Roche Holding AG (RO.SW) - Deep Dive Research Report

Prepared 25 May 2026. All figures from Roche IR disclosures, company filings, and named secondary sources. No valuation, no recommendation.

Reporting cadence used in this report: Roche reports half-yearly (FY in late January, H1 in late July) with two intervening sales-only updates (Q1 in late April, Q3 in late October). The four most recent disclosures are:

1. Q1 2026 sales update - 23 April 2026
2. FY 2025 results - 29 January 2026
3. Q3 2025 sales update - 23 October 2025
4. H1 2025 results - 24 July 2025

1. What the Company Does

Roche makes prescription drugs and the machines that diagnose disease. Those are the two halves of the business, and the company has run them under one roof for more than a century. A patient with multiple sclerosis sits in an infusion chair receiving Ocrevus; a hospital lab two floors down runs an Elecsys assay on a Roche cobas analyser to decide whether another patient has had a heart attack. Both revenue streams are Roche.

The company was founded in 1896 in Basel, Switzerland, by Fritz Hoffmann-La Roche, who attached his wife's surname to his own (it was a different era) and built one of the first companies in the world to manufacture branded medicines at industrial scale. Roche stayed family-influenced and Basel-headquartered through two world wars, a pivot into vitamins that dominated mid-century earnings, then the Valium franchise that funded the 1980s, and ultimately into the biologics era. The most consequential decision of the modern company was the 1990 minority stake in Genentech, the South San Francisco biotech that had invented recombinant DNA-based pharmaceuticals. Roche bought the rest in 2009 for roughly $47 billion. That deal turned Roche from a chemistry company into the world's largest biotech and gave it the monoclonal-antibody franchise (Herceptin, Avastin, Rituxan) that paid the company's bills for fifteen years.

The current value proposition is unusually concrete: Roche translates biological insight into drugs that work for diseases where existing options were either crude or absent. Ocrevus turned multiple sclerosis from a steady slide into something many patients can hold stable on a six-monthly infusion. Hemlibra replaced near-daily clotting-factor injections for hemophilia A with a subcutaneous shot every one to four weeks. Vabysmo gave wet-AMD patients a longer interval between needles in the eye. The technical moat is the ability to discover these molecules (the Genentech side does most of the early science), run the multi-year, multi-thousand-patient trials that prove they work, and manufacture biologics at hospital-grade purity in fermentation tanks that took a decade to build and qualify.

The Diagnostics half of the company sells the analysers, reagents, and software used in hospital and reference labs to test blood, tissue and DNA. The two divisions cross-fertilise: when Roche develops a targeted cancer drug, its diagnostics arm develops the companion test that identifies which patients have the mutation the drug treats. This is the "Personalised Healthcare" thesis that Roche has been pushing for two decades, and is unique among the big pharmaceutical companies.

The company is still effectively family-controlled. The Hoffmann and Oeri family pool held 64.97% of voting shares as of 31 December 2025 (Roche Wikipedia), a structure that has survived since 1948 and that buys Roche the freedom to run R&D on a 10-15 year horizon when the rest of pharma is on a quarter-to-quarter clock.

2. Business Segments

Roche reports two divisions. They are very different businesses sharing a parent.

2.1 Pharmaceuticals

What it does: discovers, develops, manufactures and sells prescription drugs in oncology, hematology, immunology, neurology, ophthalmology, infectious disease, and (newly) cardiovascular/metabolic. In FY 2025, Pharma generated CHF 47.7 billion in sales, +9% at constant exchange rates (CER), making it roughly 78% of group sales. The top five growth drivers (Ocrevus, Hemlibra, Vabysmo, Tecentriq, Xolair, Phesgo) generated CHF 21.4 billion combined - close to half the divisional total (Roche FY 2025 results, 29 January 2026).

Core capability: Roche's pharma engine is a hybrid. The South San Francisco Genentech site does most of the early biology; pRED in Basel does small molecules and platform science; and the global development operation runs the registrational trials. Then a manufacturing network of large-scale bioreactors (Vacaville CA, Penzberg Germany, Mannheim, Singapore) makes the biologics at purity grades that took 10-15 years of process knowledge to qualify with FDA and EMA. None of these are things a new entrant builds in a five-year plan.

Why it exists as a separate division: it operates under entirely different regulation (FDA/EMA, branded prescription) and economics (long R&D, high gross margins, patent cliffs) than the Diagnostics business.

Competitive position: Roche is in the top tier of global pharma by R&D output. In oncology specifically, the company is being outflanked: Keytruda (Merck) has become the dominant single-agent immunotherapy and Roche's own Tecentriq is the smaller franchise. In multiple sclerosis Ocrevus is the category leader; in hemophilia A Hemlibra has redefined the standard of care; in wet AMD Vabysmo is taking share from Eylea (Regeneron). In obesity, where Lilly and Novo Nordisk have an enormous head start, Roche is openly trying to become a credible third player (Pharma Day, September 2025).

How it fits in the group: this is the growth engine and the cash engine simultaneously. Pharma funds the dividend, the pipeline, and the diagnostics business.

2.2 Diagnostics

What it does: manufactures and sells in-vitro diagnostic systems - the analysers, reagent kits, sample-handling robotics and lab software used to test biological samples for everything from heart attacks to HPV to genetic mutations. FY 2025 sales: CHF 13.8 billion, +2% CER (FY 2025 results). Roughly 22% of group sales.

Roche groups the Diagnostics business into sub-areas: Core Lab (high-volume immunoassays and clinical chemistry on cobas instruments), Molecular Lab (PCR and sequencing for infectious disease and oncology), Pathology Lab (tissue diagnostics, slide scanners, digital pathology), Diabetes Care (Accu-Chek glucose monitoring), and Point of Care (rapid tests at the bedside).

Core capability: installed base of cobas analysers in thousands of hospital labs. Once a lab has standardised on cobas, the reagents have to come from Roche. The classic razor-and-blade structure, except the razor costs hundreds of thousands of dollars and takes years to displace.

Why it exists as a separate division: very different cycle (replacement of capital equipment plus consumable reagents), very different regulatory regime (CE-IVD, FDA 510(k)), different customer (hospital purchasing committees, not prescribers), different margin structure.

Competitive position: Roche is the global leader in in-vitro diagnostics by revenue. Direct competitors include Abbott (especially in clinical chemistry), Siemens Healthineers (high-volume immunoassay), Danaher's Beckman Coulter and Cepheid (molecular), and Thermo Fisher (specialty). In Europe Roche and Abbott split most of the market between them. Roche's edge is the breadth of one menu (chemistry, immunoassay, molecular all on connected platforms) and the digital pathology + companion-diagnostics tie-in with the Pharma side - the Roche pathology business is now expanding via the announced acquisition of PathAI (BioPharma Dive, 2026).

How it fits in the group: lower-growth, lower-margin, but high-recurring-revenue ballast. It is the strategic option (companion diagnostics for Pharma) more than the growth bet.

Segment summary

3. Products and Business Detail

Pharmaceuticals catalogue

Top growth drivers (FY 2025 sales):

• Ocrevus (ocrelizumab) - CHF 7.0bn, +9% CER. Anti-CD20 monoclonal antibody for relapsing and primary-progressive multiple sclerosis. Originally a six-monthly IV infusion; the subcutaneous formulation Ocrevus Zunovo (10-minute injection) launched in late 2024 and reached 24,000 global patients by Q1 2026 (Q1 2026 transcript). About half of subcutaneous patients are new starts, expanding the franchise rather than just cannibalising the IV business.
• Hemlibra (emicizumab) - CHF 4.8bn, +11%. Bispecific antibody mimicking factor VIII for hemophilia A. Replaced multi-times-weekly clotting factor with subcutaneous dosing every 1-4 weeks. Has effectively redefined the standard of care.
• Vabysmo (faricimab) - CHF 4.1bn, +12%. Bispecific (anti-VEGF + anti-Ang2) for wet AMD, DME and retinal vein occlusion. Longer dosing interval than Eylea was its initial selling point; now in a head-to-head share fight with high-dose Eylea HD.
• Tecentriq (atezolizumab) - CHF 3.6bn, +3%. PD-L1 immunotherapy. Distant second-place to Keytruda in the IO race but still a meaningful franchise.
• Xolair (omalizumab) - CHF 3.1bn, +32%. Anti-IgE for chronic urticaria, asthma, and (the new growth leg) food allergy. The food-allergy approval in early 2024 has been the main growth driver - it created a new patient population overnight.
• Phesgo (subcutaneous pertuzumab + trastuzumab) - CHF 2.4bn, +48%. The fixed-dose subcutaneous co-formulation of two HER2-positive breast-cancer drugs Roche has sold for years. Replaces hours-long IV infusions with a 5-8 minute injection. Globally 55% of eligible patients had converted by Q1 2026 (Q1 2026 transcript).

Legacy biologics under biosimilar pressure: Herceptin (-22% YoY in 2025), Avastin (-17%), MabThera/Rituxan (-4%), Lucentis, Actemra, Esbriet. Combined LOE drag of CHF 1.0bn in 2025, expected ~CHF 1.0bn again in 2026 (FY 2025 transcript via Insider Monkey).

Phase III / late-stage pipeline assets discussed on recent calls:

• Giredestrant - oral SERD for ER+ breast cancer; positive Phase III in adjuvant setting (lidERA-Breast Cancer trial showed 30% reduction in invasive disease recurrence/death risk per FY 2025 release); filing imminent.
• Fenebrutinib - oral BTK inhibitor; first in class with positive Phase III in both relapsing and primary-progressive MS (FENhance, FENtrepid trials).
• Trontinemab - brain-shuttle anti-amyloid for Alzheimer's, designed to cross the blood-brain barrier and reduce the ARIA edema/microhemorrhage side effects that limited Leqembi and Kisunla.
• CT-388 - dual GLP-1/GIP receptor agonist for obesity. Phase II showed 22.5% placebo-adjusted weight loss at 48 weeks. Moved into Phase III in H1 2026.
• Petrelintide - long-acting amylin analog, in-licensed from Zealand Pharma in March 2025 in a deal worth up to $5.3bn. Phase II showed ~10.7% weight reduction with a "placebo-like" tolerability profile, intended for combination with CT-388.
• Pegozafermin - FGF21 analog for metabolic dysfunction-associated steatohepatitis (MASH); acquired with 89bio.
• Zilebesiran - siRNA for hypertension (partnered with Alnylam).
• NXT007 - next-generation hemophilia A bispecific, intended to extend the Hemlibra franchise beyond its 2030s patent expiry.

Management has said 10 molecules moved into Phase III in 2025 alone - "a record for us" - and there are 19 potential new launches by 2030 (FY 2025 transcript).

Diagnostics catalogue

The diagnostics catalogue is organised around hardware platforms:

• cobas family - high-throughput clinical chemistry, immunoassay (including Elecsys reagents for cardiac markers, infectious disease, oncology, neurology biomarkers like pTau217 for Alzheimer's), molecular PCR.
• Ventana tissue diagnostics - automated immunohistochemistry slide stainers used in pathology to read out HER2, PD-L1, and other companion-diagnostic markers.
• Accu-Chek - blood glucose monitoring for diabetes (a legacy consumer-medical business).
• Sequencing - the Exelios sequencer (a new in-house high-throughput platform launching summer 2026 per Q1 2026 call) and prior collaborations with Illumina/PacBio.
• Digital pathology - the navify ecosystem; expanding via the announced PathAI acquisition.

Manufacturing footprint

Roche manufactures the biologics in its own large-scale fermentation facilities. The largest sites are Vacaville and Oceanside in California, Penzberg in Germany, Mannheim in Germany (also a major diagnostics site), Basel/Kaiseraugst in Switzerland, and Singapore. The US biologics network is being expanded via the $50bn US capex commitment announced in April 2025, including a new North Carolina plant. Diagnostics manufacturing is concentrated in Mannheim, Penzberg, Tucson (Ventana), and Indianapolis (Diabetes Care).

Geographies

In 2025 the regional pharma mix was: US +8% growth, Europe +5%, Japan +5%, International +14% (with China specifically +10%). The US remains the single largest market and the most price-exposed. China is the swing factor - Pharma growing nicely as Phesgo and other drugs enter reimbursement, but Diagnostics down ~27% YoY in 2025 because of the country's healthcare pricing reforms (volume-based procurement, anti-corruption campaign). Management expects the China diagnostics headwinds "to become a lot less in 2026 and gone in 2027" (FY 2025 transcript).

Currency mechanics

A pivotal piece of business detail: Roche reports in Swiss francs but earns most of its revenue in dollars, euros and yen. When the franc strengthens, reported sales fall in CHF even when underlying volumes are growing. Q1 2026 showed this starkly: +6% CER growth converted into -5% CHF reported growth, with the dollar move alone accounting for most of the gap (Q1 2026 press release). This is structural - Roche cannot move its headquarters - and is the reason management always communicates growth in CER first.

4. Customers

Roche has two very different customer bases for its two divisions.

Pharma customers. The proximate customer is the prescribing physician - an oncologist, neurologist, hematologist, retinal specialist. The economic buyer is whoever pays for the drug: in the US that is the pharmacy benefit manager (CVS Caremark, Express Scripts, OptumRx) and the commercial insurer; in Europe it is the national health technology assessment body (NICE in England, G-BA in Germany, HAS in France) negotiating with the country's payer; in Japan it is MHLW; in China it is the National Healthcare Security Administration that runs the National Reimbursement Drug List. Roche has to win the clinical argument with the doctor and the value argument with the payer separately. Sales cycles are short for a new individual prescription but multi-year for getting on a national formulary.

Why customers choose Roche: in each major franchise there is a specific clinical edge. Ocrevus stops disability progression in primary-progressive MS where no other drug had been shown to work. Hemlibra eliminates the burden of factor VIII infusions for hemophilia A patients. Vabysmo offers a longer dosing interval for wet-AMD patients tired of monthly intravitreal injections. Phesgo is faster to administer than the IV equivalent. The pattern is consistent: a clinically meaningful improvement on a specific dimension that doctors care about.

Switching costs: very high once a patient is stable on a chronic biologic. Switching MS therapies risks a relapse; switching hemophilia therapies risks bleeds. This installed-patient effect is the underappreciated source of Roche's franchise durability - patents may expire but established patients often stay on the originator if reimbursement permits.

Concentration: low at the patient level - Roche treats 39 million patients globally - but moderately concentrated at the payer level. A formulary decision by a single US PBM can move millions of dollars of weekly revenue. Loss of National Reimbursement Drug List status in China is similarly material.

Contract structure: pharma revenue is essentially per-prescription. There are no long-term take-or-pay contracts. Predictability comes from the underlying disease prevalence and the slow turnover of patient populations on chronic therapies.

Diagnostics customers. Hospital and reference labs (Quest, LabCorp, Sonic, large hospital systems). The buying decision is committee-driven: lab directors, chief medical officers, hospital purchasing. Sales cycles for a new cobas analyser placement are 6-18 months and involve menu evaluation, throughput modelling, and economic analysis. Once placed, the lab has signed up for years of reagent purchases - typically captured under a 5-7 year reagent rental agreement where the analyser is effectively free in exchange for committed reagent volume. This is the source of Diagnostics' recurring revenue.

Switching costs in Diagnostics: very high. Replacing a Roche cobas line requires validating every assay against the new platform - months of paired-sample work, regulatory documentation, staff retraining. Labs don't do this lightly.

Concentration in Diagnostics: more concentrated than Pharma. A handful of mega-labs (Quest, LabCorp, the AmeriPath networks) and large hospital group purchasing organisations can move multi-year revenue trajectories with single decisions.

5. Competitive Landscape

Pharma and Diagnostics need separate treatment.

Pharma competition

Big-Pharma peers competing across multiple Roche therapeutic areas:

Why Roche wins where it wins: a single-product clinical advantage backed by global commercial reach and the ability to outspend on Phase III. Ocrevus succeeded because the primary-progressive MS data was unprecedented. Hemlibra won because the bispecific approach was both more convenient and more effective than recombinant factor VIII. Vabysmo won because the longer dosing interval mattered to patients.

Why Roche loses where it loses: it has been late to the platform franchises that are eating pharma. It does not have a CAR-T (BMS and J&J/Legend do). It does not have a category-leading ADC (Daiichi/AZ have Enhertu, Pfizer has Padcev). It is a distant second in PD-(L)1 immunotherapy. And it is a late entrant to obesity, where Lilly and Novo Nordisk are running orders-of-magnitude ahead. Management's strategy is to build its way back via the 19-NME pipeline through 2030; whether that arrives in time is the central pharma question for the business.

Barriers to entry: high. Discovering a biologic, taking it through Phase I-III, building manufacturing, and gaining ex-US reimbursement takes 10-15 years and billions of dollars per asset. The barrier is not regulatory in the abstract - it is the cumulative capital required to put 130+ molecules through clinical development simultaneously, which only ~10 companies in the world can afford to do.

Diagnostics competition

Roche's diagnostics edge is breadth of menu on a single platform (chemistry, immunoassay, molecular all on connected cobas systems), the companion-diagnostics tie-in with the Pharma side (Ventana stainers read out PD-L1 for Tecentriq and others), and the world's largest installed base in Europe. The exposure is in the US, where Abbott is the entrenched leader, and in China, where local players are increasingly favoured by national procurement.

Industry consolidation in diagnostics has been a multi-year theme (Danaher's purchases of Beckman, Cepheid, IDT; Thermo's serial M&A). Roche has stayed organic-first in Diagnostics until the PathAI announcement.

6. Industry

Global pharmaceutical demand drivers: aging populations in developed markets, the steady expansion of insurance coverage in middle-income markets (China NRDL, India's PMJAY), the chronic-disease epidemiology of urbanisation (obesity, diabetes, cardiovascular), and the scientific frontier moving into previously untreatable conditions (Alzheimer's, primary-progressive MS, certain solid tumours). The global pharmaceutical market is roughly $1.6 trillion in 2025 and growing mid-single-digits, with biologics growing roughly twice that rate.

Industry-specific tailwinds:

• Obesity / metabolic. The GLP-1 class has created a multi-tens-of-billions-per-year category from a near-standing start in 2020. The TAM remains underpenetrated globally - in the US alone there are ~100 million obese adults vs perhaps ~3-5 million on therapy.
• Alzheimer's. After 30 years of failed trials, anti-amyloid antibodies (Lecanemab, Donanemab) have demonstrated disease modification. Roche's trontinemab, with the brain-shuttle technology designed to reduce ARIA, is positioned for what management hopes will be a "third-generation" entry.
• Personalised oncology - genomics enables ever more precisely targeted therapies and the diagnostics that match them.

Industry-specific headwinds:

• US drug pricing reform. The Inflation Reduction Act's Medicare price negotiation, plus the Trump administration's Most-Favored-Nation initiative announced via Executive Order in May 2025, are reshaping the US pricing environment. Roche's Genentech was one of nine drugmakers that reached a December 2025 deal with the administration agreeing to lower prices on certain drugs in exchange for tariff relief (Fierce Pharma). The deal also commits Roche to selling at MFN prices direct to US patients via the new TrumpRx channel.
• Tariffs. The administration has separately threatened pharmaceutical tariffs to incentivise US manufacturing. Roche's response was the announced $50bn US capex commitment over five years (Roche press release, 22 April 2025) covering both Pharma and Diagnostics.
• Biosimilar erosion of legacy biologics. Roche has now absorbed the worst of the Herceptin/Avastin/Rituxan erosion (each is now sub-$1.3bn from earlier peaks of $7bn+), but the LOE drag continues at roughly CHF 1bn/year.
• China VBP and anti-corruption. Diagnostic volume-based procurement and Pharma price negotiations have compressed prices materially. Roche says this normalises by 2027.

Cyclicality: prescription drug demand is non-cyclical. Hospital capital equipment cycles (analysers) are mildly cyclical. Diagnostic reagent volume is non-cyclical.

Where Roche sits in the supply chain: Roche is the innovator originator - it does discovery, development, manufacturing, and global commercialisation. It does not outsource manufacturing of its key biologics. It is a buyer of CDMO services for some smaller-molecule API steps but is largely vertically integrated for the biologics that generate most of its revenue.

7. Growth Triggers

Sourced exclusively from the four most recent earnings calls and disclosures.

• CT-388 obesity Phase III enrolment in H1 2026. Following positive Phase II showing 22.5% placebo-adjusted weight loss, Roche initiated the Phase III program. (Repeated across H1 2025 call, Pharma Day September 2025, Q3 2025 call, FY 2025 call, Q1 2026 call.)

  "For the efficacy estimand, we achieved a placebo-adjusted weight loss of 22.5%." - Schinecker, FY 2025 call, January 2026

• Petrelintide / CT-388 combination initiation in 2026. Roche has stated repeatedly that the combination of the long-acting amylin (petrelintide, from Zealand) with the dual GLP-1/GIP (CT-388) is the obesity asset it expects to differentiate. (Pharma Day Sept 2025; reaffirmed Q3 2025, FY 2025 and Q1 2026 calls.)

• Giredestrant filing imminent for ER+ breast cancer. Following positive Phase III (lidERA-Breast Cancer) showing 30% reduction in disease recurrence in adjuvant setting. (FY 2025 call, 29 January 2026; Q1 2026 call.)

  "Giredestrant ... 70% of the opportunity is in the adjuvant setting." - Teresa Graham (Head of Pharma), Q1 2026 call, 23 April 2026

• Fenebrutinib filing in MS in coming months. First BTK inhibitor with positive Phase III data across both relapsing-remitting MS (FENhance) and primary-progressive MS (FENtrepid). (Q1 2026 call.)

  "First and only BTK inhibitor with positive Phase III across both RMS and PPMS." - Teresa Graham, Q1 2026 call, 23 April 2026

• Trontinemab Phase III readouts in Alzheimer's. Brain-shuttle anti-amyloid intended to reduce ARIA side effects. Phase III ongoing; not yet a launch but a key readout to watch. (H1 2025, Q3 2025, FY 2025 calls.)

• Ocrevus subcutaneous (Zunovo) continues ramp. 24,000 global patients on Zunovo by Q1 2026, vs 12,500 at Q3 2025 - roughly doubling in two quarters - with ~50% of switchers being naive to Ocrevus. (Q3 2025, Q1 2026 calls.)

• Xolair food allergy continues to scale. Q1 2026 +26% CER, with management calling out food allergy as the primary driver. (FY 2025, Q1 2026 calls.)

• Phesgo conversion past 55% globally with further room as more countries reimburse the subcutaneous formulation. (Q1 2026 call.)

• Exelios next-generation sequencer launching "this summer" - first major in-house NGS hardware push from Roche after years of partnerships. (Q1 2026 call, 23 April 2026.)

• Elecsys neurology assay portfolio expanding - new menu including pTau217 (Alzheimer's blood biomarker), NfL (CE marked for MS neuroinflammation per Q1 2026 release), and IGRA tuberculosis. (Q1 2026 call.)

• Pegozafermin MASH Phase III in progress following the 89bio acquisition; positioned as "best-in-disease" by management. (Q3 2025 call.)

• NXT007 - next-gen Hemlibra successor advanced into Phase III in 2025 - protects the hemophilia franchise into the late 2030s. (H1 2025 call.)

• US $50bn capex committed April 2025, including new North Carolina plant - both Pharma and Diagnostics capacity, mitigates US tariff risk. (Discussed across Q1 2026, FY 2025 calls.)

• PathAI acquisition - expands digital pathology into AI-driven workflows. (Announced 2026, discussed Q1 2026.)

• NVIDIA AI partnership / "largest AI factory in pharma". Schinecker repeatedly framed Roche as deploying AI infrastructure at scale to compress drug discovery cycles. (Q1 2026 call.)

  "Between 2027 and 2030, we have up to 19 NMEs that could launch." - Schinecker, Q1 2026 call, 23 April 2026

8. Key Risks

1. The obesity bet may be too late, too undifferentiated, or both. Lilly's tirzepatide and Novo's semaglutide have a multi-year head start, enormous installed-prescription bases, and ferocious manufacturing scale. Roche's CT-388 produced strong Phase II data (22.5% placebo-adjusted weight loss), but the read on petrelintide was tougher. BioPharma Dive characterised the petrelintide Phase II data as "undifferentiated" (BioPharma Dive coverage), and shares of both Roche and Zealand fell on the readout. The mechanism is straightforward: by the time Roche launches in 2027-28, Lilly and Novo will have multiple oral and combination products, retatrutide-class triple agonists, and entrenched formulary positions. Roche has acknowledged this is a "be a strong number three" play, not a category-leading attempt - a credible but not high-probability scenario.

2. US drug pricing - MFN, IRA negotiation, and the TrumpRx channel. The combination of IRA Medicare negotiation (which selects 10-20 drugs per year for price-setting) and the new MFN framework will compress US pricing on the legacy franchise. Roche/Genentech was among the nine drugmakers that struck the December 2025 MFN deal. The mechanism: MFN benchmarks US prices to the lowest of an OECD basket, which structurally takes 30-60% off list prices on negotiated drugs. The franchise most at risk in the near term is anything with a meaningful Medicare exposure that reaches a 9-year (small molecule) or 13-year (biologic) IRA selection threshold.

3. Tecentriq's secondary indications attrite. The March 2026 failure of SKYSCRAPER-12 (tiragolumab plus Tecentriq vs Tecentriq alone in PD-L1-high lung cancer) marked another setback for the IO franchise (Clinical Trials Arena coverage). Mechanism: every failed combination trial removes a future growth lever and concentrates the franchise on areas where Keytruda already dominates.

4. China diagnostics may not normalise on the timeline management implies. Management has guided China diagnostics headwinds to "become a lot less in 2026 and gone in 2027". If volume-based procurement continues to expand to more product categories, or if local Chinese diagnostics companies continue to gain share via national procurement preferences, the recovery could disappoint.

"These headwinds will become a lot less in 2026 and will be gone in 2027." - Schinecker, FY 2025 call, January 2026

5. Currency translation drag is structural and persistent. With ~50% of sales in US dollars and the company reporting in Swiss francs, the Q1 2026 split of +6% CER vs -5% CHF reported shows the magnitude. Mechanism: a strong franc steadily compresses CHF-reported earnings even when the business is operationally healthy. This is not solvable.

6. Vabysmo faces tougher competition from Eylea HD. Regeneron's high-dose Eylea has counter-attacked successfully in 2025. Vabysmo growth slowed in 2025 from earlier breakneck rates. Management acknowledged that "the branded market segment in the U.S. [is contracting] because of less funding available for co-assistance foundations" (Q3 2025 call).

7. Pipeline execution risk - 19 NMEs is a lot. Roche has staked its post-2027 growth on landing roughly 19 launches by 2030. Phase III drug development has historical pass rates around 50-60%. Even if Roche executes at industry-leading rates, several of those programs will fail or disappoint. The mechanism is binary: missed Phase III readouts wipe years of expected revenue.

8. Tariff risk on pharma imports into the US. The Trump administration has threatened pharmaceutical tariffs. Roche's $50bn US capex commitment is the mitigation, but execution risk is real and the build-out takes years. In the interim, tariff exposure on European-manufactured biologics shipped into the US is non-trivial.

9. Walk the Talk

The four concalls used for this section are:

1. H1 2025 results - 24 July 2025
2. Q3 2025 sales update - 23 October 2025
3. FY 2025 results - 29 January 2026
4. Q1 2026 sales update - 23 April 2026

Most recent is within 32 days of report date (well inside the 90-day window).

Management at Roche under Thomas Schinecker (CEO since March 2023) has been notably consistent on the macro outline and notably willing to revise specifics when better data arrives.

Starting with H1 2025 (24 July 2025). Schinecker guided full-year 2025 sales to "mid single digit" growth at CER and core EPS to "high single digit". The LOE guidance was revised down to CHF 1.0bn from CHF 1.2bn earlier in the year - the legacy biologics were eroding less than management had originally feared. Pharma had grown 10% in H1, well ahead of the full-year guide; this was a hint that the guidance was conservative. The pipeline narrative was about progressing four molecules into Phase III (NXT007, prasinezumab, trontinemab, zosurabalpin).

By Q3 2025 (23 October 2025), management did exactly what their H1 trajectory implied - they raised full-year guidance. Core EPS guide moved from "high single digit" to "high single digit to low double digit". LOE drag was revised down again to CHF 0.8bn from CHF 1.0bn. Year-to-date growth held at 7%. The bigger narrative shift was Pharma Day in September - the explicit positioning as a "top three" obesity player and the disclosure that 10 NMEs had moved into Phase III in 2025 (vs 4 implied at H1). Schinecker:

"Ten moving into Phase 3 ... These are all molecules that will launch by the end of this decade." - Q3 2025 call

By FY 2025 (29 January 2026), the delivered number was 7% group sales growth at CER, 11% core EPS growth. Both inside their guidance band. The Phase III count of 10 NMEs was confirmed. CT-388 Phase II had read out at 22.5% placebo-adjusted weight loss, vindicating the obesity bet. Schinecker reiterated the 2026 guide of mid-single-digit sales and high-single-digit core EPS. He also crystallised the long-game framing:

"We had 10 NMEs moving into Phase III. That's a record for us. ... These are the 19 medicines that we can launch by the end of the decade." - FY 2025 call

By Q1 2026 (23 April 2026), the company delivered +6% CER growth and +7% in Pharma, broadly in line with guidance. Management confirmed the FY 2026 guide unchanged. New disclosures included the Exelios sequencer summer launch and the fenebrutinib filing timeline.

Promises vs outcomes table:

The pattern reads as a management team that gives conservative guidance and over-delivers on near-term metrics. The two notable over-deliveries (NMEs into Phase III, LOE drag better than feared) are both areas where the underlying biology and manufacturing operations exceeded the company's own central case. The risk in this pattern is that the under-promise/over-deliver works for as long as biology cooperates; if any of the 19-by-2030 program were to fall short, the larger commitment is open to challenge.

Assessment: this is a management team that under-promises on near-term numbers and over-delivers, while making large multi-year commitments (the 19 NMEs, the $50bn US capex, the "top three" in obesity, the China normalisation by 2027) that cannot be scored for years. On the things that can be scored quarter-by-quarter, they have been consistently accurate. They have not yet been tested on a major pipeline disappointment under Schinecker's tenure - SKYSCRAPER-12 was the largest single failure of 2025-26, and management responded by acknowledging it and moving on without dramatic guidance changes. That is a credible response but is not yet a stress test.

10. Shareholder Friendliness Index

Dividends. Roche has now raised its dividend for 38 consecutive years through FY 2024 (CHF 9.70/share), and the Board has proposed CHF 9.80/share for FY 2025, which would mark the 39th consecutive annual increase (Roche FY 2025 release; Roche AGM 2025 release). The CFO has stated dividend growth is expected to continue in 2026. This is one of the longer dividend-growth streaks among European blue chips. The three-year DPS trajectory (FY 2023 ~CHF 9.60, FY 2024 CHF 9.70, FY 2025 CHF 9.80 proposed) is growth in nominal Swiss francs every year. The payout ratio is not unusually high and the dividend is well-covered by free cash flow.

Buybacks and share count. Roche does not run a regular share buyback program. The capital-structure event of the past 12 months was technical rather than capital-returning - the conversion of non-voting Genussscheine into Participation Certificates in March 2026 (Roche release, 16 March 2026). This was a 1:1 exchange of one form of non-voting paper for another, not a share retirement. The voting share count (bearer shares held mostly by the Hoffmann/Oeri pool at 64.97%) has been static for years. Overall the share count is essentially flat - Roche neither dilutes meaningfully nor repurchases meaningfully.

Verdict: Returns Capital - via a 39-year dividend growth streak in CHF that the company has explicitly committed to extending, even with no buyback program.

11. Insider Activities

The primary regulatory source for management transactions in Roche shares is the SIX Exchange Regulation Management Transactions database (per Art. 56 SIX Listing Rules), accessible via ser-ag.com. The public interface to this database requires a session-based search that did not return retrievable results via the fetch tools available for this report. Aggregator sources (insiderscreener.com, Simply Wall St) for Swiss insider data are similarly behind anti-scraping protection.

Insider transaction data for Roche Holding (SIX-listed) was not retrievable from the primary regulatory source or major secondary aggregators within the search budget for this report. The following structural context should be read alongside that limitation:

• The Hoffmann/Oeri family pool held 69,318,000 bearer shares (64.97% of issued bearer shares) as of 31 December 2025 per Roche's own disclosure (Wikipedia synthesis of Roche disclosures). This pool has been stable since 1948 and is governed by a long-standing shareholder agreement.
• The pool's holding has only moved meaningfully once in recent memory - in 2011 when Maja Oeri withdrew her ~5% from the pool, dropping pooled control from ~50% to ~45% before subsequent buying by remaining family members rebuilt it.
• Family control gives Roche its distinctive long-horizon stance, but it also means that "insider buying" in the traditional sense (a CEO or CFO putting six-figure sums into open-market purchases) is structurally rare. Roche executives are paid in shares and options as standard, with periodic disclosed vesting events; the family does the long-term holding.
• Roche has not announced any block-sale or material divestment by family members in the past 12 months. The conversion of Genussscheine to Participation Certificates in March 2026 was a capital-structure event, not an insider transaction.

Net assessment. The signal from this section is best read as the absence of any disclosed material insider selling (which would have surfaced in press searches even with the regulatory portal inaccessible) and the continued stability of the family pool. This is a neutral-to-mildly-positive read - the controlling shareholders are not selling, but absent retrievable executive transaction data, this section cannot be made stronger than that.

12. Scenarios

Bull case. Roche's pipeline delivers. CT-388 reads out positive Phase III in obesity in 2027 and launches in 2028 with a differentiated weight-loss + tolerability profile in combination with petrelintide. Trontinemab Phase III in Alzheimer's reads out cleanly with materially lower ARIA than first-generation anti-amyloids, allowing it to be used in patients excluded from Leqembi and Kisunla therapy - a category-redefining position. Giredestrant launches in adjuvant ER+ breast cancer and rapidly becomes the standard endocrine therapy. Fenebrutinib captures meaningful MS share alongside Ocrevus, which itself continues to ramp the subcutaneous formulation. The China diagnostics drag rolls off by 2027 as guided, Exelios captures share in NGS, and PathAI integration accelerates digital pathology. The $50bn US capex insulates Roche from MFN/tariff downside. By 2028-29 the company has a visibly rebuilt growth profile with multiple newly launched megabrands, fewer than half its revenue exposed to LOE, and the obesity franchise turning from cost-line to profit contributor.

Base case. Most of the 19-by-2030 program lands but with the usual industry attrition - call it 12-14 of the 19 reaching market, with one or two breakout franchises and the rest filling out the catalogue. CT-388 launches as the third-place obesity option, generating real revenue but not catching Lilly or Novo. Ocrevus, Hemlibra, Vabysmo, Xolair and Phesgo continue their growth trajectories, with Vabysmo flattening as Eylea HD competes. The LOE drag from Herceptin/Avastin/Rituxan winds down by 2027. China normalises broadly on schedule. US pricing is reset lower under MFN but not catastrophically - the legacy franchises absorb single-digit-percent price hits and the new launches reset to lower starting points. Group sales grow mid-single-digits in CER through the decade as management has guided; core EPS grows high-single-digits. The dividend extends its growth streak. This is the path most consistent with management's own statements and recent execution.

Bear case. Two or more of the major Phase III readouts in the pipeline disappoint - trontinemab fails to clearly beat Leqembi on ARIA, or fenebrutinib runs into a tolerability signal, or CT-388 in Phase III reads out at materially lower weight loss than the Phase II suggested. Meanwhile MFN bites harder than expected - Medicare negotiation captures Ocrevus or Vabysmo in 2027-28, knocking 30-50% off US pricing on the franchise. China diagnostics doesn't normalise on the promised timeline, with VBP expanding to immunoassay and continuing to compress prices. The $50bn US capex deploys on time but doesn't fully insulate Roche from tariff drag. By 2028 the company is in a slower-growth pocket, with the next wave of LOEs (Ocrevus US patent expiry late this decade) visible on the horizon and a thinner replacement bench than today's narrative implies. Currency translation continues to erode reported CHF earnings throughout. The dividend grows but slowly; capital allocation flexibility narrows.

Sources:

• Roche Q1 2026 press release, 23 April 2026
• Roche FY 2025 results announcement, 29 January 2026
• Roche Q3 2025 sales update, 23 October 2025
• Roche H1 2025 results page
• Roche Full-Year 2025 presentation PDF
• Earnings call transcript: Roche Q1 2026 (Investing.com)
• Earnings call transcript: Roche Q3 2025 (Investing.com)
• Roche Q4/FY 2025 earnings call transcript (Insider Monkey)
• Roche on Pharma Day 2025 - obesity strategy (CNBC)
• Roche's 'top three' obesity ambition (Fierce Biotech)
• Roche to invest $50bn in US over 5 years (22 April 2025)
• Roche AGM 2025 - dividend CHF 9.70
• Genussscheine → Participation Certificates conversion (16 March 2026)
• Nine pharma giants reach US MFN pricing deals (Fierce Pharma)
• SKYSCRAPER-12 Tecentriq+tiragolumab fail (Clinical Trials Arena)
• Petrelintide Phase II - "undifferentiated" reaction (BioPharma Dive)
• Roche to acquire PathAI (BioPharma Dive)
• Roche family ownership / Hoffmann pool (Wikipedia synthesis)
• Biopharma Dive - Roche Q1 2026 / biosimilar pressure
• Roche Pharma Day 2025 transcript landing page (Seeking Alpha)
• SIX Exchange Regulation - Management Transactions database (primary source for SIX-listed insider data)

Report complete. Four concalls covered (Q1 2026, FY 2025, Q3 2025, H1 2025), all sections written, Section 13 omitted (no qualifying coverage from SemiAnalysis/Stratechery/MBI Deep Dives). Section 11 discloses the insider-data accessibility limitation explicitly rather than fabricating transactions.