Meta's leadership has acknowledged a significant misstep in how it launched its Applied AI division, prompting sweeping organizational reforms aimed at restoring employee morale and trust. The company's Chief Technology Officer Andrew "Boz" Bosworth characterized the rollout of the roughly 6,500-person Applied AI engineering unit, established in March 2026, as "atrocious" in an internal memo obtained by Wired. The candid admission marks a rare moment of executive accountability at a company that has undergone dramatic shifts in strategy and culture over the past three years.

What Went Wrong Inside Meta's Applied AI Division?

The Applied AI unit was created to accelerate Meta's generative AI (generative artificial intelligence) capabilities and compete in the rapidly evolving AI market. However, employees reported feeling alienated and undervalued almost immediately. Staff members described their work as menial and disconnected from their expertise, with one employee going so far as to characterize the division to Wired as "a gulag". This disconnect between the high-stakes mission and the day-to-day experience created a morale crisis that threatened to undermine the division's effectiveness.

Bosworth directly blamed himself and fellow executives for losing perspective during what he described as a "frantic corporate race" to compete in the generative AI market. In his memo, he acknowledged: "We've undermined the trust you have that your specific expertise and contribution will be valued, that you will grow and advance your career". The admission reflects a broader tension at Meta between the urgency of AI development and the need to maintain a healthy, engaged workforce.

How Is Meta Planning to Rebuild Trust and Stabilize the Division?

• Span of Control Limits: Managers will oversee approximately 20 direct reports instead of larger teams, ensuring more personalized attention and career development opportunities for individual engineers.
• Reduced Restructuring Frequency: The company will minimize manager shuffles and organizational changes that have historically disrupted team stability and employee confidence in their roles.
• Internal Mobility Options: Employees who feel forced into the Applied AI division can now apply for other internal roles, giving staff agency over their career paths within Meta.
• Workplace Amenities: Meta plans to upgrade office microkitchens, boost travel budgets, and expand funding for social events to foster in-person collaboration and rebuild community.

Maher Saba, vice president of the Applied AI team, announced the internal mobility initiative, signaling that leadership recognizes the need to respect employee preferences rather than mandate assignments. These structural reforms represent a deliberate reversal of Mark Zuckerberg's aggressive "flattening" strategy, which had compressed organizational hierarchies and increased span of control across the company.

What Are the Broader Challenges Meta Faces in AI Development?

Beyond internal organizational issues, Bosworth addressed the industry-wide infrastructure bottleneck that is constraining AI development across the tech sector. He warned staff to expect "tough trade-offs for a while" regarding the allocation of scarce computational resources. This reflects a reality that even well-funded companies like Meta face limits on the computing power available to train and deploy large language models (LLMs), which are AI systems trained on vast amounts of text data.

Bosworth

Bosworth also sought to address employee anxieties about automation and job security. While he clarified that Meta does not view AI as a total replacement for human labor, he cautioned employees with a memorable phrase: "AI won't take your job, but someone who knows AI might". This framing acknowledges that the competitive advantage in the AI era belongs to workers who understand and can work effectively with these technologies.

Despite the internal turmoil, Meta's external AI initiatives have produced significant results. The company's open-source Llama 3 model has become one of the most widely adopted large language models in the industry, demonstrating that Meta's AI research and development capabilities remain formidable. The challenge now is ensuring that the organization can sustain rapid innovation without burning out or alienating the talent required to build these systems.

The situation at Meta underscores a critical lesson for technology companies pursuing aggressive AI strategies: technical innovation cannot substitute for organizational health and employee engagement. As Bosworth's memo makes clear, even companies with enormous resources and talented workforces must balance the relentless speed demanded by the AI market with the stability and respect required to retain top-tier talent. Meta's next challenge will be whether these structural reforms can actually rebuild the trust that was damaged during the chaotic rollout of the Applied AI division.