McKinsey Says AI Will Create More Jobs Than It Kills, After 11 Million Workers Get Displaced
The report landed on a Tuesday. McKinsey Global Institute, the research arm of the consulting firm that sells workforce transformation advice to the same companies doing the transforming. The headline number was clean, almost reassuring in a grim sort of way: AI and automation will cut demand for about 36 million U.S. jobs by 2035. Growth elsewhere will create about 41 million. Net gain: roughly 5 million jobs.
You could almost hear the sigh of relief from the boardrooms. More jobs, not fewer. The robots didn't eat the paycheck after all.
But here's the part that gets buried under the net-gain arithmetic. The 41 million new jobs will not go to the 36 million people who lost theirs. The skills don't transfer that way. The geography doesn't align. The credentials don't match. And the transition, for the people caught in the middle, roughly 11 million American workers, about 7% of the current labor force, will be brutal for most of them.
"The next decade's challenge is mobility, not scarcity," the McKinsey researchers wrote.
Mobility. As in, moving people from the jobs that are shrinking to the jobs that are growing. As in, the largest and most sustained workforce transformation in U.S. history, according to the report's own language. And as in, the part nobody has a plan for.
What the McKinsey Report Actually Says
The Numbers Behind the Headline
The McKinsey Global Institute released its workforce analysis in late September. The base-case projection runs like this:
- Automation reduces labor demand for roughly 36 million U.S. jobs by 2035, a 21% contraction of the labor market as currently structured.
- Economic growth and AI-related industries generate demand for about 40–41 million new jobs over the same period.
- About 25 million workers affected by automation should be able to stay in their current occupations because growth in their industries offsets the impact.
- The remaining 11 million — roughly 6.5% to 7% of the workforce, depending on which denominator you use, will need to leave their occupations entirely.
That 11 million figure carries a range. Depending on how fast AI adoption accelerates, the number could be as low as 6 million or as high as 16 million. McKinsey's 2023 forecast put the figure at 12 million career switches by 2030. The new estimate is lower in total but stretches the timeline to 2035.
The Bureau of Labor Statistics publishes a number called "occupational separations", people who leave one occupation for another. The long-run annual average sits around 215,000. McKinsey's projection implies that number needs to climb to roughly 770,000 per year. That's 3.6 times the historical norm.
For context, during the pandemic disruption of 2019–2022, about 788,000 workers per year made comparable occupational jumps. The report notes that the economy absorbed those moves "without lasting damage." The difference is that the pandemic surge was a spike. McKinsey is forecasting a permanent rewiring.
Why "Mobility, Not Scarcity" Is the Real Story
The framing matters. McKinsey is not predicting a job apocalypse. The report is explicit: "the United States is likely to have more jobs available in 2035 than today, but with fewer workers because the population is aging."
The bottleneck is not the number of jobs. It's the pathways between them.
McKinsey categorizes the routes displaced workers can take:
- Direct pathways: The worker can move into a growing occupation with minimal retraining and comparable pay. Only one in seven displaced workers has this option.
- Winding pathways: The worker needs significant retraining or must accept a pay cut initially, but the move is possible.
- Unpaved pathways: The worker faces large skill gaps or credential requirements that block the transition entirely. Nearly half of displaced workers fall into this category.
And then there's the credential problem. About 85% of growing jobs require some form of credential, a degree, a certification, a license. The jobs that are shrinking, office support, retail, transportation, administrative work, often don't require them. So the worker who spent fifteen years processing insurance claims cannot simply walk into a healthcare technician role because healthcare is growing. The door has a lock, and the key costs money and time.
Who Gets Displaced, And Who Doesn't
The Low-Wage Cliff
The displacement is not evenly distributed. Not even close.
Lower-wage workers are 7.6 times as likely as higher-wage workers to need a new occupation, according to McKinsey's analysis. The jobs most exposed to automation cluster at the bottom of the pay scale: office and administrative support, retail sales, food preparation, transportation, agriculture.
The jobs that are growing, healthcare, construction, management, and what Richard Florida calls "wellness work", cluster higher up the ladder. Florida, the urbanist and author of The Rise of the Creative Class, told Fortune that displaced service workers will likely land in "a broad bucket" of fitness, dermatology, Pilates studios, wellness coaching. "I think there will be some displacement," he said. "But I think this is also the area we're going to create the most work."
That's the optimistic read. The less optimistic read is that a retail worker making $32,000 a year cannot easily become a certified nursing assistant making $38,000 without taking six months off work for training, paying for the certification, and finding childcare during the hours she's in class.
The Credential Wall
Here's a number that stops the conversation cold: 85% of growing jobs require a credential.
A credential is not the same thing as a skill. A credential is a gate. It's a piece of paper that says you sat through the required hours and passed the required tests. It costs money. It costs time. And for workers in the occupations most exposed to AI, the ones McKinsey says will need to move, the credential is often the single largest barrier.
The report acknowledges this. It classifies nearly half of displaced workers as facing an "unpaved" path, blocked by "large skill gaps or credential requirements."
The system is not designed for speed. Community colleges have enrollment cycles. Certification programs have prerequisites. Employers have hiring timelines. And the workers who need to move fastest are the ones with the least slack in their lives to accommodate the friction.
The Winding and Unpaved Paths
McKinsey's framework is useful because it refuses to pretend the transition is simple. The report identifies three types of occupational pathways:
- Direct — Little retraining, no pay cut. One in seven workers.
- Winding — Significant retraining or initial pay reduction, but achievable.
- Unpaved — Large skill gaps or credential barriers. Nearly half of workers.
"While almost every worker in the US economy has at least one pathway to a new job," McKinsey noted, "only one in seven workers has a direct pathway to new work."
That sentence should be read twice. The pathways exist. Almost everyone has one. But the pathway is not a road. It's a trail through the woods, and most people don't have the boots.
The Transition Nobody Is Ready For
770,000 Career Switches Per Year
The arithmetic of the McKinsey forecast requires a sustained, decade-long acceleration of occupational mobility. The U.S. labor market currently facilitates about 215,000 occupational switches per year. McKinsey's projection requires 770,000 per year, every year, for ten years.
That's not a spike. That's a new normal.
The labor market is not currently in a state that suggests this is possible. For two years running, the job market has been in what economists call a "low-hire, low-fire" state, a lethargic turnover environment where opportunities exist for the few but not the many. Job openings fell to a five-month low at the end of August. Voluntary quits, a gauge of worker confidence, hovered near a six-year low.
"Sluggish hiring means workers feel stuck," Daniel Zhao, chief economist at Glassdoor, told CNN. "And that means frustration and anxiety continue to build without a healthy outlet for it."
Retraining Infrastructure That Doesn't Exist Yet
The United States does not have a national retraining system. It has a patchwork of community colleges, for-profit bootcamps, employer-sponsored programs, and state-level workforce development offices. The patchwork works for some people. It does not work at the scale McKinsey is describing.
The report does not prescribe a solution. It describes the challenge. But the implication is clear: if 770,000 people per year need to switch occupations, and 85% of the destination jobs require credentials, then the credentialing system needs to expand its capacity by an order of magnitude. Community colleges need funding. Employers need to relax degree requirements. States need to coordinate.
None of that is happening at the required speed.
What History Says About Workforce Transitions
Richard Florida offered a historical frame. "We used to have most people working in agriculture," he told Fortune. "Now 1% of the workforce works in agriculture. We used to have most people working in manufacturing, 50, 60% working in manufacturing. Now 5 to 6% of the workforce works in manufacturing."
The transitions happened. They were painful. They took decades. And they were accompanied by massive social disruption, the deindustrialization of the Rust Belt, the opioid crisis, the hollowing out of entire towns.
The difference this time is the timeline. Agriculture's decline unfolded over a century. Manufacturing's decline unfolded over fifty years. AI-driven occupational displacement is projected to unfold over ten.
Ten years is not enough time for a town to rebuild. It's not enough time for a worker in her fifties to retrain for a new career and recoup the investment. It's not enough time for the credentialing system to expand.
The Skeptic's Corner
McKinsey's Own Labor Record
Here's a fact that rarely makes it into the coverage of McKinsey's workforce reports: the firm has been cutting its own workforce while publishing research about the workforce transitions its clients will face.
McKinsey cut roughly 2,000 roles in 2023. BCG's new hires dropped 40%. All three of the big consulting firms publicly deny that AI is the cause. All three sell AI-driven headcount reduction to their clients as a service.
McKinsey's CEO told the world the firm saved 1.5 million human work hours last year using AI agents. That's equivalent to 750 full-time employees. The firm's internal strategy reportedly includes growing client-facing roles by 25% while cutting back-office roles by 25%.
This doesn't invalidate the McKinsey report. The data is the data. But it's worth noting that the firm publishing optimistic net-gain projections is also the firm selling the automation that drives the displacement. There's a structural incentive to frame the transition as manageable.
When Forecasts Become Self-Fulfilling
There's a deeper problem with workforce forecasting. When McKinsey publishes a number like "11 million workers will need to switch occupations," that number enters the decision-making calculus of every HR department, every workforce development agency, every state economic planner.
The forecast becomes a planning assumption. The planning assumption becomes a budget line. The budget line becomes a program. And the program shapes the reality it was designed to predict.
This is not a conspiracy. It's just how forecasts work in institutional contexts. McKinsey's numbers will be cited in boardrooms and legislative hearings for the next decade. The 11 million figure will justify spending on retraining programs. It will also justify spending on automation, framed as "preparing for the transition."
The net effect is hard to measure. But the framing matters.
The 95% Problem
McKinsey's own enterprise research has flagged an uncomfortable finding: about 95% of companies attempting to deploy AI at scale report no measurable return.
If the productivity gains aren't materializing at scale, the displacement timeline may be slower than projected. Or it may be faster, as companies cut costs to make the numbers work. The uncertainty is the point. Nobody knows how this plays out. Not McKinsey. Not the workers. Not the policymakers.
The report says 11 million. It could be 6 million. It could be 16 million. The range is the honest answer.
What This Means for Workers, Employers, and Policymakers
For workers in occupations most exposed to automation, office support, retail, transportation, administrative roles, the window for proactive transition is open now and will not stay open forever. The credential barrier means that starting early matters more than starting fast.
For employers, the report is a warning about the talent pipeline. If 770,000 workers per year need to switch occupations and only one in seven has a direct path, employers will need to build the paths themselves. Apprenticeships, internal training programs, skills-based hiring, the mechanisms exist. The scale does not.
For policymakers, the report is a funding mandate. The credentialing system needs capacity. The retraining infrastructure needs investment. The transition is coming whether the infrastructure is ready or not.
The Takeaway
McKinsey's report is not a doomsday document. The net job creation is real. The economy will have more jobs in 2035 than it has today. That's the good news.
The bad news is that the jobs will be in different places, requiring different skills, accessible through different pathways. And the people who need to move, 11 million of them, maybe more, will mostly be moving uphill.
"The next decade's challenge is mobility, not scarcity," McKinsey wrote.
Mobility is a policy choice. It's an investment choice. It's a choice about whether the people who get displaced by the transition get help crossing the bridge or get left on the wrong side of the river.
The report describes the river. It does not build the bridge.