**Claim (2025, horizon 2045):** Over the next 20 years, US universities split into two functionally distinct systems. The **research function concentrates**: federal agencies and industry channel AI-era funding to a hub tier of fewer than ~50 campuses with the compute, data, and talent density to compete (Johns Hopkins, MIT/Stanford/CMU, UC flagships, UT/Austin + A&M, GaTech, UW, and new federally designated AI-institute sites). Minimum viable research scale rises past what a 200-PI regional can field, so the **median university exits research production** and becomes a teaching-and-credentialing institution staffed by learning engineers rather than professor-researchers. The hub tier grows in dollars, headcount, and footprint; the teaching tier persists only where states subsidize it or price is defensible. Geography follows funding: research mass migrates toward state capitals that invest and coastal industry adjacencies — the SEC's rise is state-capital-driven, not conference-driven. **The aggregate count of US universities likely holds steady to 2045 while the median campus roughly halves in enrollment and research output** — growth-vs-shrink is a distribution question, not a total.