The June 9, 2026 Medicare Trustees Report is the most consequential datapoint this scan produced: for the first time a federal actuarial body formally named "expected use of GLP-1 weight loss drugs and expensive specialty drugs" as the primary structural driver of Part D's projected 58% spending increase, from roughly 0.6% of GDP today toward 1.0% by 2099. That language hands CMS a Trustees-backed mandate to include at least one GLP-1 in IRA cycle 3, effective FY2028, with an expected 25–60% price cut on the selected molecule. The theme, read against the thesis, is three separable bets stacked on top of each other — and the Trustees report reshapes the risk profile of the largest.
Eldercare pharmacology and the IRA trap
The GLP-1 franchise reframed as an ageing drug — cardiac protection in SELECT, renal protection in FLOW, exploratory cognition trials — is the clearest manifest case of the C-10 demographic-reversal wave and the highest-conviction leg on structural demand direction. The problem is that success and danger arrive simultaneously: every positive outcomes readout broadens the eldercare-therapeutics tailwind and simultaneously accelerates IRA inclusion. Both signals fire at once. Between Eli Lilly and Novo Nordisk, LLY carries more cycle-3 exposure — Mounjaro's T2D indication is a larger Medicare Part D line than Wegovy's obesity-only label, and the dual Mounjaro/Zepbound structure is what CMS actuarial calculus finds hardest to bypass. Single-asset GLP-1 pure-plays priced without that IRA discount embedded do not belong in the portfolio. BioAge (BIOA) is the one speculative tail in the geroscience adjacent layer worth carrying: conventional clinical development informed by geroscience biomarkers, sized as a binary option.
Platform enablers
The compounding assets earn a rake on modality growth without carrying single-programme binary risk, and the BIOSECURE Act — naming WuXi Biologics, WuXi AppTec, BGI, MGI, and Complete Genomics — is a direct accelerant for US-domiciled expansion. FDA-cleared aseptic-processing qualification requires 18–36 months of process transfer per customer; that cycle is the moat. Lonza, Samsung Biologics, and Thermo Fisher's Patheon and Resilience platforms are the beneficiaries. Novo Holdings' $16.5B acquisition of Catalent in 2024 is the sovereign-adjacent template for how capacity gets locked up; NVO is the public proxy. One layer down, MaxCyte (MXCT) has its Flow Electroporation platform embedded in Casgevy and roughly twenty other cell-therapy programmes via Strategic Platform Licence agreements generating per-asset milestones and royalties — a diversified stream on modality growth. Danaher, Thermo Fisher, and Sartorius supply bioprocessing consumables across the GLP-1 fill-finish constraint and the broader cell and gene therapy build-out alike. In sequencing, Illumina holds roughly 70% short-read share but faces price-per-genome pressure from BGI and MGI, partially offset by BIOSECURE; Twist Bioscience spans synthetic DNA across every modality; PacBio is the primary long-read challenger.
CRISPR and the care-labour ceiling
Casgevy — approved for sickle cell in December 2023 and β-thalassemia in January 2024 — remains the only commercially launched CRISPR therapy in the corpus thirty-plus months on. Vertex's FY2026 Casgevy range is estimated at $150–300M, CRISPR Therapeutics' collaboration share at $80–150M: well below what XBI embedded at approval. The binding constraint is Authorised Treatment Centre throughput. Each patient requires apheresis, myeloablative conditioning, roughly six weeks of ex-vivo editing at Lonza, infusion, and a multi-month post-acute course. That post-acute leg lands in the same skilled-nursing capacity wall NIC MAP Q1 2026 flagged: 0.4% year-on-year inventory growth — a record low — against 89.5% occupancy. The 30% of home-health aides who are foreign-born (C-12) are not merely a landlord problem — though on that point, Welltower and Ventas on RIDEA structures participate in operating economics and are insulated from margin compression, while Omega Healthcare and Sabra as triple-net skilled-nursing landlords are directly exposed to their operators' care-labour costs. Structurally, that labour bottleneck is the CRISPR revenue ceiling. If FY2026 Casgevy comes in below $100M, the ATC constraint is confirmed as structural and ex-vivo multiples compress across Beam Therapeutics, Editas, Intellia, Prime Medicine, and CRISPR Therapeutics alike.
The escape valve is in-vivo delivery. CRISPR Therapeutics' CTX310/320/321, CTX340, and CTX460 programmes all use lipid-nanoparticle delivery — no ATC bottleneck, scales like a small-molecule franchise. Lilly's 2025 acquisition of Verve is the market's own signal that big pharma will pay premium multiples for in-vivo while remaining sceptical of ex-vivo economics.
The AI drug-discovery signal is the silence
The scan window closed with no Phase 2 or Phase 3 AI-drug-discovery readout sufficient to shift priors. That absence is the finding — direct support for C-04. Recursion, post its Exscientia merger, generates milestone-shaped R&D services, not subscription SaaS; its REC-994 Phase 2 in cerebral cavernous malformation delivered mixed results in 2025. Schrödinger is the only public AI-drug-discovery name with genuinely recurring software revenue. Insilico's INS018_055, a Phase 2a in IPF, is a single datapoint. Isomorphic Labs has $1.2B and $1.7B deals with Novartis and Lilly respectively but remains pre-clinical inside Alphabet — no public vehicle. Aggregate AI-attributable drug-discovery revenue across the public cohort is well under $2B per year, and commoditisation of the foundation-model layer will erode AI-DD platform pricing further.
One financing risk the public equity framing obscures: Apollo, Blackstone, and KKR direct-lending vehicles, alongside Royalty Pharma cash flows, are the primary capital conduit when public biotech windows close. A private-credit mark deterioration would compress biotech capex faster than the equity window would — and maps directly to C-08 and C-21 in the manuscript.
What would change my mind
Three things force genuine revision. Vertex FY2026 Casgevy net product revenue above $300M would indicate the ATC bottleneck is clearing faster than the skilled-nursing capacity data implies, requiring ex-vivo multiple upgrades across the cell-therapy cohort. A Phase 3 pivotal readout from an AI-designed asset at compressed timeline versus standard development would break the C-04 leg — watch Recursion, Schrödinger, and Insilico partner readouts for that signal. IRA cycle 3 exclusion of GLP-1s — possible only if CMS selects a non-GLP-1 molecule — would remove the largest known headwind on LLY and NVO near-term earnings. Before any of this goes to position-sizing, I need to clear state/thesis_revisions/2026-Q3-C-07-household-balance-sheet-floor.md and state/thesis_revisions/2026-Q3-C-10-demographic-tailwind-reversal.md; both show contradicting evidence flags in Q3 2026, and C-10 in particular is load-bearing across the entire eldercare leg.
What I'm watching next
The IRA cycle 2 final selection list is the highest-priority external fill — the CMS extraction returned a 403 on the last pull, and mid-2026 finalisation at more than 50% off list for any branded biologic with an eldercare-relevant indication resets sector-wide pricing architecture. Alongside that: direct Casgevy net product revenue from Vertex and CRISPR Therapeutics 8-K accessions 0000875320-26-000171 and 0001193125-26-204147; a verified 2024–2026 CRISPR approval ledger confirming no second approval beyond Casgevy and Lyfgenia; the Trump administration's Makary-FDA and RFK Jr.-HHS 2025–2026 regulatory actions ledger, absent from the corpus in dated primary-source form; and XBI versus SPY year-to-date 2026 — the live print on whether the capital-stack sponsorship discount is widening or compressing.