## AI disruption of US universities will show up as regional churn, not even decline (2025–2045 horizon)

**Claim:** Over the next 20 years, AI shrinks the part of US higher education that sells brand-signal credentials — small and mid-tier private liberal arts colleges and mid-tier residential publics concentrated in the Northeast and Midwest — while enrollment demand concentrates in campuses adjacent to AI-employing industry: Sun Belt metros, land-grants with scaled computer science and engineering capacity, and applied-technical/community segments feeding that work.

**Mechanism:** Credential value decays in proportion to how much of its price is justified by brand signal rather than demonstrated skill. Because the decaying and surviving segments occupy different regions, the system's aggregate size (roughly flat to slightly down) hides a redistribution of several percentage points of demand between segments and regions. "Shrink" is real but local; "grow" is real but regional.

**Falsifiable markers:** (1) Tuition discount rates at signal-premium privates breaking past ~60% by the early 2030s; (2) employer removal of degree screens starting in tech-hub AI-adjacent hiring; (3) tightening waitlists at land-grant/SEC-band engineering campuses against rising empty seats in the Northeast/Midwest private segment.