AI could automate half of U.S. work hours. 11 million face a skills cliff

Forget the robot apocalypse. McKinsey Global Institute's new report on AI and the U.S. labor market delivers a far more specific - and more urgent - warning: roughly 11 million American workers may need to switch occupations entirely by 2035, and many of them lack the skills or credentials to do it. That is the 'skills cliff' hiding inside a headline about automation absorbing more than half of U.S. work hours over the next decade. The report, from the research arm of McKinsey & Company, doesn't predict mass unemployment. It predicts something harder to fix: a massive, accelerated reshuffling of who does what.

The headline number is real. Technological and economic changes could reduce labor demand by the equivalent of 36 million jobs over the next decade, while growth - including in AI-related industries and services - could create demand for about 41 million jobs. Net, the U.S. could have more jobs in 2035 than today. But the churn is brutal. McKinsey senior partners Eric Kutcher, the firm's former CFO, and Tanguy Catlin, director of MGI, estimate that roughly 770,000 workers a year will need to change occupations over the next decade, about 3.6 times the historical rate. That is not a slow drift; it is a forced migration.

The pain will not be evenly distributed. Office and administrative support roles face particular exposure, with roughly 80% of current work hours potentially automated by 2035. That does not mean 80% of those jobs disappear, but it signals how significantly these roles will change. Meanwhile, in fields like management and law - where judgment, coordination, and accountability matter - AI is more likely to help people do their jobs than replace them outright, according to the report. It may also create new responsibilities that require training, from checking AI-generated work to ensuring its responsible use and redesigning workflows.

The bigger bottleneck is credentials. About 85% of growing employment requires a credential or certification. For 38% of those jobs, the credential is legally required; for another 47%, employers prefer it. That is a massive barrier for the 11 million workers trying to move into new roles. McKinsey suggests companies could widen their talent pools by rethinking which qualifications are truly essential and giving greater weight to proven skills and experience. This is not just a hiring issue; it is a talent strategy issue. If employers cling to outdated credential requirements, they will choke off the very supply of workers they need.

For CFOs and finance leaders, the report has a specific ask: do not just track AI cost savings. Track whether employees actually move into new roles, whether you are filling more positions internally, and how workers fare. And examine what employees do with the time AI frees up. Is it going toward better customer service, new ideas, business growth, or more complex work? Or is it simply reducing the time spent on existing tasks? The key question is not how much work AI can take on; it is what the business does with the time it saves. That is a finance question as much as an HR question.

This is a strategic issue, not just an HR issue. The companies that navigate the skills cliff will be the ones that treat workforce transitions as a core capital allocation decision. That means investing in retraining, rethinking credential requirements, and building internal mobility pipelines. The ones that do not will face the cost of churn, hiring scarcity, and the risk of losing institutional knowledge as 11 million workers try to find their next act. For boards, this suggests that workforce planning should be as rigorous as financial planning, with clear metrics for internal mobility and skills development.

For executives in every sector, the takeaway is that AI's impact on labor is not a single event but a decade-long reallocation. The report's numbers suggest a net positive job count, but the distribution of those jobs will look very different. The winners will be organizations that can retrain and redeploy talent faster than automation changes the work. The losers will be those that wait for the market to sort it out. The report's recommendation to track employee transitions and internal fills is a practical starting point. The real test is whether leadership treats the next decade as a workforce transformation project, not a cost-cutting exercise.