AI Could Create More US Jobs Than It Displaces, but 11 Million Workers May Need to Switch Careers

Artificial intelligence may not shrink the US job market. It could, however, force millions of workers to change what they do for a living.

A new McKinsey Global Institute report estimates that automation could cut US labour demand by the equivalent of 36 million jobs over the next decade. Growth in the AI value chain and the wider economy could create demand for more than 40 million jobs. That would leave the country with a larger job pool by 2035. But a major workforce shift sits beneath those headline numbers.

McKinsey estimates that about 11 million workers, or roughly 7% of current employees, may need to move between occupations. The report stresses that this figure comes from a model. It does not measure job losses or transitions that have already happened.

Jobs will change more than they vanish.

Most debate about AI focuses on how many jobs machines will replace. McKinsey paints a more complicated picture.

Automation can cut the labour needed for certain tasks without erasing an entire occupation. Meanwhile, AI can raise productivity, open new markets and create jobs around technology infrastructure. The report’s base estimate says automation could reduce labour demand by about 21% of current US work hours. That equals roughly 36 million jobs, but growth elsewhere in the economy offsets much of the drop.

About 25 million affected workers could stay in their current occupations, because growth in those fields balances the decline. The other 11 million could need to move into different roles.

That distinction matters. A job can change a great deal without disappearing. Technology may automate calculations, data processing or routine administrative work. At the same time, it can raise the value of judgement, relationship management and oversight.

McKinsey points to the spreadsheet as proof. Spreadsheets took over calculations that once ate up accounting clerks’ time. Accounting did not vanish. Some tasks got cheaper, and other forms of expertise became more valuable.

Mobility is the real challenge.

McKinsey describes the coming decade as a problem of mobility rather than scarcity. The economy may create enough jobs in total. Still, those jobs may not match the declining ones by occupation, industry, region or pay.

An administrative worker, for example, cannot always step straight into healthcare, construction, management or tech. The move depends on existing skills, training, wages and credentials.

McKinsey calls these routes pathways, and it sorts them into three types:

• Direct pathway: strong skill overlap, little retraining and no wage loss.
• Winding pathway: more reskilling, some wage loss or extra time to qualify.
• Unpaved pathway: large skill gaps, heavy credential requirements or other barriers.

Only about one in seven workers has a direct pathway into growing work. Almost half face an unpaved one. Creating jobs is therefore only half the task. Workers also need realistic ways to reach them.

The wage picture adds pressure. McKinsey estimates that roughly 60% of growing employment could sit in the top two wage quintiles, mostly in healthcare, construction and management. By contrast, more than 70% of declining employment could fall in the bottom two quintiles. Office and administrative support, retail and sales, and transportation and logistics lead that group.

Government data points the same way. The Bureau of Labour Statistics (BLS) projects US employment to grow by about 5.9 million jobs, or 3.5%, between 2025 and 2035. Healthcare support leads at 13.3% growth, followed by healthcare practitioners at 8.0%, computer and mathematical roles at 7.3%, management at 6.2% and construction at 5.1%. Nurse practitioners, solar installers, data scientists and health services managers show especially strong growth.

These projections do not mean a displaced worker can slide into any of those roles. Education, licensing and experience requirements still create real gaps.

Credentials and skills raise the bar

Skills are only part of the equation. McKinsey estimates that about 85% of growing jobs require a credential or certification. These include degrees, professional certificates and licences.

Some requirements are legally mandatory. Others reflect employer preference, and preferences can change much faster than licensing law. A worker may already have most of the skills a growing job demands. Without the right paperwork, though, a simple move can become a multiyear effort. Employers should ask whether each credential requirement serves a genuine need or simply continues an old screening habit.

Change will also reach people who keep their jobs. McKinsey estimates that more than 70% of workers could need some reinvention, because new tasks may fill over 15% of their working time. Many will work alongside AI systems, supervise automated workflows or interpret AI-generated information. The World Economic Forum’s Future of Jobs Report 2025 backs this up. Employers there expect AI, big data and cybersecurity skills to rise fast, alongside creative thinking, resilience and lifelong learning.

The demand for AI fluency shows the trend clearly. McKinsey defines it as the ability to work effectively with AI tools. Its analysis of job postings found demand for AI fluency grew about 11-fold between 2022 and 2026. Demand for adaptability rose roughly fivefold. Demand for resilience, curiosity and willingness to learn about tripled.

McKinsey groups these traits under “empowering skills”. They help workers adapt when an occupation changes. AI fluency, then, is not a final destination. It is part of an ongoing ability to keep up with new technology.

What employers and workers should do

Exposure will vary by field. McKinsey expects faster automation in office and administrative support, computer and mathematical work, and sales. Healthcare should see slower adoption, since it relies on physical presence, human interaction and accountability.

Even so, few of these jobs will vanish outright. Tasks change at different speeds. Workers may spend less time on routine activity and more on exceptions, customer relationships and analysis. The biggest shift could therefore happen without millions of termination notices.

For employers, the lesson is to build strategy around skills rather than job titles. Companies need to identify which tasks AI can automate, which still need human judgement and which new activities emerge after rollout. Hiring for transferable and adjacent skills can shorten the distance between declining and growing roles. It can also make internal moves far more practical. Firms that deploy AI without building employee capability may capture only part of its value.

For workers, the advice goes beyond “learn AI”. Start by mapping where your current skills transfer. Industry knowledge, customer service, project management and analytical ability can each open more than one route. The goal is to find the shortest realistic pathway and spot which skills or credentials are missing.

McKinsey’s research suggests the US could have more jobs in 2035 than it does today. Yet counting jobs created or lost misses the point. A labour market can add millions of positions and still leave workers behind if those roles demand different skills, credentials or locations.

The better question is whether workers can move from the jobs being reshaped to the opportunities being created. Technology may generate the demand. Building the pathways to reach it will be the harder job.

AIAI and EmploymentAI automationAI FluencyAI jobsAI SkillsAI workforceArtificial IntelligenceautomationBusiness TechnologyDigital Transformationfinancial technologyfintechFuture Jobsfuture of workJobs 2035Labor MarketMcKinseyMcKinsey Global InstituteReskillingTechnologyThe Future of WorkUpskillingUS JobsUS WorkforceWorkforce TransformationWorkforce Transition