# AI Reprices the Degree, It Doesn't Erase Demand (filed 2025)

**Core claim:** Over the next ~20 years, AI's disruption of US university enrollment is a **repricing, not a collapse**. The bachelor's bundle has three separable components — skill acquisition, sorting/credential signal, and social/network capital plus residential experience. AI compresses the price of the first (AI tutors, employer micro-credentials) and erodes the second at the median (skills-based hiring, AI-assisted screening), but the third is scarcity-priced and largely AI-insulated, and regulated occupations keep the credential legally mandatory. Demand doesn't vanish; it **migrates** — headcount drifts down slowly (~10–15% by 2040) while program mix and revenue-per-student change far faster than enrollment.

**Where demand collapses:** degrees whose value was primarily a diligence signal for routine white-collar work — generic business, general communications, similar tracks at non-selective tuition-dependent schools. First-year enrollment in these programs at exposed institutions falls steeply (25–40% by 2035); entry-level wage premium for these majors compresses toward zero.

**Where demand persists or grows:** (1) licensure-gated and physical-practice fields — nursing, engineering, health-adjacent tracks — where the university remains the legal gateway; (2) elite-brand degrees of *any* major, where the scarcity signal survives because AI makes median signals cheap, not elite ones; (3) flagships adjacent to expanding industries. Expect STEM/licensure program share of degrees to cross ~50%.

**Regional overlay:** fragmentation compounds the established map — growth concentrates at Sun Belt/SEC-region flagships and land-grants near expanding energy, defense, and health sectors; decline concentrates at small tuition-dependent privates and regional publics in the Rust Belt and rural Northeast. Florida, Tennessee,