Artificial Intelligencer: How Meta’s AI workforce transformation plans went kaput

Intuit

26/08/2026 20:14

By Greg Bensinger
Aug 26 (Reuters) - Ask just about anyone in Silicon Valley and they’ll tell you that AI is coming for their jobs. Software coders laboring to improve AI are effectively working to make their jobs obsolete.
The doomsaying is not just idle talk. Some 113,000 job cuts have been tied to AI this year, according to executive search firm Challenger, Gray and Christmas.
Amazon AMZN.O gave AI credit for a big chunk of its 30,000-worker reduction starting last October. Jack Dorsey, CEO of Square-parent Block, said a February 40% cut to staff was because AI “fundamentally changes what it means to build and run a company.” PayPal PYPL.O sent nearly 5,000 workers packing in May to “accelerate AI adoption and automation.”
The list goes on. Dell DELL.N, Visa V.N, Citigroup C.N and Intuit INTU.O are among dozens of others pulling back on paying people in favor of paying for greater AI use.
Corporations have been particularly encouraged by emerging AI software that can perform functions autonomously, known as agents, and so-called vibe coding software that can spin up new websites, apps and other services with a minimum of human input.
The exuberance over job cuts can be traced back, at least in part, to Elon Musk’s gutting of Twitter, now X, when he took it private in 2022. The billionaire eventually laid off about four out of every five workers at the social media firm and, to some technologists’ astonishment, maintained a largely functioning website.
Not all of the jobs lost can be attributed to workers competing directly with software. Many firms are committing massive capital expenditures toward building out infrastructure, such as data centers and chip manufacturing, necessary to keep the wheels of AI spinning around the clock. The money has to come from somewhere.
Oracle ORCL.N, for instance, shed 21,000 jobs in its fiscal 2026, citing infrastructure as a primary driver. Microsoft MSFT.O axed about 4,800 workers, primarily in its Xbox game division, to help fund its AI ambitions.
Many of these same companies are reporting ballooning profits and sales. Take Meta META.O, for instance. The Facebook and Instagram parent boosted its net income in this year’s first six months by 22%. It was also busy contesting allegations in federal court this month that it made its products addictive for teenagers, before reaching a $17 billion settlement. 
CEO Mark Zuckerberg drew up a secret plan to cut some teams by as much as 60% in an effort to make his workforce “AI native,” my colleague Katie Paul reports. Read on to learn how the effort, known as Project OT, failed to launch.
     
OUR LATEST REPORTING ON TECH AND AI
Mark Zuckerberg had a bold plan to replace Meta staff with AI. It imploded.
Meta agrees to pay $18 billion to settle US lawsuits over children's social media addiction
Instagram head Mosseri, at children’s addiction trial, says few teenagers knew of safety feature
The AI founders who walked away from Bezos-backed Prometheus to model the universe
AI-generated music barred from Australian charts after Madonna cover controversy
Nvidia shares set for $280 billion price swing after earnings, options show
China's Moonshot in talks with Microsoft, Amazon, Google over K3 revenue sharing, sources say
China's Robot Olympics ends with humanoid's 8.64-second 100m, faster than Bolt  
     
WORKER BACKLASH
Meta’s plan was perhaps the most ambitious AI workforce overhaul attempted by a big corporate player to date, making it a bellwether case for how far the technology can be pushed. If Project OT had succeeded, Zuckerberg would have created a template for the future of work, much as Henry Ford did at the tail end of the Industrial Revolution or as tech CEOs did a generation ago in convincing other executives to embrace jeans and ping-pong tables in offices.

For now, however, his vision has been a dud. Not to be outdone by rivals, Meta pared around 10% of its employees in May, but called off plans for a second “wave” of the restructuring the night before those layoffs, Paul reported, citing internal documents, interviews and recordings. The Menlo Park, California, company has said it expects to invest $130 billion or more in AI infrastructure and chips this year alone.
The plan included breaking down teams into “pods” with fewer employees and less management to move faster. AI tools would enable those small pods “to explore more ideas with less cost and higher fidelity,” wrote one executive in an internal memo.
Human resources was enlisted to find employees who were irreplaceable and could, alone, do the work of 10 corporate employees, supercharged in part by AI.
But Meta’s real human employees staged something of a revolt, affronted by a host of internal AI efforts that had gone sideways (not to mention Reuters reports informing them of the coming layoffs long before executives intended for them to know). Among them was Meta’s intent to log employees’ keystrokes and mouse movements to inform its AI programs. The firm’s AI agents and coding tools were also causing reliability and security problems, and not boosting employee productivity as much as expected. Zuckerberg said in a Meta town hall last month that agentic tech was not proceeding at the pace he had hoped, though he added that he believed it would improve in the next three to six months.
In response to the reporting, Meta acknowledged the existence of Project OT and said the plan encompassed reassigning some workers and would not impact all teams at the company, calling it a “scenario planning exercise.”
Ultimately, Zuckerberg scaled back Project OT, which stands for Organization Transformation.  He’ll likely be looking for other ways to transform the organization’s workforce in the coming months.

(Reporting by Greg Bensinger; Editing by Ken Li and Lisa Shumaker)
(( greg.bensinger@thomsonreuters.com ))