The Tragedy of the Frozen Middle
The fact that only 12% of companies generate real value with AI reveals that the problem isn't the algorithm, but the middle management layer that keeps stalling the transformation it should be leading.
There is a number in Genpact's latest survey of 500 senior executives that should send chills down any board's spine: only 12% of companies can be considered true leaders in artificial intelligence. According to Sanjeev Vohra, the consultancy's Chief Technology and Innovation Officer, the criteria to join this club is ruthless — you need to have AI running in production, generating measurable financial results, and maintaining governance that confirms these numbers. The other 88% are trapped in a purgatory of proof-of-concepts that never die, but also never take off.
The usual explanation for this failure tends to be an easy blame placed on immature technology or a risk-averse board. Genpact's research, however, points the finger at a less obvious and more embarrassing suspect: the so-called "frozen middle." Vohra identifies that the major bottleneck to corporate innovation is the layer of middle managers. They are too operationally exhausted to lead a transformation, yet too structurally central to the org chart to be ignored. The machine doesn't stall at the top, nor at the base, but in the middle.
This explains why so many companies confuse the starting point with the destination. According to Vohra, the implementation of AI copilots has been treated as the ultimate goal of tech strategy, when they are merely tactical assistance tools. It is the difference between giving an employee a keyboard shortcut and redesigning the production process they operate. Most companies bought the software license but didn't restructure the work. The result is marginal productivity that barely justifies the investment.
The irony of this scenario is that governance — frequently used as an excuse for paralysis — is virtually nonexistent. According to the data discussed in the survey, 99% of large companies lack a real AI governance program. Instead of establishing accountability tracks to accelerate safe adoption, corporate politics uses the shadow of risk to justify inertia. Companies that keep waiting for a flawless roadmap before acting are already irreversibly behind.
The way out of this mediocrity is not more technology, but symbolic leadership. Vohra notes that the strongest signal of AI adoption within Genpact itself came when the company's CEO started writing code on a Friday afternoon using the new tools. Transformation requires this kind of hierarchy breakdown. Vohra's goal is surgical: engineers ten times more productive and business professionals three times more capable. The catch is that this must be treated as a minimum expectation for survival, not an audacious long-term goal.
Ultimately, artificial intelligence is working exactly as it should: as a revealer of organizational incompetence. For the 12% who managed to unfreeze their own management, the technology is a margin lever. For the remaining 88%, AI is just an expensive mirror that insists on showing the bureaucratic face of those who cannot bring themselves to move.
What is the 'frozen middle' in AI adoption?
The 'frozen middle' refers to the layer of middle managers who are too operationally exhausted to lead AI transformation, yet too structurally central to be ignored. It is identified as the primary bottleneck preventing companies from scaling artificial intelligence beyond proof-of-concept stages.
Why do only 12% of companies generate real value with AI?
Most companies fail because they treat AI copilots as the ultimate goal rather than redesigning actual work processes. Additionally, 99% lack a real AI governance program, using risk as an excuse for inertia, which traps them in endless proof-of-concepts that never scale to production.
How can companies overcome the 'frozen middle' and succeed in AI transformation?
Overcoming this bottleneck requires symbolic leadership and a breakdown of traditional hierarchy. Leaders must actively use the new tools to set expectations, focusing on restructuring work processes to achieve significant productivity gains, rather than just purchasing software licenses.