Replacing workers with AI is turning out to be harder than Meta expected. At an internal town hall on Thursday, CEO Mark Zuckerberg told staff that AI agent development had not “accelerated in the way” company executives had previously predicted, according to TechCrunch.
The admission is notable given how aggressively Meta has restructured itself around AI this year. Earlier in 2025, the company laid off roughly 8,000 employees, about 10% of its corporate workforce, and moved another 7,000 workers into AI-focused teams. One of those teams is called Agent Transformation, which tells you everything about what Meta was betting on.
During Thursday’s meeting, Zuckerberg also addressed the layoffs directly. He said the cuts were not as “clean” as they should have been, and that leadership made the moves because they were worried the company would not adapt quickly enough to changes in the tech industry. It’s a candid acknowledgment that the execution was messy, even if the intent was clear.
He also said that the expected benefits from the new AI-focused structure had not “come to fruition yet.” That said, he told staff he expects to see real improvements from Meta’s AI investments over the next three to six months. Whether that timeline holds remains to be seen.
The internal picture at Meta’s AI unit does not sound especially encouraging in the meantime. Several reports have described the months-old group as a difficult, demoralizing place to work, with engineers expressing frustration about the experience of being reassigned there.
Meta is still pouring money into AI regardless. Reuters reports the company is expected to spend up to $145 billion on AI infrastructure this year alone. That level of spending puts enormous pressure on the company to show results, and Zuckerberg’s comments suggest the internal timeline is already slipping.
Meta’s situation reflects a broader tension playing out across the tech industry right now. Many large companies have made bold predictions about how quickly AI agents, systems that can take actions on behalf of users rather than just answer questions, would be ready to take on real work. The reality is proving more complicated. Building reliable, useful agents is a different challenge from building a capable chatbot, and the gap between demo and deployment keeps catching companies off guard.
For Meta specifically, the stakes are high. The company has publicly committed to a future where AI agents handle significant portions of internal work, customer interaction, and potentially software development. If that vision keeps slipping, it raises questions not just about strategy but about the $145 billion being spent to get there.
