# AI thins the median American university while the top decile grows — the system consolidates rather than shrinks

**Core answer:** Over 2025–2045, AI does not shrink the university system uniformly; it thins it at the median and grows it at the top. Total enrollment drifts from ~19M toward ~14–15M by 2045 (Unverified: projection), but institution count falls faster than degrees conferred — campuses close and consolidate while output concentrates into fewer, larger schools. The barbell: selective flagships and elite privates grow; tuition-dependent regionals and small privates close. AI is an accelerant of an already-in-motion demographic and pricing squeeze, not the primary cause.

**Demographic floor (the part already locked in):** US births fell from 4.32M (2007) to 3.59M (2023) (CDC/NCHS). High-school graduates peak around 2025–26 at ~3.9M and decline ~13% by 2041, with the demographic cliff hitting 2029–2037 and hitting the Northeast, Midwest, and Plains hardest (WICHE). This alone guarantees closure at the weakest schools regardless of AI. AI determines how many additional schools close beyond the demographic minimum.

**Credential value splits into two components:** The degree survives as an *insurance* credential — legal licensure, trust, directional competence — while losing value as a *screening* signal, because AI makes candidate competence observable pre-hire (Unverified: mechanism timeline). Selection-based signals (admission to a selective school) are AI-proof; certification-based signals (the diploma itself, undifferentiated by major or program) erode. Licensure-gated fields (nursing, engineering, teaching) hold; generalist white-collar programs erode fastest.

**Program axis cuts across every institution:** Most exposed are generalist business, communications, and similar programs at non-selective schools. Least exposed are licensure-gated and physical-practice programs. AI grades the *degree*, not the program — a selective school's generalist program survives; a non-selective school's does not.

**Regional axis:** Growth flows to the Sun Belt (TX, FL, TN, AZ); the Northeast, Midwest, and Plains contract. But geography is secondary to endowment depth and licensure portfolio: a wealthy school in a shrinking state outlasts a poor school in a growing one. Within growth states, headcount growth accrues to lower-price and 2-year institutions, so raw demographic growth does not translate into four-year tuition revenue growth (Unverified: composition detail).

**Cost side:** AI cuts instructional cost 10–20% by 2035, enabling sub-$15K degrees (Unverified: magnitude). Competitive schools do not pass savings through — they reinvest in selectivity. Price collapse hits only the non-competitive segment, where it is a survival response rather than a strategy.

**Timeline:** 2025–2032