Meta’s new AI team has 50 engineers per boss. What could go wrong?
There are flat organizational structures, and then there’s Meta’s new applied AI engineering team. The division, tasked with advancing the tech giant’s superintelligence efforts, will employ a 50-to-1 employee-to-manager ratio, according to the Wall Street Journal, double the 25-to-1 ratio that is usually seen as the outer limit of the so-called span‑of‑control scale.
The Facebook parent’s one-sided management ratio took aback even those well-versed in flat organizations. “It’s going to end in tragedy is the bottom line,” says André Spicer, executive dean of Bayes Business School in London and a professor of organizational behavior.
The idea behind a flat organization, in which managers have a large number of direct reports, is that it makes companies more agile by streamlining decision-making processes and positioning management closer to front-line workers and the customer experience. Cross-functional collaboration that isn’t muddled in hierarchy speeds up innovation. Employees who are closer to people of authority are more engaged, with a deeper sense of ownership. Or so the theory goes.
Meta is not alone in embracing a flat structure. Companies across the U.S. are flattening out, according to a January Gallup report. The average number of people reporting to managers rose from 10.9 in 2024 to 12.1 in 2025. Last year’s figure represents a nearly 50% increase in team size since Gallup first measured, in 2013.
And ultra-flat organizations account for a big part of the uptick. “The increase in average team size across the U.S. working population in the past year is largely influenced by a two-percentage-point increase in teams of 25 or more employees,” the report says.
The business world cycles through periods of tight and “loose” or flat culture, the latter being more en vogue when the economy is good, Spicer says. Delayering “will save costs in the short term,” he says. “You can show some nice quarterly report, quarterly numbers from that.”
“But then it will create medium-term problems,” he says.
Flat structures work best in “expert-oriented organizations,” Spicer says. Software engineering, for example, is ripe for flatter structures because it runs on peer coordination and is governed by professional norms. He puts his own profession of academia in the same category.
Still, things can go awry even in professions well-suited to flatness. First, says Spicer, junior or less experienced-employees will get overlooked. Second, line managers can become completely swamped and burn out. And third, a lot of the people in between will feel a lack of direction. That will result in “the loudest people or the problem cases” monopolizing managers’ limited attention.
In most cases, the natural tendency to organize large teams into smaller groups wins out, and flat teams end up establishing makeshift hierarchies in the absence of formal ones. (Spicer says research has determined that the right-sized team is seven people per manager, give or take a few.)
Zappos was once the most famous example of a radically-flat organization. In fact, the shoe retailer, now owned by Amazon, took things a step further, abiding by a decentralized “holacracy” management structure that eliminated all job titles, managers, and hierarchy. After a zealous rollout in 2015—CEO Tony Hsieh offered buyouts to anyone who wasn’t 100% committed—Zappos eventually retreated from the system and reintroduced managers in an effort to refocus workers on customers.
It’s possible that AI can ease some of the pain points that emerge in flat structures by automating the task allocation and employee counseling that typically fall to middle managers, Spicer says. (Meta did not respond to a request for comment on how its applied AI engineering team will function.)
Technology has flattened organizations before, but with only temporary effect. The computerization of the office in the 1980s and ‘90s ushered in “a huge wave of delayering of middle managers,” Spicer says. But that winnowing reversed itself as companies grew more complex and sought to serve more stakeholders. That trend lasted, he says: “If you look at where we are now, from the 1980s to today, there’s actually been an explosion of middle management.”
This story was originally featured on Fortune.com