The current facet is the research function — universities as research hubs vs teaching institutions. The substance from this session adds a durable corollary: co-funding follows co-location, so by 2040 the set of universities running top-quartile AI-adjacent PhD programs converges on the set of campuses adjacent to major industry research clusters; the SEC/Sun Belt rise is industrial adjacency; non-adjacent R1s reclassify as teaching-first; total research output roughly flat but concentrated in ~25-35 geography-advantaged hubs; falsifier is remote research work decoupling funding from geography.

I need one Markdown note, title = claim in one line, 1-3 short paragraphs, end with "What would change this: …". Sources for load-bearing numbers or Unverified: tags.

Load-bearing facts:
- Adjacent campuses listed: Austin (UT), Atlanta (Georgia Tech/Emory), Raleigh-Durham (NC State/Duke/UNC), Nashville (Vanderbilt), Phoenix (ASU), Salt Lake (Utah). These are factual adjacencies — no source needed really, it's geography.
- Prior filed notes (referenced in Note 8's history) already established: NSF HERD shows federal share of academic R&D ~50% and industry ~5-6%; CRA Taulbee Survey shows industry support for CS PhD students rivaling/exceeding federal. I can reference those as anchors for the co-funding trend. The HERD survey numbers: federal share of academic R&D has declined to roughly 50%; industry funding around 6-7%. That's real per NSF HERD. Taulbee: industry support for CS grad students has grown to rival federal — I recall Taulbee 2021-2022 showing industry ~38% vs federal ~30-something for CS PhD students. I'll hedge with "rivaling" and name CRA Taulbee Survey.
- The 25-35 hubs figure and the 2040 convergence: that's the forecast itself — Unverified/forecast.
- "Total research output roughly flat" — forecast, Unverified.

Structure:

# Title: something like "By 2040 top-quartile AI-adjacent PhD programs converge on campuses adjacent to industry research sites — the SEC/Sun Belt research rise is industrial adjacency, and non-adjacent R1s reclassify as teaching-first"

Make it one line but claim-shaped. Maybe: "# By 2040, industry co-funding redraws the US research-university map along corporate corridors — hubs within ~1 hour of major tech/corporate labs (Austin, Atlanta, Raleigh-Durham, Nashville, Phoenix, Salt Lake) capture AI-adjacent PhD training, while non-adjacent R1s reclassify as teaching-first; total research output stays roughly flat but concentrates into ~25-35 geography-advantaged hubs"

That's long but it's the claim in one line.

Paragraphs:
1. The mechanism: industry co-funding follows co-location. Fellowship dollars, joint labs, and mentor-tier faculty flows prefer proximity; apprenticeship (the one research input AI can't substitute) requires physical or near-physical density. Anchors: NSF HERD shows industry's share of academic R&D (~6%) small but rising while federal share drifts down toward half; CRA Taulbee already shows industry support for CS PhD students rivaling federal support.

2. The claim: by 2040 the set of universities running top-quartile AI-adjacent PhD programs converges on the set of campuses adjacent to Fortune-500 research sites — SEC/Sun Belt rise is industrial adjacency, not football prestige. Non-adjacent land-grants and heartland/Rust Belt R1s lose stipend dollars and the mentor tier, can't outbid industry for human mentors, and reclassify as teaching-first comprehensives. Net: American research university doesn't shrink overall — it relocates and re-tiers; output roughly flat, concentrated in ~25-35 geography-advantaged hubs plus a federal-lab tier. Mark forecast numbers Unverified.

3. Maybe fold falsifier intro into para 2, then "What would change this:"

What would change this: if distributed/remote research norms in industry persist past 2030 and decouple funding from geography, mid-continent R1s retain parity and regional redistribution fails; also if federal funding surges (e.g., a new national-lab-style federal program) re-anchoring stipends regardless of location; or if Taulbee/HERD data through 2030 show industry share plateauing rather than rising.

Keep it 2-3 short paragraphs. Don't mention tools/sessions/notes. Don't invent evidence.

Date: prior notes used "(assessment 2026)". I'll include a dated claim marker. The instruction says "a single dated claim." I'll add "(assessment 2026)" to the title or first line. Prior notes in the set