What happens to middle management when AI flattens your organization?

It’s being called the Great Flattening: a global wave of layoffs triggered by the adoption of AI that is primarily hitting middle management. Amazon is currently leading this managerial reset, aggressively streamlining its corporate structure to reduce bureaucracy and speed decision-making. And although the tech sector remains the epicenter, projections suggest that by the end of 2026, up to 20% of firms will use AI to significantly reduce middle management ranks.

The catalyst is the rise of agentic AI—autonomous tools capable of executing complex workflows, managing data streams, and generating predictive modeling for decision-making with minimal oversight. All with the promise of reduced labor costs, faster decision cycles, and the removal of human bias.

Yet historical precedents suggest these transformations come with risks. Specifically, six hidden dependencies are likely to be exposed by the sudden elimination of the middle management layer. And while streamlining may well be necessary, these dependencies must be mitigated if these changes are to be sustainably successful.

1. Loss of the human filter on strategic directives

Middle managers serve as a critical interpretive buffer between senior leaders and frontline operations. They do more than execute strategy. They translate, push back, and contextualize executive directions in the face of operational realities. Without this, organizations risk unfiltered top-down mandates that may be operationally impractical.

2. Context loss in data interpretation

Similarly, AI excels at processing data, pattern recognition, and quantitative analysis. But middle management beats it every time in qualitative contextualization—understanding what the data means within frontline insights, historical nuance, and customer subtleties—that is essential to prevent data from misleading decision-making. Research supports this, too. Purely algorithmic interpretations can miss organizational subtleties without human input, particularly in ambiguous or uncertain situations.

3. Breakdown of upward information flow

Middle managers are usually the guardians of organizational trust. The intermediaries who interpret, explain, and defend organizational decisions while advocating team concerns. Removing this layer risks eroding a critical communication conduit and the psychological safety it enables. Without such intermediaries, evidence shows that employees tend to withhold concerns and bad news more, filtering what they say to a greater extent. Upward information flow thus suffers, and as a result, decisions become less informed.

4. The centralization trap

Removing middle management invariably leads to greater centralization, as power moves to the top. In buoyant markets, that’s fine. But research shows that in downturns and more competitive markets, decentralized businesses do better as they are more agile and responsive to local market shifts.

5. Strategic drift

Middle managers can provide strategic grip: they are close enough to the bigger picture to understand it, yet close enough to execution to detect misalignment and correct it. Flattened structures concentrate authority higher up, but usually without compensating mechanisms for real-time operational feedback. And senior leaders are usually too distant from frontline activity to spot early signs of strategic drift. As a result, the gap between the plan and reality can gradually grow.

6. Erosion of the leadership pipeline

Finally, middle management functions as a developmental pipeline, exposing emerging leaders to operational complexity and stakeholder management. Flattening the structure reduces these talent development opportunities, narrows the future talent pool, and potentially leaves the business more reliant on external candidates.

AI has great potential. And the drive for greater productivity and faster and better decision-making is understandable. But we have been here before. Or, at least, somewhere similar.

The 1990s witnessed an outsourcing boom. It was a modern-day productivity gold rush. No firm numbers exist, but it’s likely the majority of firms joined in and outsourced something. And to be clear, it did produce benefits in many cases. Operating and labor costs were reduced, it allowed for a focus on core capabilities, and it provided access to specialized expertise.

But in the decades since, it’s also clear that in the rush for productivity gains, many potential issues were overlooked, and as a result, outsourcing didn’t work for everyone. It eroded internal capability, introduced hidden complexity and costs, and created new risks, including governance issues and supply chain disruptions.

Organizations are complex systems with all sorts of dependencies. And when you remove significant components, there will inevitably be knock-on effects that need to be mitigated for. Importantly, the six risks described above are not insurmountable. They can’t be eliminated, but they can at least be reduced or contained.

The solutions required will obviously vary between firms, but three core foundations stand out. As middle managers were the primary conduits of upward signals, organizations need to create replacement upward channels, institutionalize bad-news mechanisms and track the health of information flow. As flattening often centralizes power, firms need to define clear local decision rights at the frontline and consistently push operational decisions down within this framework. And they need to appoint context stewards—individuals responsible for both interpreting upward data within operational realities and translating downward decisions within frontline realities.

What is concerning at the moment is that, amid all the headlines and hype, there is so little discussion of the significant risks that need to be mitigated as well. The promise of better productivity today is a poor excuse for forgetting yesterday’s mistakes.