# Research function bifurcates from teaching (2025 claim, horizon ~2045)

**Claim:** By ~2045, US university research concentrates into a top ~50–75 hub tier that grows in scale and budget, while mid-tier schools exit research (AI makes their lab output economically valueless) and the teaching-only tail shrinks/merges/converts.

**Mechanism:** AI commoditizes literature synthesis and routine data analysis — the marginal product of a mid-tier lab — while frontier AI work demands compute scale only big-hub + industry/federal money can fund. Research becomes a scale game; teaching does not.

**Regional shape:** Hub growth tracks industry clusters and state investment — Sun Belt/SEC land-grants (Georgia, Texas, Florida, North Carolina) gain; Northeast small privates and rural Midwest publics lose most.

**Falsifiable:** If federal R&D policy deliberately redistributes to mid-tier schools, or if AI tools let small labs produce frontier-relevant output cheaply, the concentration thesis weakens.