Novartis Returns to Growth, Then Tells You Five Points of It Reverse in the Second Half

Results vs. Consensus

Novartis is a Swiss foreign private issuer reporting under IFRS, with no 8-K. Its primary operating measure is "core" (non-IFRS) results and it guides in constant currencies. Every income-statement figure below traces to the Form 6-K Exhibit 99.1 furnished to the SEC on July 21, which reproduces the media release verbatim. Growth rates are as reported by the company, with the currency basis labelled where it matters.

Q2 2026 Scorecard

Consensus for core EPS spans $2.13 to $2.20 across the sources we track, and for net sales $13.96B to $14.17B. We use the most widely cited figures above; on the lower marks the core EPS beat runs to +13.1% and on the higher revenue mark the sales beat narrows to +1.7%. Where a reliable consensus figure does not exist for a line, the column reads "n/a" rather than a manufactured number.

Q2 2026 vs. Q2 2025

H1 2026 vs. H1 2025

Both tables are the income-statement figures as presented by the company, not aggregations of brand-level or verbally quoted amounts. Subtracting the quarter from the half gives the implied first quarter, which is where the shape of this year lives: Q1 2026 net sales of $13,113M against $13,233M, core operating income of $4,897M against $5,575M, a core margin of 37.3% against 42.1%, and core EPS of $1.98 against $2.27. Novartis lost 12.8% of its core EPS in the first quarter and got all but a cent of the second quarter back. That is the guided shape working exactly as management described it in February.

Quality of Beat/Miss

• Revenue: Better underneath than the headline, and worse than the beat. The company's own Q2 bridge decomposes to volume +18pp, generic competition -14pp, pricing -3pp and currency +2pp. Eighteen points of volume growth is a business performing well. But roughly a point of the reported growth is inventory movement tied to an SAP system cutover across multiple geographies, which management confirmed under direct questioning and which shifts into the third quarter. The honest read of the operating quarter is flat constant-currency growth with an eighteen-point volume engine underneath a fourteen-point generic drag.
• Margins: Low quality this quarter, by management's own account. The printed 41.2% core margin is down 70bps cc, and about five points of the core operating income line is phasing that reverses. The CFO also flagged, unprompted, that "Q2 is generally a stronger margin quarter when we look at the phasing across the whole year." Two separate warnings against annualising this margin, in the same set of prepared remarks. The durable read is the half: 39.4%, down 230bps cc.
• EPS: Core EPS of $2.41 was flat in USD against core net income down 3%. The entire gap is share count, which fell about 2.4% year over year to roughly 1,900M on our calculation from the reported core net income and core EPS. That is a materially smaller per-share tailwind than the company generated in 2025, and it shrinks further from here as the buyback slows. The IFRS line tells the harsher story: net income down 19% and EPS down 18% cc on higher income taxes and higher interest expense, with the gap between core and IFRS operating income widening to $1,190M in the quarter.
• Cash: Free cash flow of $5,561M fell 12% on operating cash flow down 12%, and the half is down 9% to $8,891M. This is no longer a phasing story the way the Q4 2025 decline was. Two consecutive halves of cash conversion running below the prior year, against a balance sheet that has just absorbed $15.3B of acquisition outflows, is the line we would watch most closely into the second half.

Brand and Franchise Performance

Novartis has reported as a single pure-play Innovative Medicines business since the Sandoz separation, so the meaningful decomposition is by brand. The top 20 brands produced $11,781M in Q2, 81.8% of net sales, and $22,537M in the half, 81.9%. The quarter is best read as the moment the two portfolios inside that revenue line crossed.

The growth portfolio

The six brands management designates as priority growth drivers (Kisqali, Kesimpta, Pluvicto, Scemblix, Leqvio, Fabhalta) contributed $5,037M in Q2, or 35.0% of net sales, and $9,413M in the half, or 34.2%. Two quarters ago that cohort was 30.3% of a quarter. It has added nearly five points of revenue share in six months. Management put the wider growth-driver set, which includes Cosentyx and the Zolgensma group, at +36% cc.

Kisqali

The franchise crossed a threshold that matters for the peak-sales argument: US sales passed $1B in a single quarter for the first time, on 39% US growth, with total brand sales of $1,695M and 43% cc growth. The early breast cancer engine is where the durability sits, with 58% of new patients now coming from the exclusive N0 and N1 nodal populations, a 16% increase in the total prescriber base, approval in 76 countries and reimbursement in 42, and a 79% early breast cancer new-patient share in Germany. Against a Street carrying roughly $1.72B, the print was a marginal miss on consensus and a clear beat against the modelled base. The company also disclosed six-year NATALEE follow-up showing clinically meaningful overall survival in the broadest at-risk early breast cancer population, with full data reserved for an upcoming congress, and the FDA granted pediatric exclusivity adding six months to every Orange Book patent.

"Our guidance is a second half 2031, Q3 2031 guidance for LOE for KISQALI with the pediatric exclusivity, that's inclusive of the settlements that we have with generic manufacturers." — Vas Narasimhan, CEO

Assessment: The gross-to-net noise that made this brand the most misread number of the fourth quarter is fully behind it, exactly as management said it would be. What replaced it is a cleaner and more important question. Kisqali is now the company's growth centrepiece with a hard exclusivity date in the third quarter of 2031, which the company has now stated with unusual precision and confirmed is inclusive of generic settlements. That is one year past the end of the 2025 to 2030 guidance window. The $10B peak-sales goal looks more credible after this quarter, and the shape of the decade after it looks no different.

Scemblix

The strongest brand in the portfolio on any growth measure: $562M, +89% cc, with US sales up 93% and a Week 144 ASC4FIRST readout showing superior major molecular response against all standard-of-care tyrosine kinase inhibitors, including a 15.2 percentage point higher rate versus second-generation agents, with fewer grade 3 or higher adverse events and less than half the discontinuation rate. Management now expects to reach first-line new-patient share leadership in the US in the second half. Ex-US the brand is still driven by third line and beyond with 75% new-patient share in key markets, but Japan has already reached first-line leadership and Germany's first-line share is at 15%. Against a Street carrying roughly $481M, this was a 17% beat.

Assessment: This is the cleanest resolution of a Q4 open question in the entire release. Three quarters ago the first-line share plateau was the bear point on this brand, and management's defence was a diagnosis it could not prove in any single quarter: noisy rare-disease data, access perception rather than access reality. That diagnosis has now been validated by the thing it predicted. Payer acceptance improved, the share moved, and the company is guiding to outright first-line leadership within two quarters. We held management to it and they delivered.

Pluvicto and the radioligand platform

Pluvicto grew 43% cc to $651M with pre-taxane use now driving over 70% of US new patients, and ex-US new-patient growth of 83%. The infrastructure keeps compounding: more than 880 US sites and more than 650 ex-US radioligand sites. The near-term catalyst is a Q3 approval in hormone-sensitive prostate cancer on PSMAddition, which expands the eligible pool by 75%, with two-thirds of those patients already sitting with treaters who use Pluvicto today or have established referral patterns. Subgroup data presented at ASCO showed consistent radiographic progression-free survival benefit regardless of disease volume or presentation, against the previously reported 28% reduction in risk of progression or death, and AUA data showed a 58% reduction in risk of PSA progression.

"We think that RLT will behave very differently than either small molecule and potentially biosimilars just for biologics, just given the logistical complexity and as well the expectation that physicians have that the medicine is delivered on time each time, given the nature of the logistics for the office." — Vas Narasimhan, CEO

Assessment: The generic question arrived earlier than we expected. Lutathera generics were cleared to launch by the Delaware courts this year, neither filer has FDA approval yet, and the answer above is the first real test of our own view that this platform's moat is logistics rather than molecule. We still believe it, and the 880-site network is the evidence, but "we think" is doing load-bearing work in that sentence and nobody has data. The mHSPC approval remains the single largest revenue catalyst in the next twelve months that does not depend on a coin-flip readout.

Leqvio and the arrival of oral competition

Leqvio grew 59% cc to $480M with US growth of 55% and monthly prescription growth of 49%. The Medicare Part B segment, which management describes as the most important, sits at 23.3% share, up 3.6 points year to date. China's NRDL listing has roughly doubled share versus pre-listing levels, and management now frames the country as potentially the largest market it has ever built there. Two cardiovascular outcomes studies read out in 2027. The competitive event of the quarter was the approval of an oral PCSK9 inhibitor carrying an eight-hour fasting requirement.

"The fact that we are not participating in the gross-to-net battle that will ensue between the monoclonal antibodies and the orals and actually are in a segment that's insulated from that, I think gives us a strong position in the longer run for our goals of a $4 billion-$5 billion-plus product." — Vas Narasimhan, CEO

Assessment: The insulation argument is the right one and it is not costless. Novartis is deliberately competing in the physician-administered buy-and-bill segment rather than the retail pharmacy channel, which does protect net price from an oral price war, but it also caps the addressable share of a very large market at whatever fraction of patients will accept a twice-yearly injection at a physician's office. Management conceded that competitors without the food effect are coming. We treat the $4B to $5B ambition as achievable and the path to it as narrower than it was a year ago.

Cosentyx

Reported growth of 10% cc to $1,824M, ahead of a Street carrying roughly $1.72B, with US growth of 16% on hidradenitis suppurativa and the IV formulation and a new-patient share holding in the high 40s. Management was explicit that the reported number flattered the trend: underlying US growth is around mid-single-digit, "as we've guided to in the past", and part of the quarter came from one-time items. Ex-US, European growth was partly offset by a decline in China, where management confirmed under follow-up that it still holds 70% market share and leadership. The Phase III REPLENISH data in polymyalgia rheumatica was published in the New England Journal of Medicine, showing doubled sustained remission at week 52 with reduced cumulative glucocorticoid exposure, and is under review in the US, EU and Japan with approval expected in the second half.

Assessment: Management pre-empting its own beat by pointing at the underlying mid-single-digit rate is the correct behaviour and worth crediting. Cosentyx remains the dependable compounder on the road to $8B, and polymyalgia rheumatica is a genuine incremental leg in a population currently managed on chronic steroids. The China competitive pressure is the watch item, disclosed early rather than late.

The renal and immunology launches

Fabhalta grew 88% cc to $225M and received traditional FDA approval in July as the first and only complement inhibitor shown to significantly slow kidney function decline in primary IgA nephropathy, converting the accelerated approval. Vanrafia's final 30-month ALIGN results published in The Lancet showed clinically meaningful slowing of kidney function decline with sustained proteinuria reduction, consistent in patients on SGLT2 inhibitors, and traditional-approval submissions are complete in the US and EU. Zigakibart's Phase III BEYOND readout remains anticipated in H1 2027, unchanged from the timeline set at Q4. Rhapsido disclosed its first revenue at $64M, with over 4,000 US prescribers, over 10,000 patients treated, 60% of them first-line, and two of the three major pharmacy benefit managers now covering it.

Assessment: The three-asset renal position we described as the more durable point at initiation has now converted two of its three legs to traditional approval or completed submission on kidney-function outcomes, which is the specific claim that differentiates it. This is the pillar of the thesis that has advanced most quietly and most concretely this year.

The eroding portfolio

Assessment: Entresto, Promacta and Tasigna together produced $1,502M in Q2, 10.4% of net sales, against 12.4% in Q4 2025 and 19.2% across FY25. The named cohort is nearly finished eroding, and Promacta and Tasigna are now small enough that their percentage declines no longer move the company. The more consequential development is the one we flagged as an omission at Q4: Xolair has gone from -8% cc in the fourth quarter to -25% cc in this one, and Tafinlar plus Mekinist from -2% to flat with the half at -7%. Together those two brands are $923M of quarterly revenue that has turned. The erosion cohort has widened beyond the three brands everyone is modelling, and it did so without drawing a single question on the call.

Key Topics & Management Commentary

Overall Management Tone: Management was relaxed and forward-leaning in a way it has not been in three quarters, treating the return to growth as a settled fact and spending most of its energy on the second-half readout calendar rather than on defending the print. The notable feature was pre-emptive candour on the numbers, with the phasing benefit, the SAP-driven stocking and the underlying mid-single-digit Cosentyx rate all volunteered rather than extracted. That candour did not extend to the pipeline, where management declined to define what "clinically meaningful" would mean for any of the three readouts investors care most about, and where the pelacarsen success bar was described in materially lower terms than at any prior call.

1. The five points that reverse

The central disclosure of the quarter, made in prepared remarks rather than dragged out in Q&A. One-time phasing added roughly a point to net sales and roughly five points to core operating income, and it unwinds in the second half. The sales piece is inventory movement; the cost piece is clinical trial spend that was budgeted for the second quarter and slid into the third.

"Worth to note that Q2 results were also positively impacted by some one-time phasing items, which will reverse in the second half. Together, these items positively impacted net sales by approximately one percentage point and core operating income by about five percentage points." — Mukul Mehta, CFO

The CFO added a second caution in the same passage, noting that "Q2 is generally a stronger margin quarter when we look at the phasing across the whole year." Two warnings against extrapolating the 41.2% margin, delivered voluntarily, inside one paragraph.

Assessment: This is a company telling you not to believe its own headline, which is the behaviour you want and the opposite of what a management team optimising for a one-day stock move would do. It also means the print that drove a 2.9% rally is roughly a 3.6% core EPS beat rather than a 9.5% one, and that the second half must absorb approximately $300M of returning cost while still delivering the guided mid-to-high-single-digit core operating income growth. Guidance was reaffirmed with full knowledge of the reversal, so the guide is not at risk from it. The multiple applied to the quarter is.

2. Net debt nearly doubles

Net debt reached $39.4B at June 30 against $21.9B at December 31, an increase of $17.5B in six months. The disclosed bridge is $8.9B of free cash flow more than offset by $15.3B of net cash outflow for M&A, intangible assets and other acquisitions, the $9.1B annual dividend, and $3.1B of cash outflows for treasury share transactions. Long-term ratings are unchanged at Aa3 and AA-. The company also agreed in July to acquire Myricx Bio, closing in the second half, and completed the Pikavation and Excellergy acquisitions during the quarter.

Against our FY26 core operating income estimate the ratio is roughly 1.8 times, against 1.0 times at the end of last year.

Assessment: This is the largest single change to the investment case in the quarter and it drew no question on the call. The Avidity funding was disclosed and modelled, so the direction is not a surprise; the magnitude is. A pharmaceutical balance sheet at 1.8 times core operating income is not stressed and the ratings agencies agree. What it does is remove the optionality that made this company's capital allocation attractive. Novartis has spent the flexibility it built over the pure-play transition, and it has spent it in the year its earnings are down, which is the least comfortable sequencing available.

3. The buyback slows to a trickle

The company repurchased 18.2 million shares for $2.8B in the first half, of which 13.8 million shares and $2.1B ran under the up-to-$10B programme announced in July 2025, leaving $5.6B to execute by the end of 2027. The remainder was 4.4 million shares to offset participation-plan dilution and 2.0 million repurchased from employees. Against that, 12.7 million shares were delivered to employees under prior-year plans, so the total share count fell by only 7.5 million versus December 31.

Assessment: Last year the company retired 66.9 million net shares. This year it has retired 7.5 million in six months. The buyback carried per-share growth through the fourth quarter and it is now contributing roughly 2.4 points of year-over-year support rather than the 4 to 5 the model was built on. With $5.6B left across six quarters and a balance sheet that has just taken on $17.5B of net debt, the per-share bridge from here has to come from the P&L. That is the correct place for it to come from. It is also the harder place.

4. Pelacarsen: the readout slipped and the bar moved

At the February call management guided the HORIZON Lp(a) cardiovascular outcomes readout to mid-2026 and described the trial as powered for a 20% relative risk reduction at the 70 mg/dL cutoff and 25% at 90 mg/dL, while conceding a win was possible at a mid-teens reduction. The mid-2026 window closed on June 30 without a readout. It is now guided to the second half, among the readouts expected "in the coming months." In this quarter's prepared answer the powering was described differently.

"I think for pelacarsen, as we've guided in the past, we've powered the study for the kind of 13%-15% CVRR benefit, and certainly are hopeful to see that level or higher." — Vas Narasimhan, CEO

Pressed later in the call, management confirmed the original design assumption stands and that the lower number is the minimum detectable effect rather than the powering target, and went further on what would count as success.

"You are correct, 13%-15% would mean a win. It's powered for 20% for the patients who are 70 mg in DL and above or 25% for the 90 mg DL and above. I think we would say in the mid-teens is clinically meaningful given that these patients have no other option, and that this is an independent risk factor." — Vas Narasimhan, CEO

Assessment: In February the mid-teens outcome was described as something the company would have to evaluate, and as a result that would raise a commercial question rather than settle one. In July it is clinically meaningful. Nothing about the trial changed between those two statements, because management remains blinded. What changed is the framing, in the direction of a lower bar, immediately before the readout. We do not read this as misleading, since the correction came promptly and completely under questioning, and the design parameters were restated accurately. We do read it as expectation management, and it tells you which tail the company is preparing investors for. This is now the largest single binary in the story and the second consecutive quarter in which its timing has moved right.

5. Three readouts, and no definition of success for any of them

The call's opening substantive question asked management to define, in advance, what it would consider clinically meaningful for pelacarsen, remibrutinib in relapsing MS, and del-desiran in DM1. The answer declined on all three, on the stated grounds of preserving the ability to present at medical congresses.

"I think that the reality is we have to be thoughtful because we want to be able to preserve the ability to present this data at high profile congresses in the future. We'll navigate that best we can, making sure investors have clarity on what we believe the path forward is, but still preserving that ability to present the data as well." — Vas Narasimhan, CEO

The question was framed explicitly as arising from a change in the company's communication on ianalumab in Sjögren's, where the "clinically meaningful" language had shifted. Management confirmed it would focus on the primary endpoint and statistical significance, and would comment on additional secondary endpoints only "as appropriate".

Assessment: The congress-embargo rationale is real and every large pharma faces it. It is also a general-purpose answer that happens to be most useful in exactly the quarter before three readouts land. The practical consequence for an investor is that there is no pre-committed standard against which any of these three results will be judged, which means the company retains full discretion to characterise a marginal outcome favourably. We are not assigning expected value to any of the three, and this exchange is why.

6. The Avidity assets deliver ahead of schedule

Two of the acquired programmes produced concrete progress in their first full quarter of ownership. A Biologics License Application was submitted to the FDA for accelerated approval of del-zota in Duchenne muscular dystrophy patients amenable to exon 44 skipping, using dystrophin as a surrogate biomarker, on the back of the EXPLORE44 data and an existing Breakthrough Therapy designation, with first launch expected in the first half of 2027. The del-brax biomarker cohort of the FORTITUDE Phase I/II study in facioscapulohumeral muscular dystrophy met its primary and key secondary endpoints, with statistically significant reductions in KHDC1L and creatine kinase indicating target engagement and reduced muscle damage.

"Our base case remains a submission in 2028, but based on the data that we've seen in the biomarkers and ongoing work we're currently conducting... we plan to engage FDA and other regulatory authorities in the coming months." — Vas Narasimhan, CEO

Assessment: This is the best news in the quarter that nobody priced. The deal that is costing 1 to 2 points of margin and a meaningful share of the $17.5B net debt increase has, within one quarter, produced a filed accelerated-approval BLA and a biomarker package management thinks may support pulling a 2028 submission forward. That does not make the balance sheet cost disappear, but it does change the character of the trade from paying for optionality to paying for optionality that is converting. Management was appropriately careful: it has a case worth making to the FDA, not a case it expects to win.

7. Rhapsido: deliberately slow on purpose

The launch disclosed $64M in its first reported quarter, with over 4,000 US prescribers, over 10,000 patients treated, 60% of them first-line, and two of the three major pharmacy benefit managers covering with prior authorisation to label. Management repeatedly declined to promise an inflection, and explained why in terms of the multi-indication strategy rather than launch weakness. The Phase III RemIND study met its primary endpoint across the three most common chronic inducible urticaria subtypes with no liver safety concerns, and the REMIXED extension showed a 72% lower relapse risk over 18 months.

"We're trying to stay really disciplined on the gross to nets here. We just believe that if we play the long run out here, remibrutinib has the potential to be used in a broad range of indications... Any points we give now, we won't be able to get back in the future." — Vas Narasimhan, CEO

Assessment: Trading launch velocity for net price on an asset that management believes will carry six or more indications is the right decision and an unpopular one, because it guarantees several quarters of unimpressive revenue prints on the company's most-watched launch. We would rather own the discipline than the inflection. It does mean that anyone modelling Rhapsido off the prescription-tracking data will overshoot the revenue line for at least a year, and that the asset's value is almost entirely in the MS and food-allergy readouts rather than in the urticaria base.

8. Ianalumab and the Sjögren's launch

The first-ever approved therapy in Sjögren's disease is on track for a US approval in the third quarter with no advisory committee planned and mid-cycle review meetings complete. Pooled NEPTUNUS data presented at EULAR showed consistent improvement across most ESSDAI domains, and a 108-week extension showed deepening disease control with a favourable safety profile and clinically meaningful improvement in the placebo crossover group. Submissions are under review in the US, EU and Japan. The ITP first-line readout comes in the second half of 2026, with SLE, lupus nephritis and systemic sclerosis in 2027.

"This is a very heterogeneous patient population. Even the ESSDAI endpoint is covering a broad range of domains. We do see in our own data sets there are patients who are super responders and patients who respond less well." — Vas Narasimhan, CEO

Assessment: Launching into a disease with no approved therapy is the most favourable access setup available, and management's expectation of broad initial coverage is reasonable. The candour about heterogeneous response is the part worth holding onto: a launch where physicians trial broadly and a meaningful fraction of patients cycle off produces a revenue curve that ramps fast and then flattens sooner than the prescriber count suggests. The multi-blockbuster claim rests on the 2027 indications, not on Sjögren's alone.

9. M&A discipline in a repriced market

Asked about reported openness to larger deals, management denied any change in strategy: steady transactions with upfronts in the sub-$2B range, with selective larger deals of the Avidity type when an asset fits the platform or therapeutic-area strategy. The more interesting part of the answer was on price.

"You see now ourselves and our peers doing upfronts that are over $1 billion for assets that have limited or no clinical data, which is, I think, if you look at the long arc of the sector, a significant shift. Which I think just means you have to have higher levels of conviction in the science and differentiated." — Vas Narasimhan, CEO

Four transactions were named across the quarter and the month after it: Pikavation and Excellergy completed, Myricx Bio agreed in July and closing in the second half, alongside the Avidity integration.

Assessment: The stated strategy and the observed behaviour agree, which was not obvious going in. But four transactions in a quarter, in a market management itself describes as structurally repriced upward, against a balance sheet that just added $17.5B of net debt, is a lot of activity to run concurrently. The discipline claim is credible on price per asset and less credible on aggregate pace.

10. The second-half readout stack

Four readouts landed in the first half. The second half carries pelacarsen, remibrutinib in relapsing MS, del-desiran in DM1, ianalumab in first-line ITP, Rhapsido in hidradenitis suppurativa, abelacimab at 75% of events in patients ineligible for direct oral anticoagulants, and Phase II data for QCZ484 and for VHB937 in ALS. Management stated that these readouts "would allow us to raise our mid to long-term growth outlook."

Assessment: That sentence is the closest thing in this release to a forward statement on the 5% to 6% CAGR, and it is conditional on the readouts. The breadth is real and this is a genuinely dense six months of catalysts. The framing is also asymmetric in the company's favour: a positive stack raises the outlook, a negative one leaves it where it is. Investors should hold the reverse position, because the 5% to 6% CAGR already contains a probability-weighted pipeline contribution. A failed stack does not leave the outlook unchanged; it removes a contribution already embedded in it.

11. Gross margin has found its floor, and SG&A is carrying the offset

Management was direct that the gross margin is now at a level investors should model forward, with the second half of last year as the reference point, and that portfolio mix will keep pressuring it while manufacturing productivity holds the line. The offset is running through selling and administrative costs, where management identified third-party spend above $10B as the target.

"We look at Q2, we actually see while quarter-over-quarter gross margin has been a negative, but SG&A, as a percentage of sales, has actually been a positive, negating that gross margin impact." — Mukul Mehta, CFO

Assessment: The margin architecture for the next two years is now explicit: gross margin flat to down on mix, R&D up on Avidity plus the Tourmaline, Regulus and Amplyx assets, and SG&A leverage doing the work of holding the total. That is a coherent plan and a narrower one than the 2020 to 2025 period, when every line contributed. It also explains why the return to a 40%-plus core margin is not guided until 2029.

Guidance & Outlook

Full-year 2026 guidance was reaffirmed, not raised, on every line, despite a first half management itself described as landing at the upper end of the range set in January.

The Q3 versus Q4 asymmetry is the modelling point of the quarter. Management flagged, without being asked, that the two halves of the second half look nothing alike, because roughly $800M of US Entresto sits in the third-quarter 2025 base and almost none of it sits in the fourth-quarter base.

"There will be a notable difference in the sales growth rates between the two quarters, Q3 and Q4. This is because we still have about $800 million of U.S. Entresto in the sales in previous year quarter three base." — Mukul Mehta, CFO

Backing out the reported quarters, third-quarter 2025 net sales were $13,909M, so that $800M is 5.8 percentage points of headwind sitting in the third-quarter comparison alone. To average mid-single-digit constant-currency growth across the half with that drag in the first of the two quarters, the fourth quarter has to grow at roughly double the half's average rate. A third quarter that prints low-single-digit growth would be entirely consistent with guidance, and will not look that way in the headlines.

Implied core EPS math: the first half delivered $4.39 against $4.69. Applying the guided mid-to-high-single-digit constant-currency core operating income growth plus the one-point FX tailwind to the second-half 2025 core EPS base of $4.29, and allowing for core EPS growing slightly behind core operating income because of the higher net financial expense and the slower buyback, we get second-half core EPS of roughly $4.50 to $4.63 and FY26 core EPS of $8.85 to $9.05, against $8.98 delivered in 2025. That is a raise from the $8.60 to $8.90 range we set at initiation, and the raise is earned: the first half came in about five points better on core operating income than the guide it was measured against. The company still does not guide core EPS.

Street at: the reaffirmation rather than a raise was the point of contention on the day. The reading circulating on the sell side was that holding guidance after a first half at the upper end implies a step-up in second-half spending, which is consistent with both the disclosed phasing reversal and management's own statement that the second half is structurally the higher-spend half. Price-target revisions on the day were modest and upward, and at least one rating moved higher.

Guidance style: this management team guided twice-raised numbers in 2025 and delivered at or above the raised range on both lines. In 2026 it guided a low-double-digit first-half core operating income decline and delivered -7%, then declined to raise the full year. The pattern is consistent: guide to be beaten, and do not spend the beat. Applied to the second half, the guided mid-to-high-single-digit core operating income growth should be read as a floor with a returning $300M of phasing cost already inside it.

Analyst Q&A Highlights

Defining success ahead of three pivotal readouts

The call's first substantive exchange asked management to commit, in advance, to what it would treat as a clinically meaningful result for the three readouts investors care most about. The question was framed as arising from an observed change in the company's own language on a recently completed programme. Management declined across all three, citing congress-presentation constraints.

Q: "Given the change from your communication in and out of ianalumab Sjögren's on clinically meaningful, just wondering whether you've decided internally how you define clinically meaningful for three reads investors most focus on, so Pella, remi MS, DM1... Is it fair to think that any statistically significant benefit is clinically meaningful in your eyes for different reasons for each asset, and any changes in level of confidence on each?"
— Sachin Jain, Bank of America

A: "No change from previous comments. We don't know anything else that I can provide on any of the three. I think in terms of how we will read them out, I think we always focus on the primary endpoint and statistical significance, and reaching the goal and the primary endpoint in the study."
— Vas Narasimhan, CEO

Assessment: The most important question of the call and it went unanswered by design. "We always focus on the primary endpoint and statistical significance" is a standard that permits any statistically significant result to be presented as a success regardless of effect size, which is precisely the scenario at issue for the cardiovascular readout. The follow-up, which established that the company has never disclosed how either the MS or DM1 studies are powered, closed the last route to independently handicapping any of the three. An investor sitting here cannot distinguish a good outcome from a marginal one in advance, and will be told which it was by the party that chose not to define it.

Whether a mid-teens cardiovascular benefit clears the bar

A direct challenge to a figure management had used in prepared remarks, testing it against the published trial design. The exchange forced a restatement of the actual powering assumptions and, more consequentially, an explicit statement that a result well below them would still be characterised as clinically meaningful.

Q: "On pelacarsen, a quick follow-up, actually, just you clarified the 13%-15% is what the trial is minimally powered to detect. I think you said it was just what it was powered for, but I think the design paper says it's 20% on all comers. Would you view that 13%-15% as a clinically meaningful result?"
— Graham Parry, Citigroup

A: "You are correct, 13%-15% would mean a win. It's powered for 20% for the patients who are 70 mg in DL and above or 25% for the 90 mg DL and above. I think we would say in the mid-teens is clinically meaningful given that these patients have no other option, and that this is an independent risk factor."
— Vas Narasimhan, CEO

Assessment: Management corrected itself immediately and completely, which is the right response and removes any suggestion of an attempt to mislead. What remains is the substance: six months ago a mid-teens result was something the company said it "would have to evaluate" and that raised a commercial question; today it is clinically meaningful. The trial has not read out and the company is blinded, so no new information caused the change. The bar for declaring victory has been lowered in public, immediately before the event. That is the single most useful thing this call disclosed about how to position into the readout.

Decomposing the one-time items in the quarter

The first question on the numbers asked management to break the disclosed phasing benefit into its parts. The answer separated an inventory effect on the top line from a clinical-trial cost deferral on the bottom, and explained why the combination produces a five-point core operating income benefit from a one-point sales benefit.

Q: "Maybe Mukul, could you give us a little bit more color about the one-off events which impacted the Q2 top line and operating profit margin? Some color on that point would be great."
— Florent Cespedes, ODDO BHF

A: "This is primarily inventory-related changes that we saw across the whole world. This would simply move from Q2 to Q3 from an inventory perspective. On the cost side, on the R&D phasing side, we have a couple of clinical trial-related costs that... were planned for Q2 and now will move to Q3. If we put both of them together, on the top line it's an impact of 1%, and on the bottom line, the cumulative impact of the top line over-delivery added or compounded by the cost phasing leads to a 5% impact on the core operating income."
— Mukul Mehta, CFO

Assessment: A clean, specific, quantified answer with the mechanism spelled out, and the CFO's second such disclosure in his first six months in the seat. The five-point figure is the compounding of near-100% incremental margin on the deferred inventory sales and the deferred trial spend, which is exactly why a one-point revenue effect produces a five-point profit effect. Anyone modelling from the reported 41.2% margin without this adjustment is modelling a number the company has explicitly disowned.

The source of the inventory build

A follow-up pressed on which products drove the stocking, on the reasonable assumption that a concentrated build would signal channel behaviour on a specific brand. The answer located the cause in an internal systems migration rather than in commercial dynamics.

Q: "Just on the stocking in the quarter, we understand the 1% of sales beat was from stocking. Cosentyx explained some of this, but not all of it. Are you able to provide any color on what other drugs may have experienced stocking?"
— Colin White, UBS

A: "This was related to the implementation of our new SAP system, where it is often the case when we roll that out in multiple geographies, we do have to shift stocking levels for the cut over to the new SAP system. That is the driver and the reason why it is not associated with one brand, per se."
— Vas Narasimhan, CEO

Assessment: The best possible answer to this question, because a systems-migration explanation is verifiable in a way that a channel-dynamics explanation never is, and it predicts the reversal precisely rather than vaguely. It also quietly disposes of the alternative reading, which would have been distributors building ahead of a price action or a supply concern. We take this at face value and note that it makes the third-quarter revenue line easier and the third-quarter margin line harder at the same time.

Whether second-half cost control is structural or borrowed

A challenge to the arithmetic of the reaffirmed guide: if the second half must absorb returning trial costs and higher acquired R&D while still growing core operating income mid-to-high single digit, the cost line has to do work, and the question was whether that work is repeatable into next year.

Q: "The guidance for mid to high single-digit core EBIT growth in the second half seems to imply more cost control, especially taking into consideration the Avidity R&D phasing that you just mentioned... Where are you containing costs, and is that something that we need to take into consideration as we think about 2027, or is that sort of tucking in expenses that will not recur?"
— Michael Leuchten, Jefferies

A: "On R&D, we will continue to invest what the pipeline needs. This would mean incremental investments this year on the back of Avidity, but also a couple of other assets that we took on... SG&A, as a percentage of sales, is a place where we believe we as a company can do more productive efforts, specifically focused on third-party spend, which is upwards of $10 billion for the company."
— Mukul Mehta, CFO

Assessment: The answer is that nothing is being tucked in, which is the reassuring version, and that the entire margin offset now rests on third-party SG&A productivity, which is the less reassuring version. A $10B third-party spend pool is genuinely large enough to fund several years of leverage, so the plan is not implausible. But it is a single lever, and the same analyst extracted an equally specific answer on the fourth-quarter SG&A question two calls ago. Management is consistent and quantitative on this line, which is why we would treat a miss on it as a genuine signal rather than noise.

Grounds for confidence in the MS programme

A recurring line of questioning across the call probed what underpins management's confidence in an oral asset that entered Phase III without Phase II efficacy data, first on relapse rate against an active comparator and later on disability progression, where a competitor molecule in the same class failed to show statistical significance.

Q: "Please could you recap specifically what gives you confidence that remibrutinib can improve upon the annualized relapse rate of about 0.1 Aubagio has achieved in recent RMS studies?"
— Multiple analysts incl. Colin White, UBS; Graham Parry, Citigroup

A: "Well, I think for us, as you know, we don't have phase II data. We're basing this based on other BTK inhibitors performance in similar studies. We continue to monitor blinded rates for safety and relapse rates, and I think taken together, that gives us confidence that the study is performing as expected, in terms of the differences between the active and the control arm... I think for us, clearly with remibrutinib, in addition to looking at annualized relapse rate and MRI performance, how it performs in disability progression will be important to understand, is remibrutinib something that would be used in a setting post the B-cell antibodies, or could it be used in a setting in line or ahead of the B-cell antibodies?"
— Vas Narasimhan, CEO

Assessment: Candid to the point of being unhelpful, which we prefer to the alternative. The admission that there is no Phase II efficacy data and that blinded aggregate rates are the only in-study signal is a fair statement of the epistemic position, and the separate admission that disability progression cannot be handicapped is the same posture management took on the cardiovascular readout in February. The asymmetry worth noting is commercial rather than clinical: a relapse-rate win without a disability-progression win positions this asset behind the B-cell antibodies rather than alongside them, and management said as much. That is two distinct outcomes inside what the market is treating as one binary.

Whether an oral competitor changes the lipid-lowering franchise

Questioning turned to the recently approved oral entrant in the same therapeutic space, specifically whether its label or its pricing altered the company's view of its own asset's commercial ceiling. The response conceded nothing on peak sales and reframed the competitive set around channel rather than molecule.

Q: "Were there any surprises in the label or the pricing of the oral PCSK9 inhibitor recently approved that alter Novartis's view of the commercial potential for LEQVIO?... Any thoughts on how limiting the fasting ultimately will be?"
— Steve Scala, TD Cowen

A: "I think clearly an 8-hour fast plus the 30... in this particular drug. I think we'll have to see because obviously patients can find ways to manage that. I would say there are other competitors coming that, as far as we understand, may not have the food effect. Given that, I think we just have to focus on our segment and focus on the patients that we can reach in that Part B buy and bill setting in the U.S."
— Vas Narasimhan, CEO

Assessment: Management declined to take the easy win. It would have been simple to argue that an eight-hour fasting requirement blunts the oral threat, and instead the answer immediately flagged that competitors without the food effect are coming and pivoted to the defensible ground of the physician-administered channel. That is an honest read of a competitive position that has narrowed, and it implicitly concedes that the retail-pharmacy segment of a very large market is not one this franchise intends to contest.

What They're NOT Saying

1. Not a word about the most-favoured-nation pricing agreement. In February management said the December 19 agreement's implications were already reflected in both 2026 guidance and the five-year CAGR without disclosing the magnitude, and explicitly promised to clarify the ex-US launch pricing strategy during the course of this year. Two quarters later the release does not mention it, the prepared remarks do not mention it, and not one of the fifteen analyst questions raised it. A promised disclosure that quietly stops being discussed is not the same as a resolved issue.
2. No question and no comment on the $17.5B increase in net debt. The largest change to the balance sheet in years appeared in the release and drew no commentary on the call. Management did not frame it, and nobody asked. Both facts are notable.
3. Still no core EPS guidance, in the year the gap between core operating income and core EPS is at its widest and least predictable, with net financial expense guided to $1.7B and the buyback pace cut by more than half.
4. No 2027 framing at all. The one question that reached for it, on whether second-half cost control carries forward, was answered on the current year only. With three readouts landing in the next six months and a debt-funded acquisition to annualise, the absence of any 2027 anchor leaves an unusually wide range of outcomes for a company that normally frames the medium term readily.
5. No update on the 5% to 6% 2025-2030 CAGR or the 2029 return to a 40%-plus core margin. Both were reiterated as recently as February. Neither appears in this release. Management referenced only the "mid-term outlook" and said the second-half readouts "would allow us to raise" it, which is not the same as confirming the standing numbers.
6. Nothing on Xolair or Tafinlar plus Mekinist, the same omission we flagged at Q4. Xolair has gone from -8% cc to -25% cc across two quarters and Tafinlar plus Mekinist has rolled to flat, together $923M of quarterly revenue that turned, with no commentary and no question in either quarter.
7. No accretion timeline for Avidity, again. The margin dilution is quantified and now visible in the reported gross margin. Two quarters after the deal was struck and one quarter after it closed, there is still no statement of when the acquired programmes contribute revenue or when the transaction turns accretive.
8. No answer on how the MS and DM1 trials are powered. Asked directly, management confirmed it has never disclosed the powering assumptions for either and would not do so now. The cardiovascular trial's parameters are public because they were published; the other two are not.

Market Reaction

• Pre-print setup: NVS closed at $149.56 on July 20, up 8.5% year to date against the S&P 500's +8.7%, up 1.7% over the trailing 30 days and up 31.0% over the trailing twelve months. The 52-week closing range entering the print was $113.50 to $168.62, so the stock sat 11.3% below its closing high, set on February 27. That is a materially different setup from the fourth-quarter print, when the ADR entered within 1.6% of its high.
• Reaction session (before-open reporter, so the print day is the reaction day): opened at $155.39, a 3.9% gap up, ran to a session high of $159.30 (+6.5%), and closed at $153.87, up 2.9% or $4.31, on a session when the S&P 500 rose 0.9%.
• The fade is the story: the ADR finished 3.4% below its intraday high, giving back more than half the gap-up. Two independent sources put the premarket bid at 3.0% to 3.7%, so the stock was already priced for a beat before the opening bell and spent the session working through what the beat contained.
• Volume: 2.7 million shares against a 30-day average of 2.2 million, 1.2 times normal. Notably unremarkable for a print that beat both lines by this much.
• Since the print: the ADR closed at $153.08 the following session and at $156.79 on August 10, so the reaction has held and modestly extended, leaving the stock still roughly 7% below its February high.

The 1.2 times volume multiple is the detail that reframes the day. A 9.5% core EPS beat and a return to top-line growth ahead of schedule, on barely more than normal volume and a move that faded 3.4% off the high, is not a market repositioning. It is a market marking a stock up for a good quarter it had partly anticipated and declining to re-rate it. That is a rational response to a print whose headline the company itself qualified twice in prepared remarks.

The more interesting comparison is with the February print. Then, the ADR rallied 2.0% on 2.8 times volume into a fresh 52-week high on a day the index fell, against a guide to declining core profit. Now, it rallies 2.9% on 1.2 times volume from 11% below the high on a day the index rose, against a first half that beat its own guidance. In February the market paid up for a company at the top of its range and told it to prove the trough would be shallow. It has now been shown a shallower trough and has responded by paying almost exactly the same multiple. The market's view of this business has not changed in six months. Its view of the price has.

Street Perspective

Debate: Does the second-half readout stack justify owning the stock into it?

Bull view: Six meaningful readouts land in six months against a stock 11% off its high at roughly 17x. A first-in-class cardiovascular outcomes result would create a new drug class and validate an entire Lp(a) franchise, an MS win would make the oral asset the most valuable in the company, and the DM1 and ITP readouts each carry blockbuster potential. Management has said explicitly that this stack would allow it to raise the medium-term outlook. Optionality this dense is rarely available at a market multiple.

Bear view: None of the three most important readouts has a pre-committed definition of success, management has declined to disclose the powering assumptions for two of them, the cardiovascular readout has slipped past its guided window for a second time, and the bar for calling it a win has been described in progressively lower terms. A dense catalyst calendar with undefined success criteria is not optionality, it is a series of events whose interpretation is controlled by the seller.

Our take: The bears have the better of this specific argument and it does not settle the stock. Everything the bear case says about the definition of success is factually correct and materially raises the odds of a marginal outcome being presented favourably. But the breadth argument survives it: six independent readouts do not need the marquee one to land, and the Avidity programmes have already converted faster than the deal case assumed. We value the portfolio and assign no expected value to any individual event, which is the same position we took in February and the one this quarter's disclosures reinforced rather than changed.

Debate: Is the growth portfolio now large enough to end the patent-cliff discussion?

Bull view: The six priority brands are 35% of quarterly sales, up from 30% two quarters ago, and grew fast enough to carry the company back to constant-currency growth while absorbing a fourteen-point generic drag. Scemblix is up 89%, Leqvio 59%, Kisqali 43% with US sales through $1B, and Pluvicto 43% with a label expansion approving in the third quarter that adds 75% to the eligible pool. The named LOE cohort is down to 10.4% of sales and nearly finished eroding. The cliff has been climbed.

Bear view: The cliff was never one event. Xolair has gone from -8% to -25% in two quarters and Tafinlar plus Mekinist has rolled over, neither of which is in anyone's LOE model, and together they are $923M a quarter. Cosentyx faces exclusivity issues inside the guidance window, and Kisqali, the brand carrying the whole growth story, now has a company-confirmed loss of exclusivity in the third quarter of 2031, one year past the end of the 2025 to 2030 CAGR window.

Our take: The bulls win on the near term more convincingly than they did in February, and the bear point has sharpened rather than weakened. Reaching 35% of sales while growing through the largest patent expiry in company history is a result rather than a forecast, and the Q3 approval in hormone-sensitive prostate cancer adds to it. But management's newly precise 2031 date for Kisqali means the guidance window now demonstrably ends immediately before the next cliff, and the widening erosion outside the named cohort suggests the base is decaying faster than the three-brand model implies. This is an argument about duration, and it argues for how long to own the stock rather than whether to.

Debate: Has the balance sheet stopped being an asset?

Bull view: Net debt at roughly 1.8x core operating income is unremarkable for an AA- credit, the ratings are unchanged at Aa3 and AA-, free cash flow still runs near $17B annually, and the money bought a pipeline that produced a filed accelerated-approval BLA within one quarter. Deploying a strong balance sheet into late-stage assets during a trough year is what a good capital allocator does.

Bear view: Net debt has nearly doubled in six months, free cash flow has now declined for two consecutive halves, the buyback that carried per-share growth has fallen from $8.9B a year to $2.1B in a half with only $5.6B left, and the company completed or agreed four further transactions in the same quarter. The flexibility that made this a defensive holding has been spent, and it was spent in the year earnings declined.

Our take: Both are right and the bears are describing the more relevant fact. Nothing here threatens the credit or the dividend, and the Avidity assets are converting faster than we assumed. But this stock's appeal was a high-quality compounder with a fortress balance sheet and a buyback that could bridge a soft year, and two of those three have changed. The per-share bridge from here has to come from operations, in a year when management has told you the margin architecture rests on a single SG&A lever. That is a narrower proposition than the one we underwrote in February, at the same multiple.

Model Update Needed

Valuation: at the $153.87 reaction-day close, NVS trades at 17.1x FY25 core EPS of $8.98 and roughly 17.0x to 17.4x our revised FY26 core EPS range. On the $156.79 close of August 10 the range is 17.3x to 17.7x. Market capitalisation at the reaction-day close is roughly $292B on the approximately 1,900M shares implied by reported core net income and core EPS. Against our FY26 free cash flow estimate of $16.0B to $16.5B, that is a free cash flow yield of 5.5% to 5.6%, down from the 5.9% we calculated at initiation on FY25 actuals.

Valuation impact: our estimate went up and our multiple did not. The stock is 11% off its February high and trades at the same roughly 17x it traded at then, on numbers that are now about 2% higher. That is the whole quarter in one sentence: the business got modestly better and the price got modestly cheaper, in equal measure. What has changed underneath is the composition of the risk. The earnings trough is shallower than guided, which is the good news and is now largely in the price. The balance sheet has taken on $17.5B of net debt, the buyback has been cut by more than half, and the largest binary in the story slipped again while its success threshold was publicly lowered. We see fair value close to the current price and would want either a de-risked cardiovascular readout or a mid-teens multiple before paying up.

Thesis Scorecard Post-Earnings

The pillars below are the standing thesis established at initiation and carried in our thesis of record, graded against this quarter rather than re-derived.

Grading last quarter's commitments

Overall: the operating thesis strengthened and the financial thesis weakened, by roughly equal amounts. Seven of the ten commitments management made in February were delivered, one partially, and two were not, with the two failures concentrated in exactly the area we identified as the least handicappable. The growth portfolio is executing better than we underwrote. The balance sheet, the buyback and the pipeline timing are all worse than we underwrote. Net of both, the thesis is unchanged and better understood.

Action: hold. One of the three upgrade triggers set at initiation has been met, in that the first half tracked materially better than the guided decline, and our estimate rises accordingly. The other two have not: the multiple is where it was, and the cardiovascular readout is riskier rather than de-risked. The downgrade trigger, deceleration in the six-brand growth portfolio, is nowhere in sight and moved emphatically the other way. For an investor with a position, this quarter is a reason to keep it. For one without, the second-half readout stack is a coin-flip we are not paid to take at 17x, and a third quarter that prints low-single-digit growth on a known base effect may offer a better entry than this one.