A new McKinsey Global Institute report paints a nuanced picture of the American labor market over the next decade.
Far from a simple story of mass unemployment, it forecasts that AI and other automation technologies could reduce labor demand by the equivalent of about 36 million jobs, that’s roughly 21% of current work hours by 2035.
At the same time, economic growth, the expanding AI value chain, and structural shifts could generate demand for more than 40 million jobs—producing a net surplus of work.
Roughly 25 million workers will be able to stay in their current occupations, although the nature of their jobs may change, while the remaining 11 million will need to pivot into other fields.
“The good news is that the demand is going to be higher than the decline,” says Tanguy Catlin, director of the McKinsey Global Institute.
But for workers whose jobs are affected by automation adoption, which the report defines as “the share of current work hours potentially automated or meaningfully augmented” by technology, the path ahead could be complicated, Catlin says.
The real challenge, the authors emphasize, is mobility rather than scarcity. Roughly 11 million workers about 7% of the current U.S. workforce, with a modeled range of 6–16 million depending on the pace of adoption, may need to switch occupations entirely.
Another 25 million whose roles are surely going to be affected are expected to stay in the same broad occupations, though the day-to-day content of their jobs is likely to change substantially. Achieving that level of occupational movement would require about 770,000 workers to change occupational groups each year—roughly 3.6 times the long-term historical average.
Lower-wage workers are far more exposed and estimated to be 7.6 times as likely as higher-wage workers to need a full occupational switch.
Workers without a college degree also face elevated risk. More than 75% of those who may need to transition are concentrated in just three broad fields—office and administrative support, retail and sales, and transportation and logistics.
10 Occupations Projected to be Most Affected by Artificial Intelligence Automation Adoption
McKinsey ranked the occupations by a combination of the number of full-time workers, the share of current work hours expected to be automated or meaningfully augmented, and the resulting reduction in labor demand.
Here are the top 10, drawn from McKinsey’s September 2026 report:
1. Office and administrative support
18 million full-time employees | 81% of work hours affected | 38% reduction in labor demand
2. Agriculture, fishing, and forestry
400,000 full-time employees | 76% of work hours affected | 43% reduction in labor demand
3. Technology and analytics
5.2 million full-time employees | 70% of work hours affected | 22% reduction in labor demand
4. Manufacturing and production
8.5 million full-time employees | 68% of work hours affected | 25% reduction in labor demand
5. Retail and sales
13.3 million full-time employees | 67% of work hours affected | 27% reduction in labor demand
6. Business and finance
10.5 million full-time employees | 62% of work hours affected | 18% reduction in labor demand
7. Food service
13.7 million full-time employees | 60% of work hours affected | 31% reduction in labor demand
8. Arts, media, and entertainment
1.9 million full-time employees | 58% of work hours affected | 13% reduction in labor demand
9. Transportation and logistics
13.8 million full-time employees | 55% of work hours affected | 30% reduction in labor demand
10. Installation and repair
6.1 million full-time employees | 54% of work hours affected | 15% reduction in labor demand
Some jobs will be ‘reinvented’
According to Catlin, higher levels of automation adoption don’t always translate to less demand for workers. Some jobs are more likely to be “reinvented” or augmented by AI adoption, rather than replaced, he says.
For example, automation could affect 47% of work hours in the legal field, but the projected loss of labor demand is only 18%.
Lawyers “historically used to spend a lot of time doing research,” Catlin says, but today, AI can perform a lot of that research, so lawyers will spend more time on tasks that require human judgment.
“Your job role or title doesn’t change, but how you spend your time is changing massively,” he says.
With this data, Catlin says that it’s important to note that there is a difference between how much work is “theoretically automatable” versus “what is actually going to be automated” within organizations: according to the report, “Technology determines what is possible, but managerial choices determine how quickly those possibilities become reality.”
As such, the number of workers who will need to switch occupations could range between 6 and 16 million, with 11 million as the base estimate.
Navigating the AI Transition
For the 11 million displaced workers, migrating into entirely new jobs “will not necessarily be easy,” Catlin says.
One in seven will have a “direct” pathway into new jobs in expanding fields, meaning that they can find roles that involve similar skills, require easy-to-acquire credentials and offer equal or better pay than their previous job.
But almost 45% could face an “unpaved” pathway, which means that finding a new job could require them to move geographically, significantly reskill or acquire new credentials — often for lower pay than their original roles.
“To get those credentials may require an investment of time and an investment of money, and that will be source of friction” for workers, Catlin says.
High theoretical automation potential does not automatically equal job losses. In some fields for example legal work, productivity gains are more likely to be absorbed through augmentation, new AI-enabled tasks, and greater human oversight rather than pure headcount reduction. Technology sets what is possible; managerial decisions determine how fast (and how) those possibilities become reality.
Where New Demand Is Expected to Grow
Job growth is projected to concentrate in healthcare, construction, professional and technical services, and management. Many of these expanding roles sit in higher wage quintiles and demand more advanced or specialized skills. The mismatch between declining and growing occupations is why the report describes the coming decade as potentially “the largest and most sustained workforce transformation in U.S. history.”
Only about one in seven workers who need to switch is expected to have a relatively direct pathway (similar skills, accessible credentials, and comparable or better pay). Nearly half face “unpaved” paths that may require significant reskilling, new credentials, geographic moves, or acceptance of lower pay.
What This Means for Workers and Policymakers
McKinsey’s findings underscore that the next decade’s labor-market challenge is less about the total number of jobs and more about helping people move into the right ones.
Successful navigation will require coordinated action: employers investing in reskilling, governments supporting accessible education and credential pathways, and individuals cultivating durable skills such as problem-solving, interpersonal communication, collaboration, critical thinking, curiosity, and continuous learning. As one McKinsey director put it, the pace of change will never again be as slow as it is today.
In short, automation and AI are expected to create more jobs than they eliminate in aggregate. The harder and more consequential task is ensuring that the millions of workers whose current roles shrink or change can actually reach the opportunities that emerge.
