McKinsey is cutting its own back office to AI, and the message travels
The firm that sells AI transformation is now running one on itself. The numbers are modest so far. The direction is not.
- McKinsey cut roughly 200 technology and support staff in late 2025, part of a broader reduction reported at around 3,000 to 4,000 roles, close to 10% of its workforce.
- The cuts hit back-office and junior research work, where generative AI now does in minutes what analyst hours used to cover.
- It is the firm's largest workforce reduction since the 2008 financial crisis, and leadership has signaled more non-client roles will go over the next two years.
- Forecasters still expect AI to create more jobs than it removes across the decade, even as specific roles disappear.
The firm that has spent two years telling clients to reorganize around AI is now doing it to itself. McKinsey cut about 200 technology and support staff in late 2025, a move Fast Company read as a warning sign for consulting. Reporting since then has put the broader reduction at roughly 3,000 to 4,000 roles, close to a tenth of the workforce, the firm's largest cut since 2008.
What McKinsey did
The roles going are not the ones in front of clients. They are in the back office, research scheduling, compliance, reporting, and the junior analyst work that generative AI has compressed hardest. A spokesperson framed it as routine efficiency, and the global managing partner has said the non-client group will keep shrinking as the tools take on more. The plan is to hold or grow client-facing headcount while thinning everything behind it.
Why a small number carries weight
Two hundred jobs at a firm this size is not, on its own, a crisis. What makes it notable is who is cutting and why. McKinsey sells the AI transformation story to the largest companies in the world. When it applies that story to its own staff, and names AI as the reason, it hands every client permission to do the same. The message travels farther than the headcount. This is the human edge of the shift from software you operate to agents that do the work, the theme running under our primer on what GaaS is and our look at why the seat is losing to the outcome.
The wider ledger
The full picture is not only subtraction. Forecasters, including the World Economic Forum, still project that AI and automation will create more roles than they eliminate over the rest of the decade, even as specific jobs vanish and the mix of skills changes. The tension the McKinsey cuts expose is one of timing and distribution. The roles disappearing now are concrete and named. The roles arriving are a forecast. For operators and workers alike, the practical question is which side of that ledger their own work sits on, and how fast.
Sources: Fast Company, Quartz.