Why engineers in Meta's AI unit, built for it's key employee, are calling it a 'mess'

Mark Zuckerberg, the CEO of Meta Platforms Inc., recently addressed the company’s internal turmoil following a significant restructuring that included the layoff of 8,000 employees and the reassignment of another 7,000 workers to AI-related projects. In an internal memo shared with Reuters, Zuckerberg acknowledged that the company had made mistakes during its AI transformation and expressed regret over the impact on employees. While he pledged to avoid further mass layoffs in 2026, he emphasized that such assurances come with caveats due to unpredictable external factors. This statement did little to ease concerns among the remaining 70,000 employees, many of whom feel uncertain about their future within the organization.

Zuckerberg’s memo highlighted the challenges faced during the restructuring process, particularly the “manager managing managers” strategy that led to the reduction in engineering management positions. He admitted that the approach had been harsh, with layoffs occurring in waves across multiple regions. Employees who remained were often reassigned to new AI teams, including a unit known internally as “Applied AI and Engineering,” which has become colloquially referred to as “the Draft.” Many of these employees were not given a choice in their new assignments, leading to widespread dissatisfaction.

The discontent among employees is palpable. Reports indicate that during a recent company-wide livestreamed presentation, an employee interrupted the session with a strong verbal outburst, expressing frustration towards a senior executive. The incident underscored the growing tension within the company, especially among those working in the newly formed Applied AI unit. Engineers in this unit describe their tasks as monotonous and unfulfilling, focusing heavily on generating coding puzzles to train AI models. Some have likened the experience to being in a “gulag,” emphasizing the stark contrast between their previous roles, where they developed features for millions of users, and their current responsibilities.

Beneath the surface of these personnel issues lies a substantial financial commitment. Meta has increased its 2026 capital expenditure forecast significantly, aiming for a range between $125 billion and $145 billion. Much of this investment is directed toward data centers, custom chips, and model training for Meta’s Superintelligence Labs. These financial commitments necessitated the layoffs and reassignments, as the company seeks to align its workforce with its ambitious technological goals. However, the human cost of these decisions has been considerable, with employees feeling undervalued and disengaged.

Despite the challenges, Zuckerberg remains optimistic about the future of Meta’s AI initiatives. He emphasized that automation is not the primary objective but rather a means to enhance user experiences on platforms like Instagram and Facebook, develop smarter wearable technology, and create personalized AI assistants. To address employee morale, he proposed several measures, including larger budgets for offsite events, assigning desks to employees by the end of the year, and organizing a large-scale hackathon in July aimed at fostering collaboration across departments. However, the effectiveness of these initiatives remains to be seen, especially given the current climate of uncertainty and unrest.

As Meta navigates these turbulent times, the focus shifts to whether Zuckerberg’s vision for the company’s future can withstand the scrutiny of its own workforce. With stock prices declining and skepticism surrounding the success of its AI projects, the pressure on both leadership and employees continues to mount. The path ahead for Meta will depend largely on how effectively the company can reconcile its strategic ambitions with the well-being and engagement of its employees.