# AI and the American university, 2025–2045: established claims

**Core answer.** The system bifurcates and thins rather than uniformly shrinking. Total headcount falls from ~19M toward ~14–15M by 2045 (Unverified: projection), but institutions consolidate faster than enrollment falls — displaced demand migrates to surviving apex and mission-differentiated campuses rather than vanishing. The median four-year campus thins; the selective apex and the licensed-professional segment grow or hold. AI is an accelerant throughout, not the primary cause.

**Demographic base.** US births fell from ~4.32M (2007) to ~3.59M (2023), a ~17% shortfall (Unverified on exact figures), which contracts the traditional-age college pipeline through the 2030s. Because demand migrates rather than disappears, the system-wide headcount floor sits above what raw demography alone implies.

**AI as accelerant, not cause.** Demographic contraction is the primary pressure; AI's role is to remove the option value of default enrollment and to compress from two directions simultaneously — demand-side (screening decay of the generalist degree) and cost-side (automation of the instructional stack).

**Cost stack.** Instruction is roughly 25–30% of institutional expense (Unverified). AI targets the non-tenured layers — grading, sessional teaching, advising — so savings arrive through non-renewal and attrition, not layoffs. Selective institutions convert savings into selectivity or reinvestment; price-dependent institutions must reprice. AI leaves the non-teaching layers (admin, compliance, facilities, athletics) largely untouched.

**Demand split by major.** Generalist white-collar majors decay earliest; licensed and professionally-gated fields hold. Mechanism: AI erodes the degree's *screening* function (employers can observe competence directly),