Roche

Phase A — Understand the business

Company Overview

Roche is a two-engine Swiss healthcare group: a Pharmaceuticals Division (the profit engine — CHF 47.67B 2025 sales, +9% CER) and a Diagnostics Division (CHF 13.85B, +2% CER), for CHF 61.52B group sales, +7% CER / +2% CHF in 2025. It is the only company at scale that owns both a top-tier drug pipeline and the #1 in-vitro diagnostics franchise — the "personalised healthcare" thesis is that companion diagnostics feed the drug business and vice-versa (e.g. sequencing-based CDx routed straight into oncology development).

How it makes money. Branded, patent-protected biologics and small molecules sold to payers/providers, concentrated in four therapeutic areas management names as its BD targets — oncology/haematology, immunology, neurology, ophthalmology. Revenue is prescription-volume × net price, not recurring/contracted; the model lives or dies on the patent clock and the replacement rate of the pipeline. Diagnostics is a razor/razor-blade model — placed analysers (Cobas, etc.) pull high-margin closed-system reagent consumables under multi-year hospital-lab contracts.

Top products (2025 full-year sales, CHF):

The five named growth drivers — Phesgo, Xolair, Ocrevus, Hemlibra, Vabysmo — did CHF 21.4B, +CHF 3.2B CER vs 2024, more than offsetting the −CHF 0.7B from the legacy biosimilar-eroded trio (Avastin/Herceptin/MabThera + Lucentis/Actemra). The business has already digested its first patent cliff (the Avastin/Herceptin/Rituxan biosimilar wave of 2019–2023) and re-based on a newer book — that is the single most important structural fact about this name.

Customers/suppliers/competitors. End buyers are national payers and US PBMs/insurers; the US is 53% of pharma sales (Lens 4). Suppliers are CDMOs and its own manufacturing (incl. a large and growing US footprint — Lens 9). Competitors are name-by-name by franchise (Lens 3): Novartis (MS), Novo/Pfizer/Sanofi (haemophilia), Regeneron/Bayer (ophthalmology), Lilly/Novo (the obesity entry it is trying to make).

Supply Chain

Upstream → Roche → end customer, named:

• Upstream inputs: biologics manufacturing (mammalian cell-culture bioreactors, fill-finish), small-molecule API, and delivery-tech licensors — critically Halozyme (ENHANZE recombinant hyaluronidase), which underpins the subcutaneous conversions that are central to the moat defence (Phesgo, Ocrevus Zunovo, SC Tecentriq). Halozyme is a genuine single-source chokepoint for the SC-conversion strategy.
• Manufacturing: in-house network plus CDMOs. Roche has pledged tens of billions in US manufacturing/R&D capacity over five years (part of the ~$150B industry pledge struck with the Trump administration) — North Carolina among the sites alongside Amgen/Biogen/J&J.
• Diagnostics supply chain: Roche makes its own instruments and reagents; the placed-instrument install base is the lock-in.
• Distribution: Genentech (wholly-owned US arm) commercialises in the US — the most important single node, since the US is the majority of pharma revenue and the locus of the MFN/tariff risk (Lens 10).
• Downstream buyers: US payers/PBMs, ex-US national health systems, hospital labs (Dx).

Chokepoints: (1) Halozyme single-source for SC delivery; (2) Genentech/US commercial concentration overlapping the policy risk; (3) biologics capacity is capital- and time-intensive — the moat, but also the reason a tariff/MFN shock to US economics directly threatens the announced capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs..

Competitive Advantages (moats)

Durable moats — strong but franchise-specific, not monolithic:

1. Process/scale moat in biologics + the diagnostics tie-in. Roche is the largest biologics maker with the only paired top-tier Dx arm (~20% global IVD share, #1 — Lens 7) — companion-diagnostic integration is a structural advantage rivals can't easily replicate.
2. Switching-cost / formulation moat via subcutaneous conversion. The Phesgo (+48%) and Ocrevus Zunovo playbook converts patients to a more convenient, IP-fresh SC formulation before the IV biosimilar lands — Roche reports >50% of US Ocrevus patients already converted to Zunovo, targeting >60% before biosimilars. This is the moat in action, and it is the crux of the bull case.
3. Family-controlled long-termism (governance moat). The Hoffmann/Oeri pool controls ~75% of voting rights on ~8–10% of equity — Roche can run a patient, R&D-heavy, 39-year-dividend strategy without activist pressure. Double-edged (Lens 9/13).

Bargaining power. Strong over distributors and (historically) US payers — but structurally weakening on the payer axis as MFN pricing and tariffs shift power to the US government (Lens 10), and франchise-by-franchise erosion as biosimilars/next-gen rivals attack (Kesimpta vs Ocrevus growing 49% vs 6–9%; Mim8/Hympavzi vs Hemlibra — Lens 13). The moat is real on the installed franchises and the Dx razor-blade, thinner on pricing power than a year ago.

Segments

By division (2025, CHF):

Pharma is 77% of sales and effectively all the growth; Dx (+2% CER) is being held back by China healthcare-pricing reform (VBP), offset by pathology/molecular demand.

By geography (Pharma division, 2025, CHF):

Read: the US is the majority of pharma and a key growth contributor (+8%) — which is exactly why the US-policy overhang (MFN/tariffs) is the dominant macro risk for this name; International (+14%) is the fastest-growing region. The CER-vs-CHF gap (+7% vs +2% group) quantifies the franc headwind — a Swiss-cost / dollar-revenue mismatch that compresses reported earnings every strong-franc year.

---

Phase B — Measure performance

Earnings Result (FY2025, reported 29 Jan 2026)

The latest print is the full-year 2025 result:

• Group sales CHF 61.52B, +7% CER (+2% CHF) — pharma +9% CER drove it; Dx +2% CER.
• Core operating profit CHF 21.83B, +13% CER (+5% CHF) → core operating margin ≈ 35.5%. Margin expanded — profit grew nearly double sales on efficiency gains/operating leverage.
• Core EPS CHF 19.46, +11% CER (+4% CHF) — outgrew sales, confirming leverage.
• IFRS net income CHF 13.80B, +58% CER (+50% CHF) — the optical jump reflects a low 2024 base (prior-year impairments/one-offs); the core figures are the cleaner read.
• Cash/balance sheet: operating free cash flow CHF 16.2B (−12% CER); free cash flow CHF 11.8B; net debt down CHF 1.2B to CHF 16.2B — FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. + FX more than covered the CHF 8.4B dividend and CHF 3.1B of acquisitions. Healthy, lowly-levered balance sheet (net debt ≈ 0.7× core operating profit ).
• Dividend: proposed CHF 9.80/share — 39th consecutive annual increase.
• Guidance issued: 2026 group sales mid-single-digit % CER; core EPS high-single-digit % CER — i.e. continued EPS leverage over sales.
• Tone: outlook had been raised twice through 2025 (Apr +6% Q1, Oct 9M +7% with raised FY) — momentum, not deceleration.

Flag vs own history: the operating-FCF decline (−12% CER) against rising core profit is the one number to watch — driven by working-capital/US-inventory build and acquisition-related outflows; not alarming at this scale but worth tracking next print.

Earnings Calls (sentiment trend)

No transcripts on the shelf (web-only). From release cadence and management commentary across 2025:

• Recurring, escalating themes: "LOE endurance" (managing the Ocrevus/Perjeta/Actemra cliffs without a revenue air-pocket — they explicitly stress the new book offsets the old); discipline on M&A (CEO Schinecker's "three reasons for sitting out the deal spree" — debt-wariness, prices, strategic fit); obesity/CVRM as the new growth vector (post-Carmot/89bio); neuroscience (trontinemab).
• What they started saying (2025): obesity ambition ("top-three by 2030"), an explicit US-investment-at-risk warning tied to the MFN executive order.
• What they stopped leaning on: the legacy oncology triad (Avastin/Herceptin/Rituxan) as growth — now framed purely as managed decline.
• Net tone: confident on the core, openly cautious on US policy, disciplined on capital. Consistent, not promotional.

Comps

Peer table — Roche vs global large-cap pharma. Multiples are, June 2026, from aggregators (GuruFocus/StockAnalysis/Yahoo/companiesmarketcap); ROE. Where I could not source a clean figure: n/a.

Read: Roche is the largest of the EU big-three by market cap and the cheapest on both fwd P/E (16.5x) and EV/EBITDA (13.2x), while paying the joint-highest dividend (3.08%) and earning a ~45% ROE. It trades at roughly half Lilly's multiple — the market is paying a huge premium for GLP-1 growth (Lilly) and penalising Roche's patent-cliff optics. The valuation gap is the thesis: cheap on proven cash flows, optionality on obesity/Alzheimer's not in the price. Merck (~10x) is the only cheaper large-cap, on its own Keytruda-cliff overhang.

Stock-Price Catalysts (what moves ROG)

What the tape has reacted to (mostly ):

• Pipeline binaries dominate. The biggest single up-move catalysts in 2024–26 were clinical readouts, not earnings: CT-388 obesity Phase 2 (22.5% placebo-adjusted weight loss at 48 wks, Jan 2026) and trontinemab Alzheimer's amyloid-clearance data (~91% below positivity threshold, CTAD 2025). For a name re-rating on "can it find post-Ocrevus growth," obesity + Alzheimer's data points are the swing factor.
• LOE/biosimilar news (Ocrevus high-dose failure; Perjeta patent expiry) is the down-side catalyst set.
• US drug-pricing policy — the MFN executive order and the eventual Genentech/White-House deal — moved sentiment sector-wide (Lens 10).
• FX (franc strength) mechanically pressures reported CHF results each quarter (the +7% CER / +2% CHF wedge).
• Quarterly earnings matter less than for US peers (no consensus-beat ritual at the same intensity), but guidance raises (twice in 2025) supported the stock.

Pattern: ROG is a pipeline-and-policy stock with a dividend-aristocrat floor — it reacts to de-risking events on obesity/neuro and to US pricing headlines far more than to a points-beat on revenue. Current price 335.30 CHF (26 Jun 2026); consensus 12-mo target 358.83 CHF (+7%), high 428 / low 230, rating Buy (9 buy / 2 sell).

---

Phase C — Judge people & books

Management

• CEO — Thomas Schinecker (CEO since March 2023; long Roche insider, ex-head of Roche Diagnostics). Archetype: professional manager / scientist-operator, not founder. His stated frame is science-led discipline: four-TA focus, "very disciplined" BD, explicit wariness of debt.
• Track record / capital allocation (his watch). Selective, US-anchored bolt-ons rather than mega-M&A: Carmot ~$3.1B (2024 — the CT-388 obesity asset), Poseida ~$1B (2024 — allogeneic cell therapy), 89bio ~$3.5B (Oct 2025 — pegozafermin, FGF21 for MASH). This is a deliberate pivot into CVRM/obesity + cell therapy funded from FCF without levering up — coherent and on-strategy. ~45% ROE and a 39-year dividend-increase streak are the quantified scorecard of capital discipline.
• Governance / skin in the game. The Hoffmann/Oeri family pool controls ~75% of votes via the 1948 pooling agreement on ~8–10% of equity; vice-chairman André Hoffmann has stated the heirs intend to keep control. Powerful alignment toward long-term, R&D-heavy strategy — but a control structure that insulates management from external accountability (Lens 13).
• Red flags (governance, not accounting): the dual-class voting concentration; the franc/dollar structural mismatch the company chooses to live with; reliance on serial bolt-ons to refill the pipeline rather than internal discovery alone.

Forensic Red Flags

Accounting / disclosure (web-only — no filings on the shelf to forensically tie out; flag accordingly):

• Core vs IFRS gap. Roche's headline is "core" (non-IFRS), which strips amortisation of intangibles, impairments, and certain legal/restructuring items. 2025 IFRS net income +58% vs core EPS +11% shows how large the adjustments are in either direction — read core for the run-rate but watch the IFRS reconciliation. Not a red flag per se (standard for big pharma) but the spread is wide.
• FX as a recurring "below-the-line" swing — the CER/CHF wedge is disclosed and legitimate, but it means CHF-reported trends understate the operating business; don't anchor on CHF growth alone.
• Acquisition-heavy goodwill/intangible base from serial bolt-ons (Carmot/89bio/Poseida) — future impairment risk if any lead asset (CT-388, pegozafermin) fails. Watch the intangibles line on the next Finance Report.
• Operating-FCF down 12% CER while core profit rose — working-capital/inventory and deal outflows; benign at this scale but the divergence is the kind of thing to verify against the cash-flow statement once filings are on the shelf.

Regulatory findings:

• SEC (EDGAR LR/AAER): none possible — Roche has no CIK and does not file with the SEC. total_sec_findings: 0 is a structural N/A, not a clean bill.
• Non-SEC enforcement (web):
• France (antitrust): the French Competition Authority fined Roche €60M ($71M) (and Novartis ~€385M) over the Lucentis/Avastin affair — allegedly steering payers to costly Lucentis over off-label Avastin for AMD.
• US FCA (Genentech): Tarceva — $67M paid by Genentech + OSI to resolve False Claims Act allegations of misleading effectiveness claims. A separate Avastin qui tam (Petratos, alleging concealed risks) was dismissed by the Third Circuit (2017) on the heightened materiality standard.
• Item 3 / Legal Proceedings: no 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. on the shelf (foreign issuer) — would need the Roche Annual Report's litigation note to tie out fully; flag as not independently verified from primary filing.
• Summary: historical, resolved, mid-sized legal/antitrust matters typical of big pharma; no live going-concern or fraud signal surfaced. Verified via SEC EDGAR EFTS (N/A — no CIK), web search, as of 2026-06-29. The forward regulatory risk that matters is policy, not enforcement — US MFN pricing + pharma tariffs (Lens 12/13), where Genentech struck a 3-year tariff grace in exchange for US price cuts + further US investment.

---

Phase D — Project & stress-test

bolt-on · Pipeline by phase (big-cap pharma overlay)

The pipeline is the post-Ocrevus growth story. Key late-stage / catalyst assets:

The pipeline thesis: fenebrutinib (oral BTK) + Zunovo defend the MS franchise into the 2028/29 biosimilar window; CT-388 + pegozafermin are the call option on the CVRM/obesity TAM management wants top-three of by 2030; trontinemab is the Alzheimer's lottery ticket (brain-shuttle is genuinely differentiated on amyloid clearance, but clinical-benefit readout is years out and the category is treacherous).

bolt-on · Catalyst calendar (mechanism comps)

• 2026 H1: fenebrutinib FENhance 1 RMS readout → MS filing; CT-388 Phase 3 fully enrolling.
• 2026–27: trontinemab interim Alzheimer's data points; pegozafermin MASH progression.
• 2028: trontinemab TRONTIER completion; Ocrevus EU biosimilar window opens.
• 2029: Ocrevus US biosimilar window opens (the big one — CHF 7B drug).
• Mechanism comps: CT-388 vs Lilly Zepbound / Novo Wegovy-CagriSema (Roche is years behind the incretin leaders — the comp is brutal); fenebrutinib vs Sanofi tolebrutinib / Novartis remibrutinib (BTK class) and Novartis Kesimpta (anti-CD20 SC, growing ~49%); Hemlibra vs Novo Mim8/denecimig + Pfizer Hympavzi + Sanofi Altuviiio/fitusiran; Vabysmo vs Regeneron Eylea/Eylea-HD + biosimilars.

Forward Projection (EPS, 3 fiscal years)

Built bottom-up from FY2025 actuals + 2026 guidance. All; arithmetic shown. Base = FY2025 core EPS CHF 19.46; guidance = 2026 core EPS high-single-digit % CER.

Read: the base path is a high-single-digit EPS compounder through 2027 that decelerates into the 2028/29 Ocrevus cliff, with the spread between bull and bear (≈ CHF 20.5–25.8 by 2028) driven almost entirely by (a) whether the SC/oral bridge holds the MS franchise and (b) whether obesity/Alzheimer's de-risk. On 16.5x, the base case alone roughly supports the current price with the dividend as carry; the bull case is the re-rate, the bear case is the value-trap. (No our model create in watchlist mode — base call logged here only: ROG.SW FY2026 core EPS ≥ CHF 20.5, p≈0.60.)

Bull vs Bear

Bull case. Roche is a ~45%-ROE, ~36%-core-margin cash machine trading at half Lilly's multiple because the market only sees the Ocrevus cliff and not the three bridges over it. (1) The SC-conversion moat works — Phesgo +48% and >50% Ocrevus→Zunovo conversion prove Roche can defend franchises pre-biosimilar; (2) fenebrutinib gives the first oral BTK in MS, extending the CHF 7B franchise with fresh IP; (3) genuine optionality stacked on top — CT-388's 22.5% weight loss is a real obesity asset (a top-three-by-2030 ambition the price ignores), trontinemab is the most differentiated brain-shuttle amyloid programme in the clinic; (4) 39-year dividend record + lowly-levered balance sheet pay you ~3% to wait. Contrarian view the market refuses to see: Roche has already survived a bigger patent cliff (the 2019–23 biosimilar wave) and re-based — the 2028/29 cliff is smaller and better-bridged than 2019 was, yet it's priced as if there's no plan.

Bear case. (1) Ocrevus is ~11% of group sales and faces EU 2028 / US 2029 biosimilars while already losing share to Novartis Kesimpta (growing 49% vs 6–9%) — if Zunovo conversion stalls, a CHF 7B hole opens with no equal-sized replacement; (2) Roche is years late and sub-scale in obesity — CT-388 is a credible #3-at-best behind Lilly/Novo's entrenched incretin machines, and the category's bar keeps rising; (3) US policy is a structural margin threat — 53% of pharma is US, and MFN pricing + pharma tariffs directly attack the highest-margin revenue (the 3-year grace only defers it). Pre-mortem (18 months out, thesis broke): Zunovo conversion plateaued below target, a key Phase 3 (fenebrutinib safety/death-imbalance or CT-388) disappointed, and MFN pricing cut US net prices — EPS goes flat, the "cheap" multiple was cheap for a reason, and the value-trap closes. On multiples: 16.5x is not demanding, but it's only cheap if the bridge holds; against a genuine no-growth LOE outcome it's fair, not cheap.

Devil's Advocate (short-seller)

Dismantling the bull case:

• Revenue concentration that can break: Ocrevus (~CHF 7B, ~11% of group) is the keystone — its 2028/29 biosimilar cliff is the single point of failure, and it's already decelerating (+6–9% while Kesimpta does +49%). The SC-conversion "moat" is a time-limited delaying tactic, not durable protection — once IV biosimilars and SC rivals (Kesimpta) compound, the franchise erodes regardless of Zunovo.
• The obesity story is mostly hopium. CT-388 is one Phase-2 datapoint into a market Lilly and Novo own; "top-three by 2030" against two players with multi-year head starts, scaled manufacturing, and oral programmes is an aspiration, not a plan. Bulls are paying for optionality that may never clear the incumbents.
• The most dangerous competitor bulls underestimate: Novartis (Kesimpta in MS — already taking Ocrevus share) and Novo (Mim8/denecimig in haemophilia, plus the obesity moat) — both attack Roche's two biggest growth drivers simultaneously.
• Worst capital-allocation read: serial US bolt-ons (Carmot/89bio/Poseida ≈ $7.6B) loading goodwill/intangibles that impair if any lead asset fails — buying a pipeline because internal discovery hasn't replaced the cliff fast enough.
• What must hold for today's price: SC conversion >60%, fenebrutinib filed clean (the death imbalance is a real overhang), no MFN net-price shock, and obesity stays a free option. If growth disappoints 20–30% (Ocrevus cliff un-bridged + obesity miss), the EPS path flattens and a 16.5x "cheap" multiple compresses toward Merck's ~10x — meaningful downside, not just lost upside.
• The single permanent-impairment scenario: a fenebrutinib safety failure (death-imbalance escalates) kills the MS bridge exactly as Ocrevus biosimilars land — that's the structural break. Plausibility: low-to-moderate, but it's the tail that matters.

Management Questions (ordered by information value)

1. What is the latest US Ocrevus→Zunovo conversion rate, and what conversion % must you hold to keep MS-franchise revenue flat through the 2029 US biosimilar entry?
2. On the fenebrutinib death imbalance — what is the adjudicated cause, the placebo-adjusted mortality, and does it threaten the label/filing?
3. Quantify the net US price impact of the Genentech MFN/tariff agreement — what does the 3-year grace cost you annually, and what happens in year 4?
4. For CT-388, realistically — best-in-class, or a credible #3? What share of the obesity TAM is in your 2030 plan, and is the manufacturing built for it?
5. What is the expected revenue trough/air-pocket (if any) around 2028–29 as Ocrevus EU/US LOEs hit, and what specifically fills it?
6. Trontinemab — when do you expect a clinical-benefit (not just amyloid) readout, and what's your internal PoS given the Alzheimer's category's history?
7. How do you intend to defend reported (CHF) earnings against franc strength — more USD-cost matching, or do shareholders just absorb the FX wedge?
8. The Diagnostics division is growing +2% CER dragged by China VBP — what is the path back to mid-single-digit, and is the pharma/Dx integration thesis actually generating measurable revenue synergy?
9. With ~$150B industry US-investment pledges, what is Roche's committed US capex, its expected ROIC, and is it contingent on the tariff grace holding?
10. Your M&A has been serial US bolt-ons — at what point does the goodwill/intangible base concern you, and what is your impairment-test discipline on Carmot/89bio assets?
11. How sustainable is the ~45% ROE as the asset base grows via acquisition and the legacy high-margin book rolls off?
12. Hemlibra faces Novo Mim8/denecimig and Pfizer Hympavzi — what's your defence, and is the SC frequency/efficacy edge enough?
13. What is the next leg of the dividend policy — can the 39-year streak survive a simultaneous Ocrevus cliff and MFN pricing year?
14. Where would you deploy a large transformational deal if you broke your debt-discipline, and what valuation/strategic bar would justify it?
15. If you had to name the one assumption in the current plan most likely to be wrong, what is it?

---