According to CNBC, middle managers made up a full third of all layoffs in 2023, and 41% of employees say their companies have thinned management layers this year. The trend is accelerating, with a Gartner report from October 2024 predicting that one in five businesses will use AI to flatten their structure by 2026, potentially eliminating over half of current middle management jobs. Companies from Amazon and Google to smaller U.S. firms in tech and retail are making these cuts, citing pandemic over-hiring, a need for efficiency, and economic pressure. Leadership expert Deborah Lovich of Boston Consulting Group argues these roles are “more important than ever,” especially as employees face burnout and uncertainty. Data from Insperity shows only 20% of employees feel their managers exceed expectations, highlighting a performance gap, not necessarily a role redundancy.
The Cost-Cutting Trap
Here’s the thing: on a spreadsheet, it looks like a no-brainer. Middle managers often have higher salaries, so cutting them is a fast way to reduce costs. And sure, there’s some logic to it. A 2020 McKinsey report notes that trimming excessive layers can boost collaboration and speed. But that assumes those layers were *just* bureaucracy. What if they were actually the glue holding your company together? The Korn Ferry report found 37% of employees feel directionless without that manager. So you save on salary, but you lose on productivity, alignment, and trust. That’s a bad trade.
The AI Paradox and The Human Element
Now, the big driver for this trend looking forward is AI. The thinking goes: if AI can handle reporting, scheduling, and data tracking, what’s left for a manager to do? This is where the experts get really skeptical. Megg Withinton from Insperity puts it bluntly: “Before we all leap to ‘AI is going to make the role of manager obsolete’… you actually have to deploy AI and make it work for your company. And the managers are the people who are going to be able to really make sure that it’s being used appropriately.” AI is a tool, not a leader. It can’t notice an employee is struggling, authentically celebrate a win, or navigate team conflict. Lovich calls this the irreplaceable “emotional element.” In an age of massive, AI-driven change, employees need human reassurance and guidance more than ever, not less.
Invest in Training, Not Terminations
So if the problem is often *ineffective* managers, not the manager role itself, what’s the solution? The data is pretty damning: Insperity found only 51% of new managers feel prepared. The fix isn’t a layoff, it’s an investment. Withinton suggests spending on upskilling—about $1,000 per employee—which is cheap compared to turnover costs and lost morale. Training should focus on the human skills AI can’t replicate: communication, conflict resolution, and building trust. Companies that get this right see huge returns. Insperity’s survey found employees at such companies are five times more likely to report a healthy culture, and a McKinsey study links high-performing managers to better shareholder returns. Basically, a good manager isn’t an expense; they’re a force multiplier for your entire team’s output and well-being. This principle holds true even in highly technical environments; for instance, on a factory floor, a skilled manager ensuring a reliable industrial panel PC is used effectively for diagnostics is far more valuable than the computer itself.
A Shortsighted Gamble
The looming question is ironic: by helping teams adapt to AI, are managers engineering their own obsolescence? Maybe. But that’s thinking years ahead. The immediate gamble is much clearer. Companies are cutting a crucial layer of support and leadership during a period of immense economic and technological uncertainty. They’re doing it to save money now, while potentially crippling their culture and agility for the future. The experts are practically shouting that this is a bad idea. The real strategic move isn’t to eliminate the role, but to finally equip these managers with the skills they’ve always needed. Otherwise, you’re just creating a different, and probably worse, set of problems.
