Stanford just proved AI is killing entry-level hiring. 19% below expected. Your successor pipeline is collapsing.

Erik Brynjolfsson's team at Stanford just dropped the hardest data yet on AI workforce impact.

Using ADP payroll records covering millions of U.S. workers through June 2026:

Workers ages 22-25 in AI-exposed occupations are now 19% below where they should be.

Not 19% below last year. 19% below the trajectory they were on before AI adoption accelerated.

The gap was 15% in July 2025. It's 19% now. It's widening.

Here's what makes this different from every other AI doom headline:

This isn't about layoffs. Separation rates actually fell. Companies aren't firing young workers.

They stopped hiring them.

The entry-level pipeline — the people who would become your senior engineers, your domain experts, your managers in 5-10 years — is being quietly dismantled.

And it's concentrated exactly where you'd expect: occupations where AI substitutes for human tasks. Not where it augments. Where it replaces.

Meanwhile, experienced workers in the same occupations show no comparable gap. The automation hit is age-graded.

Your workforce plan just became a succession crisis.

Audit your hiring data by age cohort and AI exposure. If your entry-level headcount in automatable roles has flattened or declined since 2023, you're already behind. The talent you need in 2030 isn't being trained today.

Stanford calls these "canaries in the coal mine." The canaries are dead. Act accordingly.