America’s labor force is shrinking at a pace last seen outside a pandemic, and the exit is concentrated among the workers HR departments can least afford to lose all at once. That shortfall, not a productivity fad, is what is really driving employers deeper into AI adoption.
A faster retreat from work
The labor force participation rate among workers 55 and older fell to 36.9% in July, down from 38.1% a year earlier, according to RSM US chief economist Joseph Brusuelas. A 1.2 percentage point drop in a single year is unusual outside a recession; RSM notes it is broadly comparable to the 1.7 point decline recorded during the worst of the pandemic labor exodus. “We are simply witnessing a historic exit from the American labor market,” Brusuelas wrote in the firm’s Market Minute commentary.
Unlike the pandemic-era exit, this one has no clear reversal point. Older workers who left the labor force in 2020 and 2021 eventually returned as health risk eased. The current retreat is structural: it tracks the aging of the population, not a temporary shock.
The demographics behind the drop
There are 27 million more Americans age 65 and older than there were in 2005. That cohort made up 12% of the population in 2005; by 2025 it was 18%. The baby boom generation began aging into retirement roughly two decades ago, and its oldest members are now approaching 80. Brusuelas ties the acceleration to more restrictive immigration policy as well: negative net migration returned in 2025 for the first time in more than 50 years, with over 605,000 deportations and roughly 1.9 million voluntary departures reducing the pool of workers who might otherwise offset an aging domestic workforce.
Put together, labor supply growth slowed to 0.77% over the past year, well below the roughly 1% average the US economy relied on in the decades after World War II. RSM’s math suggests the economy now only needs to add about 35,000 jobs a month to keep the labor market stable, a fraction of what it took a decade ago.
Demand is not waiting on supply
The Bureau of Labor Statistics reported payrolls fell by 23,000 in July while unemployment held at 4.1%, an outwardly weak headline. Yet hiring plans jumped 47% from June to reach the highest July total since 2022, according to Appcast and ManpowerGroup data reported by HR Executive, and layoffs sit near a two-year low. The two data points describe the same problem from opposite sides: fewer available workers, even as employers keep plans to hire alive.
Some employers are resolving the mismatch with automation rather than headcount. Visa cited AI-driven efficiency gains in a 7% workforce reduction this year, one of several companies treating AI as a direct substitute for scarce labor in specific functions. At the macro level, AI spending is projected to reach $2.6 trillion in 2026, a 47% jump from $1.76 trillion in 2025, spending that is arriving at the same moment the labor pool it is meant to backfill is contracting.
Where the demand actually sits
The hiring that is happening is not spread evenly. ManpowerGroup regional president Ger Doyle and vice president Bekir Atahan, whose Experis division tracks placement demand directly, describe the shift in blunt terms: “the demand is showing up in aerospace, energy and manufacturing, work that happens on a floor rather than a screen.” Leisure and hospitality lost more than 80,000 jobs across June and July combined. Healthcare added 22,000 jobs in July, a gain that still came in below its historical average. Local government payrolls fell by 50,000.
Read together, the picture is not “AI is taking jobs” in the aggregate. It is a labor market splitting along two axes at once: physical, floor-based work is hiring despite a shrinking overall pool, screen-based and back-office roles are shedding headcount fastest, and the aging exit is thinning the supply available to fill either category. For an HR leader building a workforce plan, treating “labor shortage” and “AI displacement” as opposite stories misses that they are the same story, viewed from different functions.
What it means for the HR leader
Treating this as a cyclical hiring slowdown misreads the data. Workforce planning and people analytics tools need to model a structurally smaller labor pool for the rest of the decade, not a temporary dip that resolves once “the economy improves.” That has direct implications for how HCM platforms should be configured and used.
Three moves follow from the data. First, run a skills exposure audit against the 55-and-older segment of the workforce specifically, since that is the group leaving fastest and often holding the deepest institutional knowledge. Second, build phased retirement and structured knowledge transfer into HCM and succession workflows now, rather than treating it as an HR-process afterthought, given how little time some roles have before a departure. Third, be skeptical of AI-adoption business cases pitched purely as productivity plays. As the productivity payoff from AI investment is still mostly a promise in the data employers themselves are reporting. The stronger case for AI in this environment is scarcity management: covering work a shrinking labor pool cannot fill, not a general-purpose headcount cut. That distinction matters when the same automation trend is also compressing wage growth for the workers who remain.
The labor pool is not coming back to its old size. The employers who plan for that now, rather than waiting for a rebound that demographic data does not support, will be the ones best positioned when the next hiring cycle turns.
Source: RSM US
