Meta Chief Executive Mark Zuckerberg acknowledged significant shortcomings in the company's AI agent development during an internal town hall on Thursday. According to reports from Reuters, Zuckerberg told employees that the trajectory of agentic development over at least the last four months hasn't really accelerated in the way that we expected, and that the company's bets on the new structure haven't come to fruition yet. The CEO was referring to AI agents, automated systems that can execute tasks on behalf of a user, which have not progressed as quickly as anticipated despite earlier optimism. Zuckerberg noted that conversations with our top people when they started planning the restructuring in January and February were that they were worried that we weren't going to move fast enough to adapt, and that executives were super optimistic about tools like Claude Code from AI startup Anthropic at the time. In retrospect, Zuckerberg said the kind of trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected. The comments highlight the central tension in Meta's evolving AI strategy, underscoring the company's massive financial and talent investments against the practical realities of accelerating development.
The acknowledgment comes as Meta continues to grapple with the fallout from its sweeping restructuring efforts. As reported by Reuters, the company laid off about 10% of its global workforce and reassigned roughly 7,000 employees to AI-focused teams in May, moves that prompted significant employee pushback and raised concerns about morale. Zuckerberg added that the company reorganization was not as clean as it could have been and that executives had miscalculated on the timing of the changes. The changes were part of a broader restructuring aimed at funding costly AI infrastructure investments and positioning Meta to capitalize on efficiency gains from AI-assisted work. Zuckerberg told employees in May that he did not expect further companywide layoffs this year, though some workers were skeptical of these assurances. The leadership team continues to balance aggressive development targets with the necessity of maintaining workforce buy-in. The company's recalibration reflects a growing industry realization that rapid AI deployment requires robust internal safeguards and transparent communication.
Despite current setbacks, Zuckerberg expressed optimism about future AI benefits, telling employees he expects the social media giant will begin to experience more significant benefits from its AI investments within the next three to six months. According to Reuters, Meta is projected to spend as much as ₹14,500 crore on AI infrastructure this year, representing a significant portion of Big Tech's more than ₹70,000 crore outlay on AI technology. The company has now revised its capital expenditure guidance upward, with Meta expecting capital expenditures of between $125 billion and $145 billion in 2026, up from its previous range of $115 billion to $135 billion. The company attributed the increase to higher component prices and additional data center costs needed to support future capacity. Meta's spending makes it one of the largest contributors to the more than $700 billion that major technology companies are projected to invest in AI infrastructure this year.
In the same town hall, Meta's Chief Technology Officer Andrew Bosworth provided updates on the company's controversial mouse-tracking software incident. As reported by Reuters, a review of the data security incident indicated that no employee data was included in AI training, and if the program is turned back on once the review is completed, it will be on an opt-in basis. Bosworth told employees that for people who are comfortable, that's great, they can contribute to this kind of great human survey, while those uncomfortable will not be affected. The program, which tracks employee mouse movements and digital activity for AI training, was paused in June after sensitive employee information was found to be accessible across the company. The exposed material included internal prompts, private conversations, performance information and other sensitive data, according to documents reviewed by Reuters. When Meta first installed the program on U.S. employees' computers in April, Bosworth told them there was no way to opt out, but the company has now committed to providing an opt-in option going forward. Bosworth acknowledged the initiative had damaged internal trust, though he noted it yielded unexpectedly valuable data. The policy reversal follows a broader pattern of course corrections within Meta's AI division.
