Let me parse what's being asked here. This is a long-running "Mind" note-taking exercise about whether US universities grow or shrink due to AI over 20 years, with regional dependence. Many notes have been filed. The current facet is: Cost structure — AI reduces need for faculty/TA, tuition pressure.

The session substance from Note 8 gives the new durable claim:
- The credit-hour pricing model is the real cost moat, and AI erodes it
- Universities price per seat; AI-native competitors can price per verified outcome
- When instruction quality is verifiable outside the degree, the accreditation-backed seat-based revenue model loses its cover
- The flip from per-seat to per-outcome pricing is the mechanism by which AI shrinks university revenue, not just university cost

So the increment here: prior cost-structure notes covered (a) AI absorbs grading/tutoring/TA labor first, (b) instructional savings are competed away, not banked — pass-through as discounting, (c) tuition shows disruption first, (d) ≥25% instructional staffing cost cut by 2035 outside top-50 (IPEDS falsifiable). This note adds: the unit-of-pricing claim — universities price per credit-hour/seat; AI-native competitors price per verified outcome; when instruction quality is verifiable outside the degree, the accreditation-backed seat-based model loses its cover; the per-seat → per-outcome flip is the revenue-shrink mechanism.

Wait — but Note 8's opening line shows a title already drafted: "# Degrees lose signaling value asymmetrically — fastest in generalist white-collar fields where AI lets employers verify output directly, slowest where statute or failure cost enforces the credential — so by 2035 the median degree is a positional luxury marker, not" — that's a prior note title (the signaling facet). The actual new substance is the credit-hour pricing moat claim.

Also Note 7 shows that a "revise as a set" instruction happened at some point — one consolidated note synthesizing everything. But then Note 8 continues filing a new facet claim. So the current instruction (bottom of Note 8) is: write ONE Markdown note with a single dated claim for the cost-structure facet, per the standard format:

- # title = claim in one line
- 1-3 short paragraphs
- End with "What would change this: …"
- Name a source for load-bearing numbers or start sentence with Unverified:
- Don't mention tools, sessions, passes, note-writing
- Don't invent evidence
- Single dated claim

So I should NOT produce the full consolidation again — the latest instruction is the facet-level one: "Write one Markdown note: a single dated claim this Mind can stand behind." The current facet is cost structure. The new substance is the per-seat vs per-outcome pricing moat.

Let me check what the existing cost-structure notes covered so I don't repeat:
- Note from earlier (cost structure): AI absorbs grading/tutoring/TA labor first; instructional savings competed away, not banked — oversupplied market forces pass-through as discounting; tuition-dependent middle gains no margin, only price war; middle compressed from both sides; tuition, not faculty headcount, is where disruption shows first; ≥25% instructional staffing cost cut by 2035 outside top-50 (falsifiable via IPEDS).

So the increment: the pricing-unit flip. Universities price per seat (credit-hour); AI-native competitors price per verified outcome. When instruction quality is verifiable outside the degree, the accreditation-backed seat-based revenue model loses its cover. Per-seat → per-outcome pricing flip is how AI shrinks revenue, not just cost.

This connects to: revenue shock before headcount (buyers defect); signaling (verification outside the degree); program migration (credential-signal → verified-skill). But the pricing-unit mechanism is distinct and durable.

Mechanism articulation:
- The credit hour is a unit of time, not of learning. Its moat is accreditation + federal aid (Title IV flows through credit hours) + employer trust. AI breaks the third leg directly (employers can verify skill via work samples/AI assessment) and pressures the first (competency-based education already exists — Western Governors University is the real-world precedent).
- Western Governors University: real, competency-based, priced per term not per credit, ~$4,000-ish per six-month term (roughly $3,800-$4,700 depending on year). That's a citable real precedent that per-outcome/per-competency pricing can operate at scale under accreditation. WGU enrollment ~130k-150k students, one of the largest universities in the US. That's real and checkable. I should be careful with exact numbers: WGU is the largest degree-granting university in the US by enrollment (over 130,000 enrolled around 2022-2023). Tuition roughly $3,800 per six-month term