In 2013, a founder turned down an investment offer from Mark Cuban on Shark Tank because Cuban insisted AI would soon make much of the founder's specialized expertise irrelevant. The founder pushed back, believing human judgment could not be automated. Three years later, the market shifted and the founder realized Cuban had been right, though not for the reason either of them thought at the time.
The lesson from that story, recently shared as a personal essay, matches a much bigger shift now playing out at large companies worldwide. Businesses are removing entire layers of middle management, and they are doing it because AI has taken away the reason those layers existed.
Target cut 1,800 corporate jobs this year. Its CEO told staff directly that too many layers and overlapping work had slowed decisions and made it harder to get anything done. GitLab restructured its company for what it calls the AI era, removing up to three layers of management in some departments while cutting its overall staff by 14 percent. Block, run by Jack Dorsey, cut close to 4,000 jobs and described it as a permanent restructuring, not a cost cut, with a goal of having most employees report directly and letting AI absorb the work that middle managers used to do. Oracle has cut between 20,000 and 30,000 roles, with middle management the main target.
The numbers behind this trend are striking. Research firm Gartner predicts that by the end of 2026, one in five companies will use AI to flatten their organization, removing more than half of their current middle management jobs. Separately, workforce data firm Revelio Labs, which tracks over 100 million employment profiles, found that middle management job postings were still down 42 percent from their 2022 peak as of last October.
Why middle management specifically? Many of these roles exist to hold and pass along information that no single person used to be able to track alone: a coordinator catching a request and routing it, a layer of managers relaying updates between departments that cannot see each other directly. AI can now hold that same information and context without a person carrying it by hand, so the job of relaying disappears even though the job of deciding does not.
That distinction matters for anyone running a team. A role where someone receives new information and weighs it against years of experience before making a call is not going away, it just gets faster with AI. A role where the honest answer is that someone mostly notices something and passes it along is the one that is disappearing right now, at Target, at GitLab, at Block, and at hundreds of other companies quietly doing the same thing without a press release.
The businesses getting ahead of this are not just buying AI tools for existing staff. They are redrawing who reports to whom and cutting out steps that no longer need a person, then moving that freed up time toward the parts of the business that still need real judgment.