Invest3 publishers3 min readPublished
McKinsey's AI forecast needs near-pandemic rates of US career switching through 2035
McKinsey Global Institute's base case has about 11 million US workers leaving their occupations by 2035 because of AI and automation. The cost of that shift will be paid in training and relocation, since most of the growing jobs require a credential and cannot be done remotely.
The Investor · Invest desk
What happened
- McKinsey Global Institute estimates AI and automation will cut demand for about 36 million US jobs by 2035 while growth elsewhere adds about 41 million.
- About 770,000 workers a year would need to switch into an entirely different field, such as retail to healthcare, about 3.6 times the historical average.
- During the pandemic, from 2019 to 2022, about 788,000 workers a year made similar moves without lasting damage, according to the report.
- Only one in seven displaced workers has a direct path into a growing job that needs little retraining and pays at least as much.
- McKinsey found that the chance of having to change occupations is 7.6 times higher for lower-wage workers than for higher-wage ones.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Training offered only in the towns where office and retail jobs disappear will leave many workers qualified for growing jobs that are physically located somewhere else.
- cost Retraining budgets spread evenly across a workforce will put much of the money where the need to change careers is smallest, because the burden sits with lower-paid workers.
- decision Because the forecast ends with more jobs than it starts with, an employer's AI workforce plan turns more on how many outside candidates it can credential than on how many roles it cuts.
Divide 770,000 by 3.6 and the long-run average for cross-field switches comes out near 214,000 a year [2]. Against that history the base case is steep, or rather, it is steep against the long average and about 2% short of what workers actually did during the pandemic [3].
The difference is duration. The pandemic surge was brief, and outside it workers change employers less often than they did in the late 1990s and early 2000s [8]. The base case needs something close to the surge rate every year through 2035 [2].
On totals, the authors have the better of it. Net of losses, the forecast leaves the economy about 5 million jobs ahead [1]. "The next decade's challenge is mobility, not scarcity," they wrote [3].
The problem is getting into the new jobs. The shrinking work is in office and administrative support, retail and transportation, much of it lower-paid, and the growth is in healthcare, construction and management [9]. If the one-in-seven direct-path ratio holds across the 11 million, about 1.6 million people can step across without much retraining [4]. Nearly half face what the report calls an "unpaved" path, blocked by large skill gaps or credential requirements [12]. Applied to the same 11 million, that is about 5 million people [5]. About 85% of growing jobs require a credential [13], and about 76% cannot be done remotely [14].
Richard Florida, the urbanist and author of The Rise of the Creative Class, was not involved in the report and told Fortune that shifts like this have happened before [15]. "We used to have most people working in agriculture. Now 1% of the workforce works in agriculture," he said [16]. He expects displaced service workers to land in what he called "a broad bucket" of wellness work [17]. "I think that there will be some displacement," he said. "But I think this is also the area we're going to create the most work." [18]
The base case can miss in either direction. McKinsey's range runs from 6 million to 16 million [4], and the top is about 45% above the central figure [6]. The forecast has also eased since 2023, when the firm projected 12 million switches by 2030 [7]. The new number is slightly smaller and spread over five more years [7]. Fortune's account of the report does not put a cost on retraining or say whether the losses arrive before the gains.
I think the authors are right that job totals are not the constraint. That view fails if cross-field moves drift back toward the roughly 214,000-a-year average while office and retail demand keeps falling [2]. In that case the shortfall against 770,000 a year would show up as people out of work.
What to watch
- Whether annual cross-field moves hold near the 2019-2022 pace of about 788,000 or slide back toward the roughly 214,000 long-run average.
- Where actual displacement lands within McKinsey's 6 million to 16 million range as office, retail and transportation demand falls.
- Whether McKinsey or large employers put a dollar cost on credentialing the roughly half of displaced workers facing an unpaved path.