Meta CEO Mark Zuckerberg says AI agent development is progressing slower than expected, raising fresh questions about the payoff from the company's massive AI investments.
 |
| Meta CEO Mark Zuckerberg says AI agents have not advanced as quickly as expected, while the company also moves to make its employee tracking software optional after a security review. Image: CH |
Tech Desk — July 3, 2026:
Meta is discovering that building the next generation of artificial intelligence is taking longer than expected.
Chief Executive Mark Zuckerberg told employees during an internal town hall that the company's AI agent strategy has not progressed at the pace executives anticipated. The comments offer a rare glimpse into the challenges facing one of the world's biggest AI investors as the race to develop autonomous software intensifies.
The admission comes just months after Meta reorganized thousands of employees and cut around 10% of its global workforce to focus more aggressively on artificial intelligence.
Zuckerberg acknowledged the restructuring itself was far from perfect.
He said the changes were not as "clean" as they could have been and admitted company leaders had misjudged the timing. While the strategy remains in place, he suggested the expected gains have yet to materialize.
"The trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected," Zuckerberg told employees, according to a recording reviewed by Reuters.
AI agents are widely viewed as the next major step in artificial intelligence. Unlike chatbots that simply answer questions, these systems are designed to perform tasks, complete workflows and make decisions with minimal human involvement.
Many technology companies believe AI agents could transform software, customer service and workplace productivity. But Meta's latest assessment shows that turning that vision into reliable products is proving more difficult than early expectations suggested.
Zuckerberg said executives entered the year convinced Meta needed to move faster to keep pace with rapid advances in AI.
He recalled that senior leaders were especially optimistic about new developer tools, including Claude Code from AI startup Anthropic, when they planned the company's restructuring earlier this year.
That optimism, however, has yet to translate into the acceleration executives hoped for.
Even so, Zuckerberg is not backing away from Meta's AI ambitions.
The company is expected to spend as much as $145 billion on AI infrastructure this year, making it one of the largest investors in artificial intelligence globally. The spending is part of a broader wave of investment across major technology companies, which together are expected to commit more than $700 billion to AI.
Zuckerberg said he still expects Meta to begin seeing more meaningful benefits from those investments within the next three to six months.
The town hall also addressed another sensitive issue inside the company.
Meta's Chief Technology Officer, Andrew Bosworth, said a review of the company's controversial mouse-tracking software found that no employee data had been used to train AI systems.
The software, introduced earlier this year, monitored employee mouse movements and digital activity before Meta paused the program after a data security incident exposed sensitive information.
If the software returns, Bosworth said it will operate on an opt-in basis, allowing employees to choose whether they want to participate.
That marks a significant change from the original rollout in April, when U.S. employees were told there was no option to opt out.
For Meta, the latest comments reflect a broader reality facing the technology industry.
AI models have improved at remarkable speed over the past two years. But building dependable AI agents that can consistently complete real-world tasks remains one of the industry's biggest technical hurdles.
Meta is betting that challenge is temporary.
Whether the company's massive investments begin paying off over the coming months could determine how quickly it closes the gap between ambitious AI promises and practical results.