Zuckerberg concedes AI agents lag hopes at Meta town hall

The CEO of a company on track to spend as much as $145 billion on AI infrastructure this year told his own staff, at an internal town hall, that the thing that spending is supposed to build is not coming together on schedule. According to TechCrunch, Mark Zuckerberg told Meta employees Thursday that the pace of AI agent development had not 'accelerated in the way' executives had previously expected. He also said the earlier reorganization, which included roughly 8,000 layoffs and 7,000 employees reassigned into AI groups such as one called Agent Transformation, was not as 'clean' as it could have been, and that the upside of the new structure had not 'come to fruition yet.'

The reason this matters beyond Meta is that Zuckerberg has been one of the loudest voices arguing that autonomous agents are the next platform, and spending accordingly. When the person authorising that much capital tells his own workforce that the timeline slipped, it becomes a datapoint about the broader agent thesis, not just about Meta's org chart. The layoffs and the transfers were framed as necessary because staff 'weren't going to move fast enough to adapt.' Now the boss is telling employees the acceleration on the other side of that painful cut has not shown up either.

Zuckerberg reportedly said he still expects clearer returns from the AI investments within the next three to six months. That is a specific window, and it is a self-imposed one, which makes it the internal yardstick employees and probably investors will measure against later this year.

The honest caveat is that this comes from an internal meeting, so we are getting a paraphrase and a handful of quoted fragments rather than a full transcript. What the reporting does not spell out is which agent capabilities or product lines are actually lagging, against what benchmarks, or whether the up to $145 billion capex figure changes if the next two quarters underdeliver. The industry backdrop cited alongside is that only around 11 percent of enterprises that have adopted agentic AI are running it in production, per research aggregating Gartner, McKinsey and Digital Applied data, and analysts project more than 40 percent of agentic AI projects will be canceled by the end of 2027.

If that backdrop holds, the interesting move over the next two quarters is who narrows their agent bets and who doubles down. Meta has just told its own staff which side of that it wants to be on. Whether it can actually get there in three to six months is the part worth watching.