McKinsey Sees AI Reshaping Half of U.S. Work Hours While Skills Become the Main Bottleneck

Automation could absorb more than half of U.S. work hours by 2035, but McKinsey’s forecast describes a major reallocation of work rather than the disappearance of half the workforce. Technology and economic changes could reduce labor demand equivalent to 36 million jobs over the next decade, while growth in AI-related industries and other services could create demand for about 41 million jobs.

That balance would leave the U.S. with more jobs overall, but in a different occupational mix. The central pressure falls on workers whose current duties contain large volumes of repeatable administrative or routine tasks. McKinsey’s research frames the transition as a labor-market redesign involving both automation and expansion.

The hardest problem is matching workers to the jobs that grow, not simply generating enough jobs. About 11 million Americans may need to leave shrinking occupations for entirely different fields, according to McKinsey’s findings. An average of 770,000 workers a year would need to cross occupational lines during the next decade, 3.6 times the historical rate.

That pace would last far longer than the labor-market disruption caused by COVID-19, which produced a similar surge for only 1 or 2 years. Many workers can move within occupational groups they already understand, but others will need new credentials, technical abilities, or experience in unfamiliar workplace settings. The forecast places schools, employers, and public workforce systems at the center of the response.

Office and administrative support is the occupational group facing the largest job losses in McKinsey’s analysis. Automation could take on 80% of the hours worked in that group, putting routine scheduling, documentation, data handling, and coordination tasks under sustained pressure.

An administrative assistant illustrates the transition. A worker with years of experience keeping a department organized may possess valuable judgment and institutional knowledge, yet still need a structured route into a growing occupation. The relevant pathway might involve customer operations, project coordination, health services administration, or another role that rewards communication and problem-solving alongside digital fluency. The challenge is turning existing experience into recognized qualifications before the old role contracts.

AI’s influence will differ sharply by occupation. In management and law, McKinsey expects technology to support workers more often than replace them outright because judgment, coordination, accountability, and responsibility remain central to the work.

Those occupations will still change. Workers will need training to review AI-generated output, verify accuracy, manage risk, and redesign workflows around automated systems. That creates new responsibilities even where job titles survive. The same pattern extends across professional services: AI handles more production, while people take on supervision, interpretation, client communication, and decisions that carry legal or organizational consequences.

The result is a more demanding definition of workplace readiness. Technical familiarity matters, but so do communication, domain knowledge, and the ability to assess machine-produced work.

Employers will determine whether the forecast produces broad mobility or concentrated displacement. Hiring for new roles without building pathways from shrinking ones leaves experienced workers stranded, while training disconnected from actual vacancies wastes time and money.

The scale of the projected movement favors targeted programs tied to specific occupations. Companies will need to identify which tasks automation absorbs, which responsibilities expand, and which adjacent roles workers can reach with focused training. Credentials also matter because many workers cannot signal readiness through prior job titles alone.

The McKinsey analysis gives employers a practical benchmark: an economy can create more jobs while individual workers still face serious losses. Aggregate employment figures therefore won't capture the disruption. The decisive question is whether workers can cross occupational boundaries quickly enough to meet demand.

The next decade will bring a prolonged skills transition rather than a single shock. McKinsey’s estimate of 770,000 annual cross-occupation moves indicates that worker mobility must become a standing feature of the labor market, with training and credentialing available throughout working life.

The near-term signals will emerge in administrative support, professional services, and AI-enabled industries. Companies adopting automation will reveal which tasks disappear and which new oversight duties appear. Policymakers and educators will need to track those changes closely, while workers will face pressure to build portable skills before their current roles shrink.

The headline number is 36 million affected jobs, but the sharper warning concerns access. Economic growth can create 41 million jobs and still leave millions behind if training, credentials, and hiring practices fail to connect people with the openings.