In January, Mark Zuckerberg gathered his senior leadership team at his Hawaiian estate with a bold vision. The Facebook and Instagram owner would become “AI native” – a company where artificial intelligence handles the bulk of daily operations while human staff oversee virtual workers in streamlined teams. Internal documents seen by this publication reveal Meta considered slashing numerous teams by as much as 60 per cent. The restructuring would unfold in two phases, beginning with mass layoffs in May. But on the eve of the first wave, Zuckerberg reversed course. The May cuts proceeded, yet the second phase was cancelled. The social-media giant’s grand experiment in AI-driven workplace transformation had unravelled faster than anticipated.
The Blueprint for an AI-First Workforce
The initiative, codenamed Project OT (short for Organisation Transformation), represented an attempt to fundamentally reshape how Meta operates. Executives explored scenarios where AI would assume responsibilities currently handled by thousands of employees. Smaller, highly skilled human cohorts would manage these digital workers – a model the company described internally as “talent-dense.”
Meta confirmed the programme’s existence when approached for comment, describing it as a year-long effort focused on cost reduction, restructuring teams, and redirecting staff toward priority areas such as developing training data for AI systems. The company acknowledged that scenarios under consideration included reducing certain units by up to 60 per cent, though it stressed these were hypothetical exercises and denied any intention to cut the overall workforce by such a margin.
Silicon Valley had been captivated by the possibilities since OpenAI released ChatGPT in late 2022. Meta’s leadership grew particularly interested in “agentic” AI – systems capable of autonomous action rather than simply responding to queries. Internal presentations envisioned a future where “AI-ready tools and agents interact, workflows are automated, new builds are AI-first.”
Executives including Chief Data Officer Alex Schultz and Head of Product Naomi Gleit visited Asian startups last year, studying how some companies had restructured entirely around AI capabilities. Gleit later acknowledged that practices observed in Meta’s Singapore office had influenced changes implemented in California and New York.
Resistance and Technical Hurdles
By early 2024, the rollout had begun. At least eleven units, spanning engineering and research, adopted the “small tech pod” model. Each pod comprised two to three engineers and a designer, equipped with AI tools and freed from traditional product-development processes. Fixed six-month planning cycles were abandoned in favour of four-week development sprints.
The restructuring brought sweeping management changes. Middle-management layers disappeared, with pods reporting directly to senior “Org Leads” overseeing between 30 and 50 people each. Performance reviews would incorporate input from “AI systems,” according to internal announcements, though Meta insisted all promotion decisions remained with human managers.
Employees grew increasingly sceptical. Staff flooded Meta’s internal communications platform with complaints and sardonic memes – pictures of elephants, symbolising the “elephant in the room” they felt executives were ignoring. Morale measurements obtained by this publication show employee sentiment dropped from 74 per cent favourable to 55 per cent in the company’s half-year survey.
Technical problems compounded the difficulties. While AI-assisted coding increased code output by 220 per cent year-on-year, the number of new or improved features reaching users rose by just 36 per cent. Infrastructure teams warned of reliability issues stemming from AI-generated code. Service disruptions and potential data breaches surged 40 per cent compared to the previous year, with staff spending 70 per cent more time addressing emergencies.
A public illustration came in June when hackers exploited Meta’s AI-powered customer support bot to access high-profile Instagram accounts, including the dormant Obama White House page.
The Retreat and Future Uncertainties
On the evening of 19 May, hours before Meta announced 10 per cent workforce cuts, executives convened. Zuckerberg decided to abandon the planned November restructuring. The following morning, notifications went out to affected employees. Within hours, Zuckerberg posted on the company’s internal network promising there would be no further company-wide layoffs that year and expressing a desire to provide staff with greater stability.
Subsequent efforts to rebuild confidence included pausing keystroke-tracking software installed on employee devices, allowing some workers to transfer back from the new AI engineering unit, and dispatching Chief Financial Officer Susan Li to announce enhanced perks.
In July, Zuckerberg admitted to miscalculations during an internal town hall. AI agent technology, he said, had not advanced as swiftly as anticipated. He expected meaningful improvements within three to six months.
Despite scaling back the most disruptive elements, Zuckerberg has maintained careful language about future cuts. His public statements reference “company-wide” reductions and limit them to “this year.” Some employees interpret this as suggesting team-specific layoffs or performance-based dismissals could continue.
The pressure to reduce costs remains intense. Meta plans to invest at least $130 billion in AI infrastructure during 2025, a sum analysts expect will consume the company’s operating cash through 2026.
Why it Matters
Meta’s aborted transformation offers a cautionary tale about the gap between AI’s theoretical promise and practical reality. The company discovered that automating code generation does not automatically translate into delivering better products for users – a distinction that proved costly in both technical reliability and employee morale. For workers across the technology sector, the episode illustrates that while executives may envision leaner organisations powered by artificial intelligence, the path toward that future remains fraught with obstacles. Zuckerberg’s own essay on AI’s future suggests company sizes may indeed shrink, but whether that benefits shareholders or workers remains an open question. Meta’s experience suggests the transition will be messier than many in Silicon Valley would prefer to admit.
