The New Front Doors

• Eli Lilly — Insilico Medicine ($2.75B licensing deal). Lilly buys worldwide rights to preclinical candidates from Insilico’s generative AI platform Pharma.AI. $115M upfront + up to $2.75B in development, regulatory, and commercial milestones, plus tiered royalties.

  ↳ Continues “Lilly Is Buying the Future” (Issue #4). The deal targets specific preclinical drug candidates discovered by Insilico’s Pharma.AI platform — Lilly licenses the molecules; Insilico keeps the platform.

  👉 Why it matters: Big Pharma can now license drugs born in AI platforms it doesn’t own. Lilly gets the molecules; Insilico keeps the engine. The model: AI-native biotechs become specialized drug-discovery vendors to pharma at scale.  Insilico PR →

• Roche to acquire PathAI for up to $1.05B ($750M upfront + $300M milestones, May 7). PathAI’s AISight digital-pathology platform — including AI-enabled companion diagnostic algorithms — folds into Roche Diagnostics. Builds on a Roche-PathAI partnership active since 2021, expanded in 2024 to include CDx development.

  ↳ Pairs with Lilly — Insilico (above) and Labcorp + KEYTRUDA CDx / Veppanu + Guardant360 (Issue #4). Pharma is now buying the AI and the diagnostic that surrounds the drug.

  👉 Why it matters: Roche is industrializing the CDx-as-distribution playbook. If the diagnostic is the launch (as we argued in Issue #4), owning the AI engine that generates the diagnostic is the new vertical-integration play.  Roche press release →

• Musely raises growth funding from General Catalyst — without giving up ownership (May 4). Most startup deals trade shares for cash; this one is structured more like a loan repaid from future revenue.

  👉 Why it matters: Loan-style funding only works for businesses with steady, predictable profits — the kind that banks and credit funds want to lend against. Profitable direct-to-consumer health (telehealth + compounded dermatology + GLP-1) has matured into a category investors now treat like an established business, not a startup bet.  TechCrunch →

• Google launches Fitbit Air — $99 screenless WHOOP competitor (May 7, ships May 26). 5.2g pill-shaped tracker, no display, week-long battery, Gemini-powered Health Coach on optional Premium tier. Hardware does not require subscription.

  ↳ Continues “WHOOP hits $10B” (Issue #1). Five weeks after WHOOP’s valuation peak, Google launches a $99 attacker.

  👉 Why it matters: The screenless wellness band category is being commoditized in real time. WHOOP’s subscription is now the only moat.  Source →

• Garmin CIRQA approaching launch (summer 2026). Trademark filings + “future of wearables” teaser confirm a screenless band is imminent.

  👉 Why it matters: Three players in 8 weeks. The category formerly known as “WHOOP” now has three competitors.  TechRadar →

• FDA launches HALO + Elsa 4.0 — the agency becomes AI-native (May 6). HALO (“Harmonized AI & Lifecycle Operations for Data”) consolidates 40+ FDA application, submission, and portal systems into a single platform. Elsa 4.0 — the FDA’s internal AI tool — now sits on top of it with custom AI agents, document generation, quantitative analysis, OCR, and charts. Generative AI use among FDA staff has gone from 1% in early 2025 to 80%+ today.

  ↳ Continues FDA AI Guiding Principles (Issue #3) and the post-January 2026 wellness/CDS guidance. FDA isn’t just regulating AI; it’s running on it.

  👉 Why it matters: The speed at which novel devices, drugs, and CDx clear FDA over the next 24 months will increasingly depend on how well Elsa + HALO move through complex submissions. “Previously, FDA staff would bring data to Elsa. Now, Elsa sits on top of our data,” — Jeremy Walsh, FDA Chief AI Officer.  FDA | Nextgov →

• Lilly inaugurates LillyPod — first NVIDIA DGX SuperPOD with B300 systems built for pharma. Drug discovery, genomics, clinical development on dedicated supercompute.

  ↳ Continues “NVIDIA + Lilly $1B Co-Innovation Lab” (Issue #3). January announcement; May infrastructure.

  👉 Why it matters: JPM announcement → live supercomputer in four months. Sets the pace every other pharma-cloud deal will be measured against.  NVIDIA blog | HPCwire →

• TytoCare receives FDA De Novo for AI-powered eardrum analysis (May 7). Creates a new regulatory category for AI-powered ENT image analysis.

  👉 Why it matters: First de novo means every AI ENT tool that follows now has a regulatory road. Proof the post-January 2026 FDA AI guidance is moving products through.  Source →

• Mayo Clinic REDMOD detects pre-diagnostic pancreatic cancer in normal-looking CT scans (1,462-scan study, reported May 11–12).

  ↳ Pairs with Revolution Medicines RAS in pancreatic (Issue #4). RAS gives us therapy; REDMOD could give us the screen.

  👉 Why it matters: Pancreatic mortality is a late-detection problem. A screen + a drug class moving simultaneously is how an “untreatable” cancer becomes a market.  Source →

• ATA Nexus Annual Conference kicks off in Orlando — May 12–14, 2026. Top virtual care event of the year. Nine Innovators Challenge finalists pitch live, including TytoCare, Twentyeight Health, Conscientia Health.

  👉 Why it matters: Watch this stage for the next wave of virtual-care partnerships and acquisition signals.  Source →

• Twentyeight Health launches “Complete Care” — first nationwide insurance-enabled all-in-one virtual women’s clinic. 100+ payers including Medicaid; ~16M women in network.

  👉 Why it matters: Insurance-enabled virtual women’s primary care has been a category waiting for a winner. This is the most credible national attempt to date.  Twentyeight Health PR →

• FDA panel recommends Truqap (capivasertib) + abiraterone for PTEN-deficient prostate cancer (same April 30 ODAC). Committee voted 7-1 with 1 abstention based on CAPItello-281 Phase III data.

  👉 Why it matters: Same meeting, opposite verdict — both biomarker-defined. The FDA’s signal: clear molecular target + mature efficacy = yes.  AstraZeneca press release →

• FDA approves Veppanu (vepdegestrant) + Guardant360 CDx for ESR1m HR+/HER2- breast cancer (Arvinas / Pfizer, May 1). First-in-class PROTAC protein degrader, co-approved with its liquid-biopsy companion diagnostic.

  ↳ Pairs with the camizestrant ODAC rejection above (same patient population, opposite verdict). Continues the drug-CDx co-launch pattern from Labcorp + KEYTRUDA CDx (Issue #4) — second co-approval in two weeks.

  👉 Why it matters: Protein degraders just leapfrogged next-gen SERDs in ESR1-mutated breast cancer, and liquid-biopsy CDx co-launched with the drug is now the default in biomarker-defined oncology. The diagnostic is the launch.  FDA | Guardant Health →

• FDA approves Shanghai Henlius IND for cetuximab biosimilar HLX05-N (May 10). Phase 1 in metastatic colorectal cancer.

  👉 Why it matters: Chinese-originated biosimilars entering the US pipeline at scale. Leading indicator of the China-pharma wave already reshaping licensing.  Source →

👉 The moat collapsed in 6 weeks. WHOOP’s premium hardware-plus-subscription model now sits next to a $99 Google device with no mandatory subscription. The band isn’t the moat anymore.

👉 Why this matters: ~120M U.S. adults have hypertension; most managed on three readings a year. The first wearable that delivers reliable continuous BP from the wrist wins a market 3–5× larger than fitness tracking — and ties directly into CMS ACCESS reimbursement.

📋 The DTC Landscape · 2026

👉The Lilly playbook from Issue #4 is also about owning the patient end-to-end. Roche just bought PathAI for $1.05B this week to do the same thing on the diagnostic side. Wearables and DTC are running the same play from the other direction. Whoever closes the loop first owns the patient layer for the next decade.