Zuckerberg tells Meta staff agentic AI push is behind schedule, four months after restructuring

In a July 2 town hall recording obtained by Reuters, the CEO conceded the reorganization that cut roughly 8,000 jobs has not accelerated Meta's agent work — even as the company heads toward $145 billion in AI infrastructure spending this year.

Mark Zuckerberg told Meta employees at a July 2 internal town hall that the company’s agentic AI push is running behind expectations, four months after a restructuring designed explicitly to speed it up. “The kind of trajectory of the agentic development over at least the last four months hasn’t really accelerated in the way that we expected,” he said, according to a recording obtained by Reuters. The reorganization, he conceded, has “haven’t come to fruition yet.”

The math on that reorganization is worth stating plainly. Meta cut roughly 8,000 corporate jobs, about 10% of the workforce, across January and February. Some 7,000 employees were reassigned into new AI groups, one of them internally named Agent Transformation, per TechCrunch citing earlier Bloomberg reporting. The company is projected to spend up to $145 billion on AI infrastructure in 2026, part of a Big Tech capex wave exceeding $700 billion. And the CEO is telling staff, on tape, that the internal justification isn’t landing yet.

Zuckerberg framed the miscalculation as a planning-stage error. Executives had been “super optimistic” about tools like Anthropic’s Claude Code when they mapped out the restructuring, he said, and “our top people” “were worried that we weren’t going to move fast enough to adapt.” He also acknowledged the job cuts hadn’t been as “clean” as they should’ve been. A Meta spokesperson declined to comment.

The forward guidance is a three-to-six-month window in which Zuckerberg expects “more significant benefits” from the AI spend, per The Next Web. No products named. No teams named. In May he’d told employees he didn’t anticipate further companywide layoffs this year.

The broader industry context sharpens the pressure. In May, Five Eyes cybersecurity agencies including CISA, the NSA, and the UK’s NCSC issued joint guidance on deploying agentic systems, cataloguing 23 distinct risks and more than 100 recommended practices, and urging organizations to prioritize “resilience, reversibility and risk containment over efficiency gains.” Meta’s town hall admission and that guidance describe the same technology from opposite ends of the incentive stack.

Zuckerberg tells Meta staff agentic AI push is behind schedule, four months after restructuring

Sources