Analysis
Artificial intelligence will eliminate demand for roughly 36 million U.S. jobs by 2035 while creating about 41 million new ones, but about 11 million workers -- roughly 7% of the workforce, in a range McKinsey puts at 6 million to 16 million -- will need to leave their current occupation entirely, according to a new McKinsey Global Institute analysis reported by Fortune. The net jobs math is positive. The mobility math is the problem: McKinsey's authors frame it directly, writing that "the next decade's challenge is mobility, not scarcity."
Why Net-Positive Still Means Mass Disruption
McKinsey's own number is stark: only about one in seven workers who need to change occupations has a direct path into a growing field requiring little to no retraining. Everyone else needs a real transition -- new credentials, new skills, often a new industry. To close that gap over the next decade, McKinsey estimates roughly 770,000 people a year would need to switch occupations, about 3.6 times the historical average pace of occupational change, sustained year after year. Bloomberg's earlier coverage of the same report, published when McKinsey first released the data on September 29, cited the identical 11-million figure and noted the research lands as AI-driven job anxiety has become a mainstream economic story rather than a tech-industry one.
“Everyone else needs a real transition -- new credentials, new skills, often a new industry.”
Where the Pain Concentrates
More than 75% of the workers likely to be displaced sit in just three fields: office and administrative support, retail and sales, and transportation and logistics. That concentration tracks with a narrower but directionally similar finding Pulse covered earlier this year, when a Stanford study on AI and entry-level jobs found hiring for recent graduates falling fastest in the occupations most exposed to AI automation. McKinsey's new numbers suggest that pattern scales well beyond entry-level roles and well beyond the tech sector.
What the Headline Misses
McKinsey's 41-million new-jobs figure is a forecast of aggregate demand, not a guarantee that retraining happens fast enough, cheaply enough, or in the right geography -- a warehouse worker in a small logistics hub does not automatically become a solar technician or a nurse. The report also doesn't model what happens if AI capability gains accelerate past its 2035 horizon, or if employers cut exposed roles faster than they hire into growing ones. Prior waves of automation forecasting, including McKinsey's own earlier estimates on enterprise AI adoption that Pulse has tracked, were generally not wrong on direction but were frequently wrong on timing -- usually underestimating how slowly large organizations retrain existing staff instead of simply hiring fresh talent for new roles.
What To Watch
The open question for investors is who actually builds the mobility infrastructure McKinsey says doesn't exist yet at scale: reskilling platforms, apprenticeship pipelines, and placement services sized for a 770,000-person-a-year flow rather than today's niche bootcamp market. Most reskilling startups that raised money in the last cycle were built for a much smaller, slower version of this problem. McKinsey's own prescription -- treating mobility as infrastructure, not a corporate benefit -- is a bigger, more structural thesis than most of the workforce-tech pitches currently circulating.