On an intelligence watch floor, the person who keeps the place moving is rarely the one with the most rank. It is the First Class Petty Officer who has stood enough watches to know which report matters and which one is noise, and who runs the floor through the night while a junior officer signs for it. Tasking comes down written in the careful, hedged language senior people use when they want room to be right later. It is the PO1 who turns those words into the handful of concrete things the watch will actually do before sunrise. One night, his diligence catches a bad assumption buried in the work. It is the kind of error that would have gone out into the world and sent a lot of people down the wrong road. The Lieutenant Junior Grade in charge was relieved, not annoyed, that it surfaced before it left the building. His reaction was not written in anyone's job description. It came from the culture the crew had built on that floor, where surfacing a problem early was more important than being right. No one had to pretend the officer was the smartest person in the room. You notice what that petty officer does mainly when he goes on leave and the floor slows to a crawl.

American companies are removing the corporate version of that person, and they are doing it on purpose. Gartner projects that by the end of 2026, one in five organizations will eliminate more than half of their middle management roles, a shift reported by Fortune as part of a broader push to flatten structures and fund AI.1 The pitch is easy to say out loud in a board meeting. Fewer layers and faster decisions, with a lighter payroll to show for it. The savings are real and they arrive this quarter. What leaves with those managers is harder to see on a spreadsheet, which is exactly why it gets cut.

The job nobody wrote down

We talk about middle managers as if their work were status meetings and approvals, a tax the organization pays to feel supervised. Some of it was that. Most of it was translation. The middle layer took strategy written for an annual report and turned it into what a team should do on Tuesday. It carried context downward and friction upward, so the executive setting a target heard early that the target collided with a supply problem nobody upstairs had seen. It absorbed the small conflicts that would otherwise reach a vice president. Strip that out and the strategy still gets announced. It just stops arriving.

Who employees actually trust

There is data on where that trust sits, and it is not flattering to the people doing the cutting. A 2025 survey from Firstup, a workforce communication company with an interest in the answer, asked a thousand non-managerial employees at organizations that had been through layoffs who they rely on. Fifty-two percent named their direct manager as their most trusted source for company updates. Ten percent said the same about senior leaders.2 Eighty-six percent said they depend on their manager to understand what a business change actually means for their own role.2 When that manager disappears, the message from the top does not route around the gap and arrive intact. It hits an empty seat.

Treat the middle layer as overhead and you miss what it was actually doing, which was holding the wiring that let leadership reach the work at all.

The work does not vanish, it lands on you

Here is the part the efficiency math misses. Coordination is conserved. Cut the layer that did it and the work does not evaporate, it redistributes. Some of it climbs to senior leaders who were already running at capacity, who now approve decisions in a messaging app between meetings with no time to coach anyone through why a call was wrong. Some of it simply stops, and you find out which parts months later when an initiative that looked fully staffed simply fails to move. The same Firstup survey found more than a third of employees saying their manager already seems stretched thin and less reachable since the cuts.2 That is the wiring browning out before it goes dark.

The reorganization framed the flattening as removing a layer of cost. What it removed was the organization's capacity to convert intent into action without the top doing it by hand. In practice the company cut a nerve and seemed surprised the hand stopped working.

Before you cut the layer

None of this argues for keeping every manager or for the bloated, seven-approvals version of the middle that earned its reputation. It argues for knowing what a given manager was actually doing before you decide it was nothing. The honest question is not how many layers you can remove. It is who was translating strategy into action, and whether the trust that makes a directive land walks out with them. If you cannot answer that, the flattening is a bet that the organization can run without a nervous system, with the bill coming due on someone else's watch.