Mark Zuckerberg just told his own staff that Meta’s AI agents aren’t moving as fast as he’d hoped. According to TechCrunch AI, which cites Reuters reporting, the CEO made the admission at an internal town hall on Thursday, saying the pace of AI agent development had not “accelerated in the way” executives previously expected.
That’s a notable thing to hear from the person who bet the company on it.
What Zuckerberg actually said
The comments landed at an all-hands meeting where Zuckerberg walked staff through the state of Meta’s AI push. A few things stood out from the TechCrunch AI report:
- The perceived upside of Meta’s new AI-focused company structure hasn’t “come to fruition yet.”
- He expects improvements from the company’s AI investments over the next three to six months.
- He conceded that earlier job cuts weren’t as “clean” as they should have been.
- Those cuts happened because top officials “were worried that we weren’t going to move fast enough to adapt” to a shifting industry.
So the message is mixed. Progress is behind schedule, but the boss is still forecasting a turnaround by year-end.
The reorganization behind the comments
To understand why this matters, you need the backdrop. Earlier this year Meta laid off roughly 8,000 employees, about 10% of its corporate workforce, and reassigned another 7,000 people into various AI groups, including one named Agent Transformation, according to Bloomberg reporting cited by TechCrunch AI.
That’s not a side project. That’s a company restructuring itself around a bet that AI agents would carry more of the load. When the CEO now says the upside hasn’t arrived, he’s grading his own reorg.
And the internal picture doesn’t look great. TechCrunch AI notes that several investigative reports have described Meta’s months-old AI unit as a “soul-crushing gulag,” in the words of some engineers assigned to it. Morale problems in the exact group meant to deliver the future are worth paying attention to.
Why this matters for the industry
Here’s what stands out to me. Meta is expected to spend as much as $145 billion on AI infrastructure this year, per Reuters via TechCrunch AI. That’s one of the largest capital bets in tech history, and the person making it is telling employees the results are running late.
The broader story the industry has been telling for two years is that AI agents would soon handle real work, replace headcount, and reshape how companies operate. Meta ran that experiment at massive scale. The early verdict, in Zuckerberg’s own words, is that replacing people with agents is harder than the hype suggested.
That’s a useful reality check for everyone downstream:
- For enterprises: If Meta’s own teams can’t get agents to perform on schedule with near-unlimited resources, temper your timelines. Agents are still an emerging capability, not a plug-in replacement for staff.
- For engineers: The “AI will take your job next quarter” narrative just got a public dent from one of the biggest spenders in the field.
- For the market: Investors have been pricing in near-term returns on enormous AI capex. A CEO saying the payoff is three to six months out, again, is the kind of line analysts will remember.
What to watch next
Zuckerberg gave himself a deadline. He expects to see improvements from Meta’s AI investments in the next three to six months, which puts the check-in point somewhere around late 2026 to early 2027. That’s the window to judge whether this was a temporary lag or a deeper miscalibration.
A few signals worth tracking:
- Whether Meta ships agent products that show up in real usage, not just demos.
- Whether the reassigned 7,000 employees start producing visible results.
- Whether that reported morale problem in the AI unit gets addressed or gets worse.
- Whether Meta’s capex guidance holds or shifts as results come in.
TechCrunch AI reports that it reached out to Meta for comment. For now, the takeaway is simple: the company that spent the most and cut the deepest is saying the agents aren’t there yet. When the biggest believer starts managing expectations, the rest of the industry should listen.
More details are available in the original TechCrunch AI report.