Meta has been betting heavily on artificial intelligence, ramping up investments in AI infrastructure while cutting costs elsewhere. The company laid off around 8,000 employees in May, reassigned nearly 7,000 workers to AI-focused teams, and reorganised large parts of its workforce. However, CEO Mark Zuckerberg has now admitted that one of Meta's biggest AI ambitions—developing AI agents—is progressing more slowly than expected, and acknowledged that the company's restructuring has not gone entirely to plan.
According to a Reuters report, Zuckerberg told employees, "The trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected." AI agents are software powered by large language models that can perform tasks autonomously on behalf of users.
Despite the slower progress, Zuckerberg said Meta remains committed to its long-term goal of achieving "superintelligence" and expects its AI investments to begin delivering more meaningful results over the next three to six months. The company is projected to spend as much as $145 billion on AI infrastructure this year.
Meta has also strengthened its AI efforts through acquisitions, including Moltbook, the viral social platform for AI agents, whose team has since joined the company's AI division.
Restructuring hasn't gone as planned
Speaking about Meta's internal overhaul, Zuckerberg admitted that the changes "haven't come to fruition yet." He revealed that early this year, senior executives were concerned the company was moving too slowly in adapting to the AI race, particularly as tools such as Anthropic's Claude Code gained momentum.
In March, Meta created a dedicated Applied AI division and moved roughly 6,500 engineers and product managers into AI-related roles. Alongside that, the company cut about 10 per cent of its workforce, reassigned another 7,000 employees to AI-native teams and cancelled around 6,000 planned hires.
The restructuring, however, drew criticism from employees. Meta Chief Technology Officer Andrew Bosworth later admitted in an internal memo that the company had done an "atrocious" job explaining the purpose and vision behind the organisational changes.
No mandatory mouse tracking
Meta also addressed concerns over its controversial employee monitoring programme. In April, reports that the company was using software to track mouse movements and keystrokes to train AI agents sparked backlash over privacy, prompting Meta to pause the initiative.
Bosworth told employees that an internal review found no employee data had been used to train AI models. He also clarified that if the programme returns, participation will remain voluntary.
"For people who are comfortable, that's great, they can contribute to this kind of great human survey. To people who are not, it is not an issue," he said.
