The Reskilling Math Doesn't Work If Workers Can't Afford to Do It

The optimists are technically right. McKinsey Global Institute (MGI) finds that AI is, on net, creating more jobs than automation displaces across the U.S. economy. But buried in that headline is a number that should stop every chief human resources officer (CHRO) cold: approximately 11 million American workers will require active help transitioning into those new roles over the course of this decade, according to that same McKinsey MGI research. "Net positive" is a macro-economist's comfort; it is not an organizational strategy.

Here is the self-contained finding that every board presentation this quarter should carry: McKinsey MGI research estimates that approximately 11 million U.S. workers will need structured transition support into growing roles by 2030, even as AI-driven automation produces a net increase in total employment. The aggregate looks fine. The distribution is the crisis.

What does it actually cost a worker to reskill?

The answer, it turns out, is less than most employers assume — and that gap is the problem. Workers report they would spend, on average, no more than $2,500 USD out of pocket to prepare for a new career, according to a Penn Foster survey cited by HR Dive (2025). A separate HR Dive roundup (2025) confirmed the same ceiling. That $2,500 USD figure is not a sign of low ambition — it is the signature of a cost-of-living squeeze that has already consumed household slack. Workers are not refusing to invest in themselves; they are capped out before the investment starts.

The implication is uncomfortable: if your reskilling strategy is built on employee co-investment, it is already broken for the workers who need it most. The $2,500 USD threshold self-selects for workers who already have financial cushion, which means employer-subsidized programs are likely reaching the employees least at risk and missing the 11 million who are most exposed.

Why is employer-led reskilling still moving so slowly?

Part of the answer is structural confusion about where the real cost sits. Moving an artificial intelligence (AI) tool from pilot to production can nearly quadruple its total cost, according to McKinsey experts (2025). Organizations that have already absorbed that shock on the technology side are now reluctant to take on a second large cost center in workforce transition. The sequencing is backward: companies deployed the automation, deferred the human investment, and are now discovering that the gap between displaced roles and available talent is not closing on its own.

There is also a judgment problem at the strategic layer. McKinsey's analysis of AI and corporate strategy (2025) makes the point that analysis now runs on AI time, but judgment still runs on human time — and decisions about workforce architecture fall squarely in the judgment category. Nobody's large language model (LLM) is going to tell you which 200 roles in your logistics division need retraining versus managed exit. That call requires leadership attention that has, so far, been rationed toward product and engineering.

Workforce transition, plainly defined: Workforce transition — sometimes called labor market reallocation — is the process by which workers whose current roles are automated or eliminated gain the skills, credentials, and connections required to move into occupations where demand is growing. It is distinct from upskilling in that it assumes the worker's existing role may not survive, not merely that it will evolve.

The workforce transition problem is not, at its core, a training content problem. Platforms offering micro-credentials are not scarce. The bottleneck is financial runway: can a 45-year-old warehouse supervisor afford six months of reduced income while retraining for a healthcare support role that pays comparably? At a $2,500 USD personal spending ceiling, the answer is almost certainly no, regardless of how good the curriculum is.

What should senior leaders actually do with this?

The frame that matters here is not "how do we offer training?" but "how do we fund the transition period itself?" Income support, internal mobility bridges, and cohort-based redeployment programs are structurally different from course libraries — and far more expensive. Before your next board conversation on AI investment, the honest question is whether your workforce transition budget is sized to match your automation ambition. If the AI spend is in the tens of millions USD and the transition budget is in the low seven figures, you are not managing a workforce strategy; you are managing a liability that has not yet matured.

The 11 million workers who need help this decade are not an abstraction. A meaningful share of them work for organizations represented in this readership. The net-positive headline does not absolve anyone sitting in the CHRO or chief operating officer (COO) seat — it just gives them a talking point while the clock runs.