skills of Meta’s AI models. Many staffers derided the work as rote and boring in internal posts seen by Reuters.
Meta told Reuters that the company is beginning to see progress with its redeployments, noting that data produced by the Applied AI Engineering unit helped train an AI model it released last month.
Between the reassignments and the job cuts, headcount in some engineering units dropped as much as 30% by the end of May, according to one document and people familiar with the planning.
Reuters also reported in April that Meta had mandated that tracking software be installed on U.S. employees’ devices to capture their keystrokes and mouse clicks to teach its AI agents to replicate how humans interact with computers. Believing they might be training their own AI replacements, and irate over the lack of detail about the layoffs, many employees flooded Meta’s in-house communications network – called Workplace – with angry screeds and gallows humor.
Staffers replied to executives’ internal posts with pictures of elephants, symbolizing that layoffs were the elephant in the room, according to examples seen by Reuters. Some employees directly clashed with Andrew Bosworth, Meta’s chief technology officer, who oversaw the AI transformation and was defending it in posted comments. One person trolled Zuckerberg, sarcastically likening his internal announcement of an AI initiative for small businesses to Prometheus, the figure from Greek mythology, giving fire to humanity. Meta declined to make Bosworth available for comment.
Staff morale sank. Meta’s internal measure of employee sentiment dropped from 74% favorable to 55% favorable, according to the company’s half-year Pulse survey. Labor organizing efforts gained momentum.
Amid the rebellion, other internal data indicated that the tech at the heart of the plan wasn’t delivering. Employees’ use of AI had resulted in a vast increase in the code they generated, but with questionable impact on productivity, according to internal posts seen by Reuters. For instance, code changes made to the internal software platforms and infrastructure employees use on the job were up 220% year-over-year, according to a post by Bosworth in early June. But changes that led to new or upgraded features reaching Meta users were only up 36%.
As early as March, infrastructure teams were also warning of “reliability warning signs” caused by the AI coding surge, according to an internal post. Another post, in April, said that unchecked AI agents were performing “large-scale, disruptive actions that humans are unlikely to execute.” The result: Major technical and security incidents, such as service disruptions and possible data leaks, spiked 40% from the previous year, with the time staffers had to spend “firefighting” them up 70%, according to the internal posts.
The public saw a glimpse of the problem in early June after hackers exploited Meta’s new AI-powered customer support bot to gain access to high-profile Instagram accounts, including one for the dormant Obama White House page.
Meta declined to comment on the internal data about the AI disruptions.
With pressures piling up, Zuckerberg changed course. Hours before Meta carried out its first wave of Project OT layoffs on May 20, he conferred again with his top lieutenants and called off the second wave of restructuring planned for November. The next morning, Meta went ahead with the 10% cut. But after the notifications went out, Zuckerberg posted a memo on Workplace telling Meta’s remaining employees he did “not expect other company-wide layoffs this year” and expressed his desire to give them more “stability.”
Executives then tried to shore up morale. They encouraged unit heads to acknowledge employees’ feelings, paused the mouse-tracking program, let some employees in the new AI engineering unit transfer back to their old teams and dispatched Chief Financial Officer Susan Li to boost perks. In the weeks after the layoffs, a stream of empathetic executive posts arrived on Workplace, along with pledges to improve snack quality in office microkitchens and increase spending on travel and social events. Meta declined to make Li available for comment for this report.
In early July, Zuckerberg made a surprise appearance at an internal town hall and conceded miscalculations on the reorganization’s timing. AI agent technology, he said, had not “accelerated” as quickly as he had anticipated. He added that he expected the tech to improve and begin showing more benefits in the next three to six months.
Excerpts from Mark Zuckerberg’s town hall with employees on April 30, 2026, in which he describes his thinking behind the company’s reorganization and coming layoffs.
‘BETTING ON PEOPLE’
As Zuckerberg has dialed back the most disruptive aspects of the internal AI transformation, he has also launched a public-relations blitz positioning Meta as people-centric. The push included a video advertisement proclaiming the company is “betting on people.”
In the campaign, which also included a recent 6,500-word essay outlining his vision for the future of AI, the CEO has highlighted Meta’s plans to make creating AI agents more user friendly while portraying unnamed competitors as the true job killers. 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 – rather than primarily focusing on automating work.”
Excerpt from a Meta video advertisement proclaiming the company is “betting on people.”
Still, Zuckerberg has stuck to the words “company-wide” and “this year” in discussing layoffs with employees, according to his internal communications. That has prompted some employees to speculate that he’ll continue trimming the ranks via team-specific cuts or performance-based dismissals – or delay company-wide headcount reductions until next year. The company is facing a cash squeeze and investor scrutiny due to its dizzying AI spending, heightening the pressure to find savings. Meta plans to invest at least $130 billion in AI chips and other infrastructure this year, which analysts expect will eat up its operating cash for 2026, estimates from LSEG show.
“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 — rather than primarily focusing on automating work.”
From an internal post Mark Zuckerberg shared with Meta employees in June
In his essay on the future of AI, titled “The Future is for Everyone,” Zuckerberg predicts there could be “an abundance of jobs in the future.” But some companies might end up with fewer employees.
“Company sizes may shrink – just as they did in the transition from industrial giants to tech companies,” he writes. “But this doesn’t mean fewer jobs overall. It implies a larger number of companies with fewer people each.”
Learn about the latest breakthroughs in AI and tech with the Reuters Artificial Intelligencer newsletter. Sign up here.
Katie Paul reported from New York. Additional reporting by Jeff Horwitz in Oakland. Illustration by Catherine Tai. Design by John Emerson. Edited by Steve Stecklow and Kenneth Li.
Comments
Post a Comment