Meta’s new AI team has 50 engineers per boss. What could go wrong?

Meta is implementing a radical organizational experiment within its new applied AI engineering division. According to a Wall Street Journal report, the team tasked with advancing the company's 'superintelligence' efforts will operate with a staggering 50-to-1 employee-to-manager ratio. This structure doubles the 25-to-1 ratio typically considered the outer limit of managerial 'span-of-control,' startling even experts familiar with flat organizations. The move is a bold bet that extreme flattening will streamline decision-making, position management closer to frontline engineers, and accelerate innovation by removing hierarchical bottlenecks.

However, prominent organizational behavior experts are sounding alarms. André Spicer, Executive Dean of Bayes Business School in London, bluntly stated, 'It’s going to end in tragedy is the bottom line.' While the theory suggests such structures boost agility and employee engagement by fostering direct access to authority and cross-functional collaboration, the practical reality of one manager overseeing fifty direct reports is fraught with risk. It challenges the manager's ability to provide adequate support, mentorship, and oversight, potentially leading to burnout, misalignment, and project failures. This high-stakes gamble underscores the intense pressure Meta faces to rapidly advance its AI capabilities in the competitive race against rivals like OpenAI and Google DeepMind.