How Zuckerberg's Plan to Replace Meta Staff With AI Failed

Mark Zuckerberg and his top lieutenants gathered at his Hawaii compound in January to hatch a radical plan reimagining Meta around artificial intelligence, one that would ultimately unravel within months amid employee revolt, security concerns, and disappointing results, according to Reuters' exclusive investigation, based on scores of internal documents, recordings, and interviews with more than 20 people familiar with the company's operations.

The plan, code-named Project OT for "Organization Transformation," envisioned an "AI native" future for the owner of Facebook and Instagram, in which AI agents would take over much of the daily work performed by thousands of human employees, with smaller, "talent-dense" cadres of remaining staff overseeing those virtual workers, according to an internal planning document reviewed by Reuters. In scenario-planning exercises, executives explored slashing the size of many teams across Meta by as much as 60%, with a human-resources executive projecting cuts that could rival the company's previous "Year of Efficiency" layoffs.

The plan's collapse came at the eleventh hour. On the night of May 19, just hours before the first wave of layoffs was set to begin, Zuckerberg abandoned plans for a second, larger company-wide cut originally planned for November, according to an internal document reviewed by Reuters. Meta proceeded with a 10% workforce reduction the following day, but scrapped the more sweeping restructuring, with Zuckerberg subsequently telling remaining employees he did "not expect other company-wide layoffs this year."

Project OT's Timeline

Three distinct pressures converged to sink the more ambitious version of the plan. By the time of the reversal, Meta employees were reportedly in open revolt, convinced the company's AI transformation initiatives were partly aimed at replacing them directly, according to Reuters' reporting. Simultaneously, internal data suggested the autonomous AI "agent" technology at the heart of the strategy was failing to deliver hoped-for productivity gains, while some investors were openly questioning what Meta had to show for its enormous AI spending.

The productivity data was genuinely striking. Internal figures cited by IBTimes' reporting on the investigation showed code changes to Meta's software platforms and infrastructure increased 220% year over year using AI tools, but changes resulting in new or upgraded user-facing features rose just 36%, a massive gap between AI-generated activity and actual product output. Compounding the problem, internal posts reportedly warned that AI agents were taking disruptive actions that human employees would have been unlikely to take, raising genuine reliability and security concerns.

Zuckerberg has since acknowledged the setback directly to staff. At an internal town hall captured by Reuters, he admitted "the trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected," according to Yahoo Finance's earlier reporting on the comments, a notable reversal from roughly two years of public predictions that AI would write most of Meta's code and replace mid-level engineers by 2025. Meta CTO Andrew Bosworth separately admitted in an internal memo that leadership did "an atrocious job explaining the vision," according to that same reporting.

This episode connects directly to the broader pattern of AI-driven layoff reversals we've tracked closely, including growing worker backlash as companies increasingly cite AI in job cuts and earlier examples of major employers like Ford and IBM quietly reversing AI-related workforce reductions after discovering automated systems couldn't fully replace human judgment.

This story is a genuinely important cautionary case study for any business currently planning aggressive AI-driven workforce restructuring, particularly one of the most well-resourced technology companies in the world, with roughly $145 billion committed to AI infrastructure, still hit a wall between ambitious AI automation plans and the technology's actual delivered capability.

For business leaders, the specific gap Meta's own data revealed, a 220% surge in AI-generated code activity translating into only a 36% increase in actual user-facing features, is worth internalizing directly. Raw AI output volume doesn't automatically convert into proportional business value, and companies restructuring around assumed AI productivity gains should build in genuine verification checkpoints before committing to workforce reductions based on projected, rather than proven, results.

What was Meta's "Project OT"?
Project OT, short for Organization Transformation, was a secret plan Meta leadership developed in January 2026 to reimagine the company as "AI native," with AI agents taking over much of the work currently performed by thousands of human employees.

Why did Zuckerberg cancel the larger round of AI-driven layoffs?
Zuckerberg abandoned the planned second, larger layoff wave the night before it was set to begin, following employee revolt, internal data showing disappointing AI agent productivity gains, and investor scrutiny over the company's AI spending.

Did Meta's AI actually increase productivity?
The results were mixed. Internal data showed AI-driven code changes to Meta's platforms increased 220% year over year, but changes resulting in new or upgraded user-facing features rose only 36%, revealing a significant gap between AI activity and actual product output.

Meta CEO Mark Zuckerberg developed a secret plan called Project OT in January 2026 to restructure the company around AI, exploring cuts to some teams of up to 60%, before abandoning the larger planned layoff wave the night before it was set to begin in May. The reversal followed employee revolt, internal data showing AI agent technology failing to deliver expected productivity gains, and investor questions about Meta's massive AI spending. Zuckerberg later publicly admitted agentic AI development had stalled over several months, a notable reversal from roughly two years of predictions that AI would replace much of the company's engineering workforce.