Four Wins and Three Misses Put the Pipeline on Its Own Assumed Base Rate; a Tax Credit Carries the EPS Beat

Results vs. Consensus

Q2 2026 Scorecard

Consensus figures are the mid-point of independently compiled Street estimates. Dispersion this quarter was 0.5% on the revenue line against 3.4% in Q1 2026, so the direction of both verdicts is unambiguous.

Year-Over-Year Comparisons

The reported-to-core bridge for the quarter is $100m of restructuring, $1,546m of intangible asset amortisation and impairments, and $348m of other acquisition-related items, totalling $1,994m at the operating line and $1.02 per share.

Quarter-Over-Quarter Comparisons

Revenue

Total Revenue of $15,384m grew 5% at constant currency and 6% on an actual basis, against a consensus that clustered tightly between $15.39bn and $15.47bn. The shortfall against the mid-point is 0.4%, which on a $15bn revenue line is noise rather than signal, but the composition deserves attention because the mix moved more than the total did.

Product Sales, the part of revenue that AstraZeneca books from its own commercial operations, grew 4% at constant currency. Alliance Revenue, the profit share taken from partners who record the sale themselves, grew 33% to $874m and now represents 6% of the total. That line carries essentially no associated cost of sales, so its $220m increase supplied roughly a quarter of the $927m revenue increase and, falling straight through, roughly a quarter of the $877m gross profit increase. Collaboration Revenue was nil, against $8m a year ago and $77m in the first quarter, so the milestone-driven revenue that flattered Q1 did not repeat.

Assessment: The underlying commercial engine grew 4%, the partner-booked profit shares grew 33%, and the reported total grew 5%. That is a slightly lower-quality revenue mix than the headline implies, and it is the second consecutive quarter in which the composition of the revenue line has been the more informative number than its level.

Margins

Core gross margin improved a point to 84%, and core operating margin expanded two points to 34%. Both are genuine, and both were achieved while the company absorbed the first full quarter of US Farxiga generic competition. Core total operating expense grew 4% at constant currency against revenue growth of 5%, so operating leverage was positive on a year-over-year basis for the second consecutive quarter.

The sequential picture is different and matters more for the second half. Core operating margin fell 1.5 points from Q1's 35.0% to 33.5%, because core operating expense rose $396m quarter-over-quarter while revenue rose $96m. That step-up is the Baxfendy launch and the tozorakimab pre-launch build, both of which management flagged and neither of which reverses. The CFO also told the call to expect a lower gross margin in the second half and higher core finance costs, which means the Q2 margin is close to the high-water mark for the year.

"While the margin improved in the second quarter compared to the first quarter, we expect lower gross margin in the second half, consistent with prior years, reflecting seasonal demand patterns for lower-margin medicines such as FluMist and Beyfortus."
— Aradhana Sarin, CFO

Assessment: Year-over-year leverage held and the operating-leverage pillar of our thesis survives the quarter intact. But the second-half margin is guided down on three separate lines simultaneously (gross margin, finance costs, and a launch cost base that has already stepped up), and a company that reconfirmed rather than raised its earnings guide at the half is telling you where it thinks the year lands.

EPS

Core EPS of $2.63 grew 18% at constant currency. The core tax rate was 15%, down six points from the prior-year quarter, on what the company describes in its own reconciliation table as a benefit from adjustments to deferred tax assets following certain internal legal entity changes. That is a balance-sheet event, not an operating one, and it is not repeatable.

"Our tax rate in the second quarter benefited from a onetime adjustment to deferred tax assets following certain internal legal entity changes."
— Aradhana Sarin, CFO

The symmetry with last quarter is worth stating plainly. In Q1 the core tax rate rose to 21% from 16% because the prior-year base carried a one-off release of tax liabilities, and core EPS growth of 5% understated an operating business compounding at 12%. This quarter the rate fell to 15% from 21%, and core EPS growth of 18% overstates an operating business compounding at 10%. Two consecutive quarters in which the tax line has been the largest single determinant of the headline earnings number.

Reported EPS of $1.61 grew 2% on an actual basis and declined 2% at constant currency, because the core adjustments widened to $1.02 per share from a materially smaller figure a year ago. Depreciation, amortisation and impairment rose 39% to $1,928m, and $345m of that is impairment charges recorded against products in development inside R&D expense.

Assessment: Neither the Q1 headline nor the Q2 headline described the business. Take the two quarters together and core operating profit grew 11% at constant currency in the first half against core EPS growth of 11%, which is the number to carry. The tax line will keep generating noise, and the company's own guided range of 18% to 22% against a first-half outturn of 18% tells you the second half normalises upward.

Segment Performance

Total Revenue by Therapy Area

Total Revenue by Region

Top Medicines by Product Revenue

Oncology

Oncology reached 48% of Total Revenue and grew 15% at constant currency, which is a point below the 16% it delivered in Q1 and still the whole reason the corporate algorithm survived the CVRM cliff. Imfinzi at $1,854m and 27% growth is now within $90m of Tagrisso and will overtake it. Calquence crossed $1bn in a single quarter for the first time, and the composition of that growth is unusually clean: the segment head attributed it to front-line CLL leadership against direct competition rather than to price or stocking.

"Turning to Calquence, which grew 16% in the quarter, generating more than $1 billion in revenue for the first time in a single quarter. Calquence maintains its position as the leading BTK inhibitor in frontline CLL across major markets despite intense competition."
— David Fredrickson, EVP Oncology Haematology Business Unit

Enhertu at $888m and 31% growth reflects the widening of the label into the curative setting, with simultaneous US approvals in DESTINY-Breast05 and DESTINY-Breast11 landing this quarter. Datroway at $55m against a low base is early. Lynparza declined 3%, which is the expected shape of a mature first-generation PARP inhibitor facing generic competition in China and destocking in the US. The two newly branded launches, Etcamah in ESR1-mutated breast cancer and the first sales of Truqap in prostate cancer following the CAPItello-281 approval, contributed almost nothing to the quarter and are 2027 stories.

Assessment: Oncology is doing exactly what the thesis requires of it, which is to be large enough and diversified enough that no single asset's deceleration threatens the corporate growth rate. The one thing to watch is that Tagrisso, the historic flagship, is now the slowest-growing of the top four oncology brands, and its succession plan runs through combination readouts that do not report until later this year.

Cardiovascular, Renal & Metabolism

CVRM fell 18% at constant currency to $2,772m and is now 18% of Total Revenue, down from a fifth. This is the quarter the loss-of-exclusivity cycle became fully visible. US Farxiga product sales fell 48% to $219m in the first quarter of multiple generic entrants, worldwide Farxiga fell 19%, Brilinta fell 63%, and roxadustat fell 82%. Nothing here is a surprise: every one of these events was flagged and dated in advance.

"As expected, generic competition for Farxiga entered the U.S. at the start of the quarter. And this, along with loss of exclusivity in some other markets and VBP in China saw Farxiga decline by 19% overall, resulting in $1.8 billion of revenue for the quarter."
— Ruud Dobber, EVP BioPharmaceuticals Business Unit

The offsets inside the segment are real but small. Lokelma grew 26% to $221m, Wainua grew 58% to $70m on the polyneuropathy indication, and Baxfendy, the newly branded baxdrostat, booked its first $3m following US approval in May.

Assessment: The decline is the plan working, not the plan failing, and the important observation is that a segment shedding 18% of its revenue while representing 18% of the company still left Total Revenue growing 5%. That said, Q2 is the first of four quarters in which US Farxiga will lap an uneroded base, so the segment's drag on the growth rate persists through mid-2027 regardless of what baxdrostat does.

Respiratory & Immunology

Respiratory grew 11% at constant currency to $2,431m and was the offset that let BioPharmaceuticals decline only 7% despite the CVRM collapse. The biologics did the work: Tezspire grew 45% to $390m, Fasenra 13% to $570m, Breztri 20% to $346m, and Saphnelo 24% to $209m. Symbicort declined 8% on a new US generic entrant, which is the expected end-of-life shape for the inhaled franchise's oldest asset.

The strategic point in this segment is the pre-launch spending rather than the current numbers. Tozorakimab has not launched, will not launch this year, and is already consuming SG&A.

Assessment: Respiratory is the segment where the second-half margin pressure will be most visible, because it is carrying launch investment for an asset whose pivotal data has still not been presented. That is the right investment to make if the data is what management says it is, and it is an uncomfortable one to be making on undisclosed effect sizes.

Rare Disease

Rare Disease grew 8% at constant currency to $2,490m, a deceleration from the 11% the segment delivered in the first half, and the internal composition explains it. Ultomiris grew 12% and Strensiq 36%, but Soliris fell 28% as the conversion to Ultomiris continues alongside biosimilar pressure. At $389m Soliris is still large enough that its decline absorbs a substantial share of the growth Ultomiris and Strensiq generated.

Strensiq at 36% growth is the number that matters for 2027, because it is the commercial infrastructure that efzimfotase alfa will inherit. Management has repeatedly framed efzimfotase at $3bn to $5bn peak, against a Strensiq franchise currently running near $2bn annualised.

Assessment: The segment is a net grower with a large declining asset inside it, which makes the reported 8% a poor guide to the underlying trajectory: the three growing brands compound at 12%, 36% and 27% respectively. The segment's 2027 depends almost entirely on the breadth of the efzimfotase label.

Infectious Disease

Infectious Disease fell 30% at constant currency to $131m, driven by Beyfortus down 37%. The segment is under 1% of revenue and the quarterly numbers are meaningless in isolation because both products are acutely seasonal. The one item worth flagging is that AstraZeneca has commenced an arbitration process regarding Beyfortus in the US and will not discuss it.

Assessment: Immaterial to the model, potentially material as a signal. An arbitration against a partner in a product this size is unusual enough to note and impossible to size from what has been disclosed.

Key Topics & Management Commentary

Overall Management Tone: Management was noticeably more defensive than in April, and the defensiveness was concentrated in one place: the framing of the pipeline as a risk-adjusted portfolio rather than a set of individual bets. The CEO opened with that framing before any analyst raised it, returned to it unprompted twice more, and closed the call with it, which is the posture of a team that expected the CARDIO-TTRansform question and decided to answer it pre-emptively. On the financial lines the CFO was precise and forthcoming on direction while declining every request to size the components, and on the pipeline the tone was unchanged and confident. The one exchange where management was visibly irritated rather than merely firm was on whether further business development is required to bridge the post-2032 patent cliff.

1. CARDIO-TTRansform: the first genuine test of the conversion thesis

The Phase III CARDIO-TTRansform trial of Wainua in transthyretin-mediated amyloid cardiomyopathy missed its primary endpoint. This was announced on 9 July, three weeks ahead of the print, and the ADR gapped 8.4% lower on the day and lost 13.1% cumulatively across four sessions. It is the largest single pipeline disappointment since we initiated coverage, and the way management handled it on this call is more informative than the result itself.

"In this contemporary patient population treated with standard of care, including 57% on a stabilizer, adding Wainua did not provide a statistically significant benefit on the composite outcome of CV mortality and recurrent CV events. However, in a prespecified subgroup analysis of patients treated with Wainua monotherapy as compared to placebo, fewer primary composite events were observed, and this result was nominally significant. In patients who were on stabilizer therapy at baseline, no treatment effect was observed."
— Sharon Barr, EVP BioPharmaceuticals R&D

The structure of the failure is specific and the disclosure was unusually complete. Fifty-seven percent of the enrolled population was on a stabiliser at baseline, and in that group there was no treatment effect at all. The nominally significant monotherapy subgroup is a scientifically interesting observation and a commercially near-worthless one, because contemporary practice puts most of these patients on a stabiliser. Full data goes to the European Society of Cardiology in August.

Assessment: The result kills the largest single expansion opportunity in the cardiovascular portfolio and it does so cleanly, which is preferable to an ambiguous outcome that would have consumed another two years of development spending. The thesis question is not whether this specific trial failed but whether the Q1 conversion rate was a signal or a sample. This quarter answers that: it was a sample.

2. The pipeline converted at the base rate, and management said so first

Seven registrational readouts were disclosed since the prior results announcement. Four hit: VOLGA and NILE for Imfinzi in bladder cancer, CLARITY-Gastric01 for sonesitatug vedotin in gastric cancer, and a paediatric myasthenia gravis study for Ultomiris. Three missed: EMERALD-2 for Imfinzi in adjuvant hepatocellular carcinoma, CARDIO-TTRansform, and TMA-313 for Ultomiris in adult transplant-associated thrombotic microangiopathy. That is a 57% hit rate against the four-for-four of Q1.

"the average probability of success across Phase III for the industry is 60%, 65%. I think Aradhana mentioned it earlier. The probability -- the aggregate probability of success that we use across our pipeline, and we looked at it again recently, it's around 60%. So we are more or less planning as if we were going to develop the industry average, in fact, a little bit lower."
— Pascal Soriot, CEO

Disclosing the aggregate probability-of-success assumption is genuinely useful and, as far as we can tell, not something the company has quantified this precisely before. It converts the $80bn 2030 ambition from an assertion into a model with a stated input. If the plan assumes 60% and the company has historically run at 75% and this quarter ran at 57%, the ambition has slack in it.

Assessment: This is the most important disclosure on the call and it cuts both ways. It defuses the "one failure breaks the thesis" argument, which is the right defence. It also removes the basis for extrapolating the Q1 conversion rate, which is what we did in April when we called the base rate "well above what the Street had underwritten." One quarter of four-for-four was not evidence of a structurally superior hit rate.

3. sone-ve: the wholly owned ADC portfolio delivers its first Phase III

CLARITY-Gastric01 met its overall survival dual primary endpoint for sonesitatug vedotin in previously treated Claudin 18.2-positive gastric and gastroesophageal junction cancer. This is the offset to CARDIO-TTRansform, and it is a structurally more valuable asset than the trial's second-line-plus indication implies.

"CLARITY-Gastric01 is the first Phase III trial to demonstrate an overall survival benefit with a Claudin 18.2 targeted antibody drug conjugate in the second-line plus setting."
— Susan Galbraith, EVP Oncology Haematology R&D

Two details make it more than a niche win. The expression threshold used was low, at 25% at any staining intensity, which widens the eligible population well beyond what competing Claudin 18.2 agents can claim. And the trial met a key secondary endpoint in patients treated with at least one prior line, which opens the second-line population rather than just third-line-plus.

"the survival data were demonstrated in patients with Claudin 18.2 expression as low as 25% at any staining intensity, a lower threshold than that required by other Claudin 18.2 targeted therapies. meaning that sone-ve could potentially benefit around 50% of patients with second-line plus cancer -- gastric cancer, representing more than 180,000 patients across the U.S., EU5, China and Japan."
— Susan Galbraith, EVP Oncology Haematology R&D

Progression-free survival showed a trend that did not reach statistical significance, which management disclosed unprompted. The asset is wholly owned, which is why the economics compare favourably to Wainua even at a similar headline peak-sales figure.

Assessment: A $3bn to $5bn peak framing on the first Phase III readout, before data presentation, is exactly the pattern that our Bear 3 pillar exists to track. But the ownership point is real: a wholly owned $4bn asset is worth materially more to shareholders than a partnered $4bn asset, and the gastric franchise now has three positive Phase III readouts in two years across three different mechanisms.

4. Tozorakimab moves to "more than $5 billion" without the data being shown

Management raised its peak-year revenue expectation for tozorakimab to more than $5bn, from the $3bn to $5bn range framed in April. The pivotal OBERON and TITANIA data still has not been presented and now goes to the European Respiratory Society Congress in September, one congress later than the April call implied.

"a good example of this is the strong data we obtained for tozorakimab, a product that very few people thought was going to work. We ourselves had a low probability of success for it. We tried because we thought we have a different mechanism of action, and we have a chance, and it actually worked. You will see the data very soon. As a result, we have increased our peak revenue expectation to more than $5 billion."
— Pascal Soriot, CEO

What has changed since April is regulatory rather than clinical: filings have been accepted in Europe and China. That is meaningful evidence that the package supports a broad label. It is not evidence about effect size, which remains the entire question, and it is the fourth consecutive quarter in which the peak-sales number has moved before the data has moved.

Assessment: Raising a peak-sales estimate on an asset whose pivotal effect sizes remain undisclosed, in the same quarter that a different asset's confirmatory trial failed, is a difficult combination to underwrite at face value. We carry tozorakimab at $2bn to $3bn and will revisit after September, unchanged from our April position.

5. Other operating income is not normalising, and that creates a 2027 problem

Other operating income was $152m in the quarter against $79m a year ago, and $341m in the half against $192m. In April we flagged this line, then running at $189m, as containing what the CFO called nonrecurring milestones, and said not to annualise it. The guidance now given is that the second half should be broadly similar to the first, which implies a full-year figure near $680m against $192m in the first half of 2025.

"Other operating income was $341 million in the first 6 months, consisting of royalties and small regional divestitures, and we anticipate a broadly similar level in the second half."
— Aradhana Sarin, CFO

Two separate analysts pressed on how much of this repeats in 2027, which tells you the Street has noticed. The answer offered was directional and unquantified: a portion is royalties and milestones that continue, a portion is divestiture gains from cleaning up the legacy portfolio, and there is no split. On our arithmetic this line supplied roughly half a point of the two-point core operating margin expansion in the quarter.

Assessment: A line that has more than doubled and is guided to stay there, with no disclosure of the recurring versus one-off split, is a real hole in the 2027 bridge. If the divestiture component is even a third of the total, 2027 core operating profit faces a headwind of roughly $200m before anything operational happens.

6. Grading Tagrisso: the destocking reversed, but not to the rate management quoted

Our April note set an explicit test. Management had disclosed mid-teens underlying US demand against a reported 5% constant-currency print, attributed the gap to destocking, and said it did not see it going further down. Q2 was the quarter that verdict came due.

US Tagrisso product sales rose to $845m from $733m, a 15% sequential increase, and year-over-year US growth accelerated to 11% from 8%. The destocking did unwind. But worldwide constant-currency growth moved only from 5% to 6%, because Europe decelerated from 12% to 4% over the same two quarters and Emerging Markets remain roughly flat against a more competitive Chinese EGFR market.

"we're seeing double-digit volume growth and, in fact, double-digit revenue growth in the U.S. in the face of the competition with Tagrisso right now."
— David Fredrickson, EVP Oncology Haematology Business Unit

Assessment: Half a pass. The destocking explanation was honest and the US number confirmed it, which removes the share-loss concern we were carrying. But double digits is not mid-teens, and the worldwide rate barely moved because a second region weakened as the first recovered. Tagrisso is now a mid-single-digit grower with three combination readouts as its succession plan, and the first of those does not report until later this year.

7. China deteriorated sharply within the half

China revenue of $1,587m fell 13% at constant currency in the quarter, against a first-half decline of 5%. That arithmetic locates essentially the entire deterioration in Q2. Volume-based procurement is the stated cause, with roxadustat down 82%, Farxiga hit by both generics and VBP, and Lynparza facing generic competition there.

Emerging Markets excluding China grew 11% at constant currency, which is the more important number and the one that gets lost when the region is reported in aggregate at 0%. The company's emerging-market growth engine is intact; China is a separate and worsening problem.

"China, still very bullish for reasons that have evolved, I must say. I mean the potential in China is still large. But on top of it, the innovation potential is also enormous, as you've seen from the various deals we've made, but also other companies have made."
— Pascal Soriot, CEO

The reframing in that answer is the substance. The bull case on China has shifted from "large and growing commercial market" to "source of licensable innovation," which is a materially different proposition for a business that books 10% of revenue there. Two of the three business development transactions announced this year are with Chinese counterparties.

Assessment: A 13% constant-currency decline in a tenth of the company, with the CEO's stated rationale for optimism having migrated from the commercial opportunity to the innovation supply chain, is a genuine change worth marking. It does not break the algorithm, because Emerging Markets ex-China at 11% offsets it almost exactly and the region printed flat in aggregate. But that offset is doing more work each quarter.

8. Baxfendy launched, and reimbursement is a 2028 event

Baxdrostat was approved in the US in May, branded Baxfendy, and launched. This was the Q2 regulatory decision we were watching. It booked $3m. The disclosure that matters is the access timeline.

"In May, we secured the approval and launch of Baxfendy in the U.S., and we are now building early market access through affordability programs. We anticipate commercial access for Baxfendy will broaden over the next few quarters in anticipation of Medicare Part D reimbursement from 2028."
— Ruud Dobber, EVP BioPharmaceuticals Business Unit

Hypertension is overwhelmingly an older-patient indication, which makes Medicare Part D the dominant channel, and it does not open until 2028. Meanwhile a Phase III in primary aldosteronism is recruiting rapidly, with management arguing that the existence of an approved aldosterone synthase inhibitor will itself drive screening rates that do not exist today.

Assessment: The approval de-risks the asset and the revenue is a 2028 story. Anyone carrying meaningful Baxfendy revenue in 2027 should look again at the reimbursement timeline. The primary aldosteronism indication is the more interesting long-term option and it requires building a diagnostic market from close to nothing.

9. Business development is accelerating while the balance sheet absorbs it

The CSPC collaboration closed on 14 April with a $1.2bn upfront, of which $1.1bn was capitalised as an intangible asset, with up to $3.5bn of contingent consideration on regulatory milestones. In July the company added Zegfrovy from Dizal for $600m upfront plus up to $900m in milestones, and an inhaled PDE3/4 inhibitor from Sino Biopharmaceutical's CTTQ subsidiary for $200m upfront plus up to $1.9bn.

"We have announced new BD transactions totaling just over USD 2 billion in upfront payments year-to-date, including the most recently announced Dizal transaction."
— Aradhana Sarin, CFO

Set against that, the CEO's answer on whether more business development is needed for the post-2032 period was an unqualified no, delivered with visible impatience. Both things are true simultaneously: the company does not think it needs deals to hit its targets, and it is doing roughly one a month.

Assessment: The deals themselves look sensible and cheap relative to the assets acquired, particularly Zegfrovy, which is already approved in two markets and directly plugs the exon 20 gap in the EGFR franchise. The tension is between the strategic message and the spending pattern, and the spending pattern is more informative.

10. Cash conversion deteriorated materially

Net cash inflow from operating activities was $6,224m in the half against $7,099m a year ago, a 12% decline, while reported operating profit grew 3%. Three things drove it: a $1,438m working capital outflow against $771m, cash tax paid of $2,051m against $1,549m, and the absence of the prior-year Lynparza milestone.

Cash tax paid represented 31% of reported pre-tax profit against 24% a year ago, which sits oddly beside a reported tax rate of 17% for the half and is worth holding in mind when reading the Q2 deferred tax benefit. Net debt rose $3,538m in six months to $26,912m. Capital expenditure of $1,513m is guided to rise roughly a third for the full year, and milestone payments related to prior transactions are guided at approximately $2.5bn.

"We expect these working capital effects to persist through the remainder of the year before normalizing."
— Aradhana Sarin, CFO

Assessment: This is the pillar that moved most negatively this quarter. A company whose accounting earnings grow double digits while operating cash flow falls 12%, whose cash tax rate is nearly twice its book rate, and whose net debt is rising $3.5bn per half against rising capex and a monthly deal cadence, is one where the reported earnings quality is drifting away from the cash. None of the individual items is alarming; the direction of all of them together is.

11. Elecoglipron enters Phase III on the data management showed in June

The obesity and type 2 diabetes Phase IIb data we flagged as an April commitment was presented at the American Diabetes Association meeting in June, and the Phase III programme has started dosing.

"In VISTA, we observed a clinically meaningful and statistically significant weight loss of up to 11.8% at week 36, importantly, without evidence of a plateau. In SOLSTICE, there was an up to 1.9% reduction in HbA1c at week 26 with the vast majority of patients with type 2 diabetes reaching their glycemic goals."
— Sharon Barr, EVP BioPharmaceuticals R&D

The Phase III commitment is large: EMBOLD in obesity, five trials under the ELUMINATE programme in type 2 diabetes, and a planned ELEVATE outcomes programme in heart failure with preserved ejection fraction and chronic kidney disease. A separate amylin receptor agonist read out Phase II this quarter and is going into a Phase III monotherapy trial.

Assessment: The commitment was met and the data is respectable for an oral small molecule, without being competitive with injectable incretins on weight loss. The strategic logic is the combinability that only a small molecule offers, and the company is now building fixed-dose combinations with dapagliflozin and with its oral PCSK9. That is a differentiated position and an expensive one, and it is a meaningful part of why R&D is guided to the upper end of the low-20s as a percentage of revenue.

12. The $80bn ambition, restated as a portfolio claim

The CEO front-ran the CARDIO-TTRansform question in his opening remarks, before any analyst had spoken, with the framing that carried the entire call.

"when we set our $80 billion revenue ambition for 2030, we did so based on the strength of a broad and diversified portfolio, not on a single program. As you know very well, the $80 billion is a risk-adjusted forecast. If everything worked, we would be above the $80 billion. So we have, of course, expected setbacks to happen."
— Pascal Soriot, CEO

Twenty-five key Phase III readouts are planned over the next eighteen months, with pivotal data for six new molecular entities in 2027 alone. Eleven new molecular entities have been approved against the stated target of twenty by 2030.

Assessment: The framing is correct and it is also the only framing available after a quarter like this one. Eleven of twenty approved NMEs, with the target set for 2030, is on pace. The honest read is that the 2030 ambition remains achievable and that the evidence for it got no stronger this quarter.

Guidance & Outlook

Implied second-half ramp: First-half Total Revenue grew 6% at constant currency, so a mid-to-high single-digit full year requires the second half to grow at broadly the same rate. That is a harder ask than the first half was, because US Farxiga contributed a full uneroded quarter to Q1 and contributes nothing comparable to either second-half quarter, and because China is now declining at 13%. The offsetting arithmetic is that Brilinta and roxadustat have already collapsed to immaterial levels and cannot subtract much more.

Core EPS: First-half core EPS of $5.21 grew 11% at constant currency, at the low end of the guided low-double-digit range. The second half faces a lower gross margin, higher finance costs, a stepped-up launch cost base, and a core tax rate that must rise from the first half's 18% to land inside the guided 18% to 22%. First-half core EPS is seasonally the larger half for this company, and the guide implies the second half grows at a similar rate off a smaller base rather than accelerating.

Street at: Consensus for full-year core EPS sits near $10.21, which is consistent with the low-double-digit guide and implies the Street has not extrapolated the Q2 tax benefit.

Guidance style: House style is to set the algorithm in February and move it at the half. That the half passed without a raise, in a year where the first half delivered 6% revenue growth and 11% core EPS growth against guidance of mid-to-high single digit and low double digit respectively, is the single most informative thing about the guidance section. Management had the room to raise and chose not to.

Grading the Commitments from Last Quarter

Our April note closed with an explicit checklist of what the second quarter had to show. Scoring it is the point of writing the checklist.

Assessment: Four delivered, two partial, four not delivered. The two that matter most are the tozorakimab data deferral, because the peak-sales number moved again without it, and the absence of a guidance raise, because we said in April that a raise would be the confirmation of the thesis and its absence would be the risk case. Both argue for lower conviction rather than a different rating.

Analyst Q&A Highlights

Whether the 2026 readout cohort still carries its risk-adjusted peak-sales frame

The most precise question on the call went to whether the aggregate value of this year's readout cohort survived the failures. Management had put a figure of more than $10bn of combined risk-adjusted peak sales on the 2026 cohort at the full-year results in February, and the exchange forced an explicit reconciliation of the year's wins against its losses.

Q: "So at the start of the year, you highlighted over $10 billion combined and risk-adjusted peak sales from the 2026 readouts. And today, we see a couple of increases with tozo and then sone-ve and also hearing consistent messages from Susan on SERENA-4 and AVANZAR kind of balanced and consistent. So does that guidance still stand? Or have there been some risk adjustment changes maybe that were negative for the cohort?"
— Luisa Hector, Berenberg

A: "So the $10 billion was the peak year estimates with the risk unwind. So what that means is as risk unwinds for the 2026 cohort, we were still -- we would say $10 billion. Now some of them have unwound in a positive manner like tozo. Some of them have unwound in not so positive manner like Wainua. But I think, again, all of those being probability adjusted, I still think we'll unwind close to $10 billion. Now that $10 billion is not a 2030 number. That's a peak year sales number. And the peak for some of these products may hit beyond 2030."
— Aradhana Sarin, CFO

Assessment: A straight answer to a difficult question, and the caveat at the end is the part to keep. "Close to $10 billion" rather than "$10 billion" is a small downgrade delivered honestly. The clarification that this is peak-year revenue and that peak may fall beyond 2030 is a reminder that the 2026 readout cohort and the 2030 ambition are less tightly coupled than the catalyst slides imply.

Whether further business development is needed to bridge the post-2032 cliff

The sharpest pushback of the call was not about this quarter at all. It was the argument that the multiple compression the shares have suffered reflects the 2032-plus patent expiry cohort rather than any doubt about 2030, and that the current Phase III pipeline is insufficient to bridge it. It drew the only visibly irritated answer of the ninety minutes.

Q: "I just want to touch on BD. More specifically, is the current Phase III pipeline sufficient to deliver a stable profile during the main patent expiration period in 2032 plus. It seems to us that the current pressure on the multiple has much less to do with 2030 and much more to do with 2032 and beyond and that larger BD may be necessary to drive growth in that time frame."
— Zachary Dunn, Guggenheim

A: "Well, I think on this one, I can be very clear. The answer is no, we don't need more BD to deliver. But of course, to deliver this growth post 2030. But again, it assumes that the aggregate probability of success across our pipeline of new products will be at least as we plan it, because, again, there's no 1 or 2 projects. It's the aggregate view of the pipeline."
— Pascal Soriot, CEO

Assessment: The answer contains its own condition. "We don't need more BD" is true if and only if the aggregate probability of success lands at or above plan, which is the same 60% assumption disclosed elsewhere on the call, and which this quarter's cohort ran below. The dismissive coda about people who have not analysed the pipeline in detail was the wrong register for a question that correctly identified where the multiple pressure is coming from. On the evidence of the deal cadence, the company is hedging the answer it gave.

How much of the elevated other operating income repeats

Two separate analysts came back to the same line, which is the reliable indicator that the disclosure was insufficient the first time. The concern is not this year's earnings but next year's base: a line that has roughly doubled and is guided to stay doubled becomes a headwind the moment it stops.

Q: "Maybe if I could just go back to Aradhana on the other operating income question, so we have that covered. How much of that is sustainable going forward? How much does the higher run rate cause and a higher base that makes '27 a little bit more challenging?"
— Michael Leuchten, Jefferies

A: "So again, we don't provide the split of other income, but there is a portion of that, obviously, that is relating to royalties and some milestones that we get. And then there's a portion for this year that also relates to, like I said, smaller regional divestitures we have done, again, as we clean up the portfolio. Again, all of these are small legacy products that we continue to sell and monetize those. But there is a portion of that, that is consistent and will likely continue into 2027."
— Aradhana Sarin, CFO

Assessment: Three uses of "a portion" and no numbers. The answer confirms that some of the increase is divestiture gains, which by construction do not recur once the legacy portfolio is cleaned up, and declines to say how much. This is now the least well-disclosed material line in the P&L and it will be a debated item in the 2027 guide.

Whether first-half cost growth is the right proxy for 2027

The most useful modelling question of the call asked directly whether the 6% first-half growth in research and selling costs is the run-rate to carry forward, or whether the launch build accelerates it. The answer deferred entirely to the budgeting cycle.

Q: "So as we think about 2H '26 and '27 cost growth, is the 1H R&D SG&A cost growth of 6%, a good proxy? Or should we think about acceleration into second half of this year and into next year?"
— Sachin Jain, Bank of America

A: "So I think for 2027, we will obviously give guidance for 2027 at the beginning of 2027. We will start our sort of annual budgeting process later this year. So we start that in a few months. And as you know, there's a lot more readouts still to come and investments still to make, whether it's on tozo or we'll see how some of the other events read out. So all of that is going to be part of our planning."
— Aradhana Sarin, CFO

Assessment: A non-answer, but an informative one. Naming tozorakimab specifically as an investment still to be sized, alongside the observation that further readouts will ungate further spending, is a signal that the 2027 cost base is not yet settled and that the direction of the uncertainty is upward. The sequential cost step-up already visible in Q2 supports that reading.

Whether the efzimfotase filing reaches the adult population

The question we flagged in April as determining whether roughly 80% or 20% of the addressable hypophosphatasia population is reachable came back on this call and got a more complete answer than last quarter, though not a complete one.

Q: "So I had one on efzimfotase alfa, you said your filing for a broad label. Can you confirm that, that does include adults? And was that based on any discussion with regulators to date is the acceptability of the HICKORY trial and the trend benefit in the pediatric onset adult population. And does the $3 billion to $5 billion sales outlook include the adult population? Or is that just in pediatrics?"
— Graham Parry, Citi

A: "So the broad population, so the filings will be above 2 years of age for a broad population. As I said in my prepared remarks, this has been done in discussion with regulatory authorities. But obviously, they will need to look at the detail of the 3 studies that we have completed and submitted. To your second question on the -- so the adult population is segmented into 2 parts, adult with pediatric onset, which is in several countries already obtained with Strensiq and adults with adult onset where Strensiq is not approved today."
— Marc Dunoyer, CEO Alexion, AstraZeneca Rare Disease

Assessment: The filing is for above two years of age, which is broad, and management added separately that the $3bn to $5bn range holds even on less than the full label filed for. That is a meaningfully better answer than April's, because it decouples the peak-sales frame from the label outcome. The unanswered half is whether any regulator has indicated it will accept the paediatric-onset adult data, which is where the incremental population sits.

Competitive response to the IL-33 data in COPD

A question on whether a competitor would abandon its own IL-33 programme in light of the AstraZeneca data drew a candid refusal to speculate, and an unprompted defence of the mechanism.

Q: "Two quick ones, Pascal, one for Ruud on IL-33 and $5 billion target for tozo . Are you anticipating any competition here?"
— Peter Verdult, BNP

A: "Now I think, Peter, it's a fair question, but it's so difficult for me to answer this one because I simply don't know what Roche or Sanofi are planning to do. What I can tell you is that the data set, hopefully, you will see at the ERS is very convincing. And we get many questions about why do we think that our IL-33 was successful in those 2 pivotal trials. And we truly believe that the mechanism of action is very specific to our anti-IL-33."
— Ruud Dobber, EVP BioPharmaceuticals Business Unit

Assessment: "Hopefully" and "we truly believe" are hedging words attached to a peak-sales estimate that was raised above $5bn on the same call. The admission that management is fielding many questions about why its IL-33 worked when the class has struggled is the tell: the scientific community has not yet accepted the result, and it cannot until September.

Most-favoured-nation pricing and the new launch playbook

The obligatory pricing-policy question produced the most substantive answer on the topic we have heard from this management team, largely because it was answered from the commercial side rather than the policy side.

Q: "And then secondly, the obligatory sort of MFN pricing question. When you think about cami approved in Europe, hopefully will get approved in the U.S., how have your thoughts on pricing strategy evolved if we compare it to historical precedence like Calquence and Tagrisso."
— Peter Verdult, BNP

A: "That said, our pricing approach across all of launches now has really evolved in response to MFN. And with the U.S. now referencing a basket of countries and setting its price, we are seeing productive engagement with payers in the wealthy nations within that context. With Etcamah specifically, we're still early in the commercial life cycle in the markets where we have approvals, but the early negotiations are reflecting this new reality that funding for innovation needs to rise in line with the country's GDP per capita."
— David Fredrickson, EVP Oncology Haematology Business Unit

Assessment: The useful disclosure is that the pricing approach for every new launch has already changed, and that ex-US negotiations are now running longer. That is a working-capital and launch-curve effect that shows up before any headline price change does. The stated direction of travel, ex-US prices up and US prices down somewhat, is the outcome the sector would choose if given the option, which is a reason to treat it as an aspiration rather than a forecast.

What They're NOT Saying

1. The recurring versus one-off split of other operating income. Asked twice, declined twice. This line has roughly doubled year over year, is guided to stay at the new level through the second half, and by management's own description contains divestiture gains that by construction cannot recur. It is the largest undisclosed item in the 2027 bridge.
2. Any sizing of the deferred tax adjustment. The company disclosed the rate movement and the cause but not the dollar amount, and did not address it in prepared remarks beyond a single sentence. It is the entire difference between an earnings beat and an earnings miss this quarter.
3. Tozorakimab effect sizes, for the second consecutive quarter. April deferred to a medical meeting; July deferred to a later medical meeting while raising the peak-sales estimate. The commercial claim has now moved twice without the clinical evidence moving once.
4. Whether China's Q2 decline is the trough. The 13% constant-currency decline was attributed to volume-based procurement, with no indication of which molecules enter the next round, no sizing of the 2027 exposure, and no framing of when the national reimbursement drug list additions offset it.
5. Second-half revenue phasing. Guidance is a full-year range and the second half now carries a full uneroded Farxiga comparison, a deteriorating China, and the seasonal respiratory products. No quarterly shape was offered.
6. The Beyfortus arbitration. Confirmed to exist, referred to the legal disclosures, and closed off. An arbitration involving a partnered product is not routine and the potential outcomes are entirely unsized.
7. Whether EMERALD-2 changes the adjuvant liver-cancer strategy. The trial's failure appears in the milestones table and was not mentioned once in prepared remarks or Q&A. A missed Phase III in a major Imfinzi expansion passing without comment on an earnings call is a disclosure choice.
8. What the 2027 cost base looks like. The direct question was deferred to the budgeting cycle, with the caveat that tozorakimab investment is still to be sized and further readouts will ungate more spending. The uncertainty is acknowledged and its direction is upward.
9. The gap between the 17% book tax rate and the 31% cash tax rate. Both figures are disclosed in the same document and the divergence went unmentioned and unasked.

Market Reaction

• Pre-print setup: The ADR closed at $169.26 on 24 July, down 3.6% year to date against the S&P 500's gain of 8.3%, up 16.5% over the trailing twelve months, and down 10.2% over the trailing thirty days. The 52-week closing range entering the print was $141.77 to $209.48, placing the shares roughly 19% below the high.
• Print-day session: Results were released at 07:00 UK time, before the US open, with the call at 11:45 UK time. The ADR opened at $170.08, a 0.5% gap higher, traded a $167.79 to $173.24 range, and closed at $169.64, up 0.2% or $0.38.
• Volume: 3.9 million shares against a 2.4 million thirty-day average, 1.6 times normal.
• Market context: The S&P 500 was unchanged on the session, so the move was idiosyncratic rather than market-driven.

The trailing thirty-day figure is the number that frames the whole session. Almost all of that 10.2% decline happened in four sessions in July following the CARDIO-TTRansform announcement: the ADR closed at $189.28 on 8 July, gapped to a $173.45 open on 9 July, printed an intraday low of $172.76, and closed at $164.50 on 14 July, a cumulative decline of 13.1%. It had recovered to $169.26 by the pre-print close. The market had already taken its pipeline markdown three weeks before the print, which is why a quarter containing a revenue miss produced no reaction at all.

The shape of the print-day session is a fade rather than a rally. The ADR gapped modestly higher on the headline earnings beat, reached $173.24 at the high, and gave nearly all of it back to close up two tenths of a percent. That is consistent with an initial read of the $2.63 core EPS figure followed by a closer read of what produced it. Anyone who worked through the tax reconciliation in the first hour arrived at a number below consensus, and the session's path suggests plenty of people did.

The most useful comparison is with April. That print delivered four positive Phase III programmes and the stock fell 0.8% on 2.4 times normal volume. This print delivered a revenue miss, a tax-driven earnings beat, and a mixed pipeline quarter, and the stock rose 0.2% on 1.6 times normal volume. The market's reaction function to this name is currently insensitive to the quarter itself, which is what happens when a stock is being priced on a policy overhang and a 2032 patent cliff rather than on its current earnings power.

Street Perspective

Debate: Does the CARDIO-TTRansform failure damage the 2030 ambition?

Bull view: The $80bn target is explicitly risk-adjusted at an aggregate probability of success near 60%, and the company has historically converted above 75%. One failure inside a cohort that also produced the first Phase III win from the wholly owned ADC portfolio does not move a portfolio-level number, and the asset lost was partnered while the asset gained is not.

Bear view: The bull argument works only if the aggregate hit rate holds, and this quarter it was 57%. The company assigned Wainua an industry-standard probability of success and tozorakimab a low one, which means the plan systematically underwrites the wrong assets. Meanwhile the shares are being marked on the 2032 cliff, where the pipeline is thinner and management insists no additional business development is needed.

Our take: The bull case is right about the arithmetic and the bear case is right about the epistemics. A single quarter at 57% tells you nothing about a long-run rate, and neither did the previous quarter at 100%, which is precisely the point: we over-read the April cohort as evidence of structural superiority and are correcting that here. The 2030 ambition is undamaged. What changed is the confidence interval around it, and that belongs in the multiple rather than in the estimates.

Debate: Is the earnings beat real?

Bull view: Core operating profit grew 10% at constant currency with the operating margin expanding two points, in the quarter that absorbed the first full impact of US Farxiga generic entry. The tax rate is volatile in both directions, as the Q1 headwind demonstrated, and looking through it in one quarter while treating it as decisive in the next is inconsistent.

Bear view: Eight of the eighteen points of core EPS growth came from a tax adjustment that management itself calls one-time. Adjust for it and the quarter missed on both lines. Reported operating profit fell 13% at constant currency, operating cash flow fell 12%, and the cash tax rate is nearly double the book rate.

Our take: Both quarters' headline EPS were tax artefacts and the honest read is the half: core operating profit and core EPS both grew 11% at constant currency in the first half. That is a good number, delivered through the worst of the loss-of-exclusivity cycle, and it is neither the 18% the Q2 headline advertised nor the 5% the Q1 headline implied. The bear point that lands is the cash statement, not the tax line.

Debate: Is the de-rating an opportunity or a re-rating?

Bull view: The shares trade near 16.6 times the mid-point of our full-year core EPS estimate, down from 18.2 times at the April print, on unchanged guidance and a first half that tracked it. A large-cap pharmaceutical business guiding low-double-digit earnings growth with twenty-five Phase III readouts in the next eighteen months should not trade at a discount to the market multiple.

Bear view: The de-rating is the market correctly repricing terminal value. Most-favoured-nation pricing is unresolved, the 2032 patent cohort is large, management denies needing further business development while announcing a deal a month, and the cash conversion is deteriorating. Cheap on next year's earnings is not cheap if the terminal margin structure is uncertain.

Our take: The bear case has improved since April and it still describes a sector risk rather than a company-specific one. What has genuinely changed is that we can no longer argue the multiple embeds none of the pipeline while also arguing the pipeline converts above the industry rate, because this quarter it did not. We resolve that by lowering the multiple we are willing to apply rather than by changing the rating, because at 16.6 times forward core earnings the shares now pay for less of the in-market business than they did in April, let alone the pipeline.

Debate: Does the guidance reaffirmation signal a soft second half?

Bull view: The company sets its algorithm in February and moves it when it has to. A first half at the low end of the core EPS guide, with the largest loss-of-exclusivity event only one quarter old and three named second-half headwinds on gross margin, finance costs and tax normalisation, is not a setup where a prudent management team raises. Reaffirmation is discipline.

Bear view: Six percent first-half revenue growth against a mid-to-high single-digit guide and 11% core EPS growth against a low-double-digit guide is a first half tracking at the bottom of both ranges, in the half that had the easier comparison. The second half has a full uneroded Farxiga base to lap and a China business declining at 13%.

Our take: The bear reading is the more likely one and it is already visible in the guidance detail rather than hidden in it. Management named the second-half gross margin decline, the higher finance costs and the elevated cost base without being asked. A company reconfirming a range whose components it has just guided lower is signalling the lower half of that range. We model the full year at the bottom of the guided band.

Model Update & Valuation Framework

Valuation. At the 27 July close of $169.64, the shares trade at roughly 16.6 times our FY2026E core EPS of $10.20 and 15.2 times our FY2027E of $11.20. That is a full turn and a half cheaper than at our April initiation, on estimates that are essentially unchanged. On statutory earnings the screen looks far worse, because intangible amortisation and impairments ran at $0.79 per share in this quarter alone and the total core adjustment was $1.02 per share. Investors screening this business on reported earnings are looking at a number that understates the cash-relevant one by nearly 40%.

Price target: $200, lowered from $215. That is roughly 17.9 times our FY2027E core EPS of $11.20 and implies about 18% upside from the $169.64 close. The multiple cut from 19 times is deliberate and specific: in April we justified a premium multiple partly on a conversion rate that had just run four for four, and this quarter's four for seven says that rate should be marked at the company's own stated aggregate assumption rather than at the April observation. The estimates are unchanged; the multiple carries the entire adjustment.

Bear case, $168. Fifteen times FY2027E, roughly flat with the current price. That is the case where the second half lands at the bottom of the guided band, the other operating income base becomes a 2027 headwind, and the policy overhang persists without resolution.

Bull case, $232. Twenty times an FY2027E lifted toward $11.60 by a tozorakimab launch on convincing September data and early sone-ve regulatory momentum. Roughly 37% above the current price.

What would change our mind. A tozorakimab presentation in September showing effect size concentrated in high-eosinophil patients would take us to Hold, unchanged from April. A second consecutive quarter of operating cash flow declining while accounting earnings grow would do the same, because that is the pattern that precedes a balance sheet becoming the story. On the other side, a disclosed recurring share of other operating income above two thirds, or a second-half guidance raise at the nine-month mark, would restore the conviction we are taking down here.

Thesis Scorecard Post-Earnings

Overall: Weakened, but at the margin rather than structurally. Three of four bull pillars were confirmed and the fourth was the one we over-read in April. Two bear points escalated, and neither is about the operating business: one is about the credibility of forward claims and one is about cash. The core proposition, that this is a pipeline-conversion story trading at a multiple that pays for the in-market business and very little of the pipeline, survives, and the multiple has moved further in its favour than the evidence has moved against it.

Action: Hold the position and add on weakness, but stop treating the pipeline as an above-average converter. The single most important date now is the September respiratory congress, where the tozorakimab effect sizes finally become public. That presentation resolves the largest unpriced item in the story in one direction or the other, and it is the reason we would rather carry a reduced position through it than a full one.