Meta Platforms is confronting a gap between expectations and development pace for its artificial intelligence agents, CEO Mark Zuckerberg told employees in a company meeting, according to an internal recording. The executive said progress on agent development over the last four months has not "accelerated in the way we expected," leading to a re-evaluation of the firm’s recent restructuring efforts.
Zuckerberg acknowledged earlier concerns among Meta’s leadership that the company might not be moving swiftly enough to match rapid advances in AI. He said the company had been optimistic about tools such as Anthropic’s Claude Code, but that those expectations have so far failed to materialize as anticipated.
Despite the slower development timeline, Meta is continuing to commit substantial resources to AI infrastructure. The company is expected to spend up to $145 billion on AI infrastructure this year as part of a broader push among large technology firms to build capacity for advanced AI workloads. Zuckerberg said he still expects those investments to yield more meaningful returns within the next three to six months.
Separately, Meta’s Chief Technology Officer Andrew Bosworth addressed an internal review of a paused employee activity-tracking program. The review concluded that no employee data from that program had been used to train AI systems. The findings were shared publicly within the internal meeting recording.
The combination of an intensive infrastructure spend and slower-than-expected product development has prompted company leaders to reassess internal plans and restructure assumptions about timing for returns. Executives conveyed both a recognition of the gap between prior optimism and recent development realities, and an expectation that continued investment will produce clearer results in the near term.
While the company moves forward with its capital commitments to AI systems, the internal messaging underscores an active recalibration of strategy driven by execution timelines that have slipped relative to management’s earlier outlook.