Stanford just proved AI isn't replacing workers. It's replacing the pipeline that creates them.

Erik Brynjolfsson's team at Stanford just updated their ADP payroll analysis through June 2026.

The number that should keep every workforce planner awake: employment for 22-to-25 year-olds in AI-exposed occupations is now 19% below where it should be.

Up from 15% one year ago.

Here's what makes this different from every other AI-jobs headline.

Nobody is being fired. No WARN notices. No layoff announcements.

The jobs simply aren't being created. A team requests two junior headcount, gets approved for one experienced hire, and the two graduates who would have filled those seats never learn the roles existed.

Repeat that across enough firms and the effect accumulates entirely at the entrance while the building above it stays fully occupied.

Experienced workers show zero comparable gap. The adjustment is running exclusively through hiring.

This is the succession crisis hiding inside your AI efficiency gains.

Every skipped junior hire is a senior you'll have to buy on the open market in five to ten years, at market rates, in a market where every competitor made the same choice.

SignalFire's data confirms it: new-grad hiring at major tech employers is down 65% against 2019. At early-stage startups, it's down 76%.

The Harvard team tested whether firms were backfilling juniors into more complex work as AI absorbed the routine layer. They found no evidence of it happening at scale. Roles were removed rather than upgraded.

Audit your seniority mix before importing anyone's headline number. If junior share has fallen faster than total hiring, you have the pattern locally regardless of what causes it nationally.

Model the cost of a senior you'll need to recruit in 2032 against this year's headcount saving before treating the saving as free.

SOURCE: https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/
VERIFIED: Stanford Digital Economy Lab (August 2026 revision), ADP Research, AI2Work analysis (September 1, 2026), Fortune (June 27, 2026)
SIGNAL: The invisible workforce crisis. Companies think they're cutting costs with AI. They're actually deferring a leadership vacuum that will hit in 5-10 years when every firm needs seniors and nobody trained them.