A Settlement-Free Beat, and a £1.9bn Programme That Turns the Patent Cliff From Flat to Rising

Results vs. Consensus

Q2 2026 Scorecard

GSK reports in sterling on a Core (non-IFRS) basis and lists in the US as an ADR where one ADS represents two ordinary shares. Both consensus bases are shown, because the quarter's coverage produced at least one widely syndicated comparison that mixed them.

The two EPS beats agree almost exactly, at +7.2% in sterling and +7.1% per ADS, because the translation is mechanical: 50.5p at two ordinary shares per ADS and the quarter's $1.34 average rate produces $1.35, against $1.36 reported. The two revenue beats diverge because the dollar estimate carries a different currency assumption than the quarter actually delivered. Sterling is the honest basis for this company, and on that basis the top line beat by two points, not four.

Year-Over-Year Comparison (Core basis, £m)

Prior-year comparatives are read from the Q2 2025 results announcement and tied back to the growth percentages printed in this quarter's release. Growth rates in the AER and CER columns are as printed by the company; the gross-profit growth figure is computed from the two sterling amounts because the release does not print a Core gross profit growth rate.

Quarter-Over-Quarter Comparison

Q1 2026 amounts are the year-to-date figures in this release less the Q2 column, and each of the four that GSK separately published in April (turnover, Core operating profit, and the three product groups) ties exactly. Core EPS is taken from the published quarterly figures rather than by subtraction, because each quarter rounds independently and the two quarters sum to 97.0p against a printed half-year of 97.1p.

Segment Performance

Product Group Turnover

Regional Turnover

The regional split contains the sharpest number in the release. US General Medicines fell 17% CER in the quarter and 12% year to date, against a group US line that still grew 5%. The US is simultaneously GSK's best Specialty market, at +15%, and the epicentre of its established-portfolio decline. Europe is the mirror image: Vaccines up 13% on Shingrix demand in the Nordics and Austria and Bexsero outbreak demand in the UK, with General Medicines merely stable.

Individual Product Detail

Specialty Medicines: £3,782m, +14% CER

Specialty crossed 45% of group turnover, from 41.7% a year ago and 42.3% last quarter. In absolute terms it added £455m of sales year on year while General Medicines subtracted £225m, so the growing half of the company is now covering the shrinking half twice over. That ratio was roughly 1.25 times in Q1. The mix shift is running ahead of the roughly three points a year we assumed at initiation.

Inside Specialty, the composition improved as well. Respiratory, Immunology and Inflammation grew 19% CER, the fastest of the three sub-groups, on Nucala's COPD launch and Benlysta's bio-penetration gains. Oncology grew 17% CER despite Zejula falling 34%. Excluding Zejula from both years, the rest of Oncology grew from £333m to £468m, or about 40%. Jemperli, Ojjaara and Blenrep together added £133m year on year against Zejula's £50m decline.

"So commercial momentum continued in the second quarter, driven by key products across our specialty and vaccines portfolio. Growth was driven by Specialty Medicines, which grew 14% and vaccines, which grew 8%." — Nina Mojas, Chief Commercial Officer

Assessment: This is the pillar working exactly as designed, and faster than modelled. The one caution is that Nucala's 23% CER included 12 percentage points of US channel-mix pricing adjustments, which the company itself flags. Strip that and Nucala grew about 11%, still respectable for a product in its COPD launch year, but the headline overstates the underlying volume story by roughly half.

Vaccines: £2,284m, +8% CER

The comp wall we flagged at initiation arrived on schedule, and the segment beat it anyway. Shingrix decelerated from about 21% CER in Q1 to 3% in Q2 as European and Japanese programme volumes annualised, with US sales described as broadly stable. What carried the segment was Meningitis at +21% on Bexsero outbreak demand in the UK and International, and Arexvy more than doubling on an Australian tender win. Neither is a run-rate.

The release is unusually explicit about how much of the growth was not demand:

"Growth in Q2 benefitted 3ppts from prior period rebate adjustments." — GSK Q2 2026 results announcement, Vaccines commentary

The same three points appear again at product level: "favourable pricing including prior period rebate adjustments which added 3ppts to Shingrix Q2 growth." Shingrix therefore grew roughly zero on an underlying basis, and Vaccines roughly 5%. Against that, GSK raised the Vaccines guide from a low-single-digit decline-to-stable range up to stable-to-low-single-digit growth.

Assessment: The guide raise is justified by the year-to-date position, not by the quarter's underlying run rate. Vaccines is up 6% CER year to date with a much easier H2 comparator on Arexvy, so the raise will very likely be delivered. But an investor reading "Vaccines +8%" and inferring that the Shingrix comp problem was overstated would be reading the wrong number. Shingrix at flat, with the US cumulative immunisation rate at 45% and the ten priority ex-US markets averaging around 12%, is a franchise whose growth is now an international-expansion project rather than a US one.

General Medicines: £2,343m, -9% CER

This is the quarter's clear negative and the one place where the guide moved down. General Medicines fell 9% CER against 6% in Q1, and the full-year outlook was cut from a low-single-digit decline-to-stable range to a mid-single-digit to low-single-digit decline. Trelegy fell 7%, of which five percentage points were channel-mix pricing adjustments, so the underlying decline was closer to 2%. The rest of the segment was worse: Respiratory ex-Trelegy and Other General Medicines both fell on generic erosion and competitive pressure, and Other General Medicines also carried a reduction in contract manufacturing sales.

"So look, this is a portfolio of mature, older assets that continue to be in many regions under pressure, under pricing pressure and generic medicines coming on board. That's one element. The other element is Trelegy. And I think you have seen -- we spoke about this in the first quarter already. We do see increased abandonment rates in the U.S. for Trelegy and Trelegy is not the only one." — Nina Mojas, Chief Commercial Officer

Management's second-half case is specific rather than hopeful: abandonment rates are falling from an elevated level, and Q2 carried a prior-year rebate and returns true-up on Trelegy that does not repeat.

"So in the second half, we're anticipating there'll be less abandonment going on, and we haven't got a tough comp like we had in Q2 because of the rebate and return adjustment." — Julie Brown, Chief Financial Officer

Assessment: At initiation we said a Q2 print with General Medicines still at -6% and no second-half recovery visible would be a downgrade trigger. It printed worse than that, at -9%, and the guide was cut. The trigger is half-hit: the deterioration is real, but the second-half path is articulated with named mechanics rather than asserted. What saves the group number is that General Medicines is now only 27.9% of turnover and shrinking as a share every quarter. Its £225m of year-on-year decline cost the group 2.8 points of growth, and that cost falls mechanically every quarter as the segment's weight does.

Key KPIs

Long-acting injectables as a share of HIV turnover is computed from the release's product table: £593m of Cabenuva and Apretude against HIV turnover of £2,078m in Q2 2026, and £442m against £1,880m in Q2 2025. Q1 2026 free cash flow is the half-year figure of £2,809m less the £1,994m reported for Q2.

Key Topics & Management Commentary

Overall Management Tone: Confident and unusually specific, and structurally different from the last two calls because this was a three-hour combined results and strategy event rather than a one-hour print review. Where Q1's posture was defensive under questioning about earnings quality and the camlipixant bar, this event put seven R&D leaders in front of the room with dated trial commitments and let them take questions directly, which is a governance signal as much as a communication one. The one place management would not go was the 2027 margin, declined twice and explicitly.

1. A Beat With No Legal Settlements In It

Last quarter's central complaint was that three of the ten points of Core operating profit growth came from net legal settlements, a benefit the release quantified in SG&A at around four percentage points. That charge does not stick to this quarter. The Q2 release attributes the settlement favourability to the year-to-date column and specifically to Q1: Core operating profit growth "primarily reflected higher turnover, favourable product and regional mix, and favourable net legal settlements and expenses in Q1 2026," and the SG&A note places the roughly two-point year-to-date settlement benefit in the same place.

Q2's operating profit growth came instead from the gross-margin line.

"Sales grew 5% and gross margin improved 250 basis points due to product mix benefits driven by the growth of specialty and vaccines, together with supply chain optimization charges that we took in Q2 last year." — Julie Brown, Chief Financial Officer

That sentence contains its own qualifier. Part of the 250 basis points is a low prior-year comparator rather than a structural gain. But the mix component is durable, and it is the direct financial expression of Specialty crossing 45% of turnover.

Assessment: One quarter does not retire a quality-of-earnings concern, but it materially weakens it. The seven points of Core operating profit growth this quarter are operating points, and they were delivered while R&D rose 13% and royalty income fell 17%. That is a harder quarter to produce than Q1's was.

2. But the Assists Moved to the Rebate Line

What the release gives with one hand it discloses away with the other. Across three separate product commentaries, GSK quantifies non-demand contributions to the quarter's growth: three percentage points to Vaccines and to Shingrix from prior period rebate adjustments, and twelve percentage points to Nucala's growth from US channel-mix pricing adjustments.

"In Q2, US channel mix pricing adjustments positively impacted total growth in the quarter by 12 ppts and YTD by 6 ppts." — GSK Q2 2026 results announcement, Nucala commentary

The same mechanism runs the other way on Trelegy, where channel-mix pricing adjustments cost five points of growth in the quarter. Netted at group level these are smaller than Q1's settlement benefit and they are at least symmetric, since GSK discloses the negative one with equal precision. But an investor building a run-rate from the quarter's product growth rates without reading the commentary column will overstate Nucala by roughly half and Vaccines by three points.

Assessment: The bear point on earnings quality is confirmed again, in a different costume. It has now appeared as legal settlements, as prior-period rebate true-ups and as channel-mix adjustments across two consecutive quarters. The pattern is not fraud and it is not even unusual for a company of this scale; it is that GSK's reported growth rates carry a persistent one-to-three point layer of non-demand noise, and the market is right to discount them.

3. Camlipixant: £1.3bn Written Off, and the Second BELLUS Lesson

On 17 July GSK disclosed that CALM-1 met its primary endpoint at the 50mg twice-daily dose at week 12 while CALM-2 missed the same endpoint at week 24, that the 25mg dose failed in both, and that key secondary endpoints including the Chronic Cough Diary measure missed target thresholds in both. Development in refractory chronic cough stopped. The Q2 accounts carry the consequence: impairments of £1,895m against £476m a year ago, of which camlipixant is £1,334m, with a residual carrying value of £104m for the irritable-bowel indication that continues in phase IIb.

That single charge, plus a £486m contingent consideration remeasurement and £371m on the terminated Alector collaboration, took Total operating profit to £481m and Total EPS to 10.8p, down 69%.

"And in the end, we had one study that worked and one that didn't. But then we looked at the totality of the data, we just thought, okay, this is not going to shift care. And so we took that decision to reallocate the resources." — Luke Miels, Chief Executive Officer

The R&D chief's post-mortem was more technical and, read carefully, more troubling:

"The study read out in the 3 dimensions that we were anticipating, the Hawthorne effect overwhelmed the outcome to a greater extent than we anticipated." — Tony Wood, Chief Scientific Officer and President, R&D

Camlipixant came to GSK through the roughly $2.0bn BELLUS Health acquisition in 2023. The stated lesson is a shift toward hard endpoints, which management says is already visible in the accelerated portfolio.

Assessment: We carried camlipixant at zero at initiation, so there is no estimate change. Two things do change. First, the failure removes the most commonly cited candidate for closing the gap between the Street's roughly £35bn 2031 revenue build and GSK's more-than-£40bn outlook, and the outlook was reaffirmed without a new bridge. Second, a placebo-controlled cough programme failing on a Hawthorne effect is a design problem, not a biology problem, and management's answer was that the effect was anticipated but underestimated. Against a portfolio now built on 20-plus phase III starts, that is a live question about trial design capability, not a closed one.

4. Accelerate Growth: £1.9bn to Convert a Flat Margin Into a Rising One

The programme is the substantive news of the day. Three years, £1.9bn of annual savings fully realised by 2029 with 90% of the cumulative annual benefit by 2028, for £2.4bn of one-time cost of which £2.1bn is cash, charged to Adjusting items as a Major restructuring programme with the majority of charges falling in 2026 and 2027. The CFO put the payback at two and a half years before reinvestment.

"To fund investment in the late-stage portfolio and R&D, we are starting a 3-year cost savings programme to simplify the organisation and to reallocate capital and resources. Savings will primarily be reinvested, with some used to improve margins and profitability in the dolutegravir patent expiry period (2028-2030)." — Luke Miels, Chief Executive Officer

The consequential sentence is in the outlooks section: the programme "strengthens GSK's outlooks for growth of: sales of more than £40 billion by 2031; a stable to improving operating margin for the dolutegravir LoE period (2028-2030); and for accelerating growth from 2031 onwards." Stable became stable to improving. That is the first upgrade to the cliff-period outlook since it was set.

The savings mechanics are conventional: technology and AI enablement, streamlined support services and process redesign including procurement, reallocation of resources from mature brands to Specialty growth drivers, and further supply chain and site network simplification. The last of those is where the Stevenage-to-Cambridge move sits.

Assessment: The market has discounted GSK for years on the arithmetic of the dolutegravir cliff, and management's answer has always been a revenue answer: new molecules will arrive in time. This is the first cost answer, and cost answers are more credible because they are more controllable. The reservation is that only "a portion" of £1.9bn reaches the margin, the rest being reinvested, and management would not quantify the split. A programme that funds R&D is not the same as a programme that protects earnings, and the release is careful never to say which dominates.

5. Nuvalent: £7.1bn, Two Lung Assets, and the First Below-the-Line Bill

GSK completed the Nuvalent acquisition on 15 July, thirteen days before the print, at $124 per share in cash for $10.6bn of equity value, or $9.4bn net of cash acquired. It buys zidesamtinib, a selective ROS1 inhibitor, and neladalkib, a selective ALK inhibitor, both in non-small cell lung cancer, plus a phase I HER2 asset. Zidesamtinib was approved in the US on 22 July as Jideytro for pre-treated ROS1-positive disease, ahead of its 18 September target date. Neladalkib carries a 27 November action date.

"On the 15th of July, we completed the acquisition of Nuvalent at a net cost of GBP 7.1 billion, increasing net debt to GBP 22 billion, just under 2x net debt to 2025 core EBITDA." — Julie Brown, Chief Financial Officer

Reported net debt at 30 June was £15,132m against £14,453m at the December year end, so the £22bn figure is a post-period-end pro-forma. Leverage of just under 2x compares with the 1.4x we cited at initiation. The immediate income-statement effect is the reason Core EPS guidance moved down within its range.

"EPS is now expected to be in the lower half of the range, and this is predominantly due to the additional interest of around GBP 160 million following the acquisition of Nuvalent." — Julie Brown, Chief Financial Officer

Assessment: The deal is defensible on its merits and the timing is not a coincidence. Announced on 9 June and completed on 15 July, two days before the camlipixant discontinuation was disclosed, it gave GSK two late-stage oncology assets, one of them approved within a week of closing, at exactly the moment a phase III cough programme vaporised. At announcement GSK described Nuvalent as incremental to the more-than-£40bn 2031 ambition; by 28 July the CFO was declining to itemise increments in either direction. The honest reading is that Nuvalent now backfills camlipixant inside an unchanged target rather than adding to it, and shareholders paid £7.1bn plus roughly a point of near-term EPS growth for the privilege of the target not moving.

6. Phase III Starts Double, and What That Actually Costs

The headline pipeline number is a doubling: "Now expect 20+ phase III trial starts in 2026 (previously 10)." The mechanism management credits is a governance change, a fortnightly executive portfolio review begun in January in which project teams present directly to the CEO, the R&D chief and the chief commercial officer.

"Now since these reviews were started in January, we have identified acceleration opportunities across 7 assets, 18 indications and 25 studies." — Luke Miels, Chief Executive Officer

The seven accelerated assets sit in Oncology, Respiratory, Hepatology and Vaccines: five phase III studies for Mo-rez in gynaecological tumours, four for Ris-rez in lung and prostate, six across indications for the ultra-long-acting TSLP antibody, and efimosfermin extended into F4 MASH patients alongside F2 and F3, with the first F4 patient recruited the week before the event. The R&D chief also claimed a durable speed gain.

"We're 25% quicker than we were. We're in upper quartile as far as that's concerned." — Tony Wood, Chief Scientific Officer and President, R&D

The cost is visible in the P&L already. Core R&D grew 13% CER in the quarter to 20.5% of sales, up 90 basis points sequentially and 140 basis points year on year, and the full-year guide now reads "significantly ahead of sales" where it previously read ahead of sales. Total R&D, which includes the impairments, was £3,466m against £2,024m.

Assessment: A doubling of phase III starts in a single year is a bolus, and management said so. It converts a 2026 and 2027 P&L cost into a 2029-and-beyond revenue option at whatever the portfolio's true probability of success is. The CEO's own framing was that phase III programmes carry roughly 75% probability. Applied to 20-plus starts that implies five or more failures over the coming three years, each of which will produce a camlipixant-shaped impairment headline. Investors who cannot tolerate that cadence should note it now.

7. HIV: The Bridge Finally Gets Named Molecules and a Started Trial

At initiation this was the pillar we opened at risk, because the twice-yearly successor regimen had moved firmly into the next decade and management would not disclose the HIV component of the 2031 outlook. This event was where that gap was meant to be filled, and it largely was.

The near-term commercial position strengthened. HIV grew 10% CER with long-acting injectables at 80% of that growth, up from 73% in Q1, and at 35% of US HIV turnover. Cabenuva grew 33% and contributed 60% of total HIV growth; Apretude grew 39% and contributed 20%. The full-year HIV guide was raised on the call.

"Given our continued growth momentum, we are raising our 2026 sales growth guidance to high single-digit growth from mid- to high single-digit growth." — Deborah Waterhouse, Chief Executive Officer, ViiV Healthcare

The pipeline detail is what was new. The three-times-yearly treatment regimen now has a started registrational trial:

"I am pleased to share that QUATTRO, our 3 times a year Phase III registrational study in suppressed switch patients has begun, supporting potential approval in 2028." — Charlotte Allerton, Head of R&D and Chief Scientific Officer, ViiV Healthcare

The twice-yearly regimen was named for the first time as VH184, a third-generation integrase inhibitor now in phase IIb, combined with VH499, a capsid inhibitor whose phase IIb starts in the second half of 2026. Phase III is targeted for 2028 and approval for the latter part of 2030. Three-times-yearly prevention has registrational data anticipated in H2 2026 and approval targeted for 2027. On intellectual property, cabotegravir's new chemical entity protection runs into 2031, with additional patents granted into 2040 on the current two-monthly presentation and further patents pending into the mid-2040s on the newer regimens.

Assessment: Bull-2 moves back on track. Not because the twice-yearly regimen got faster, since late-2030 is the same answer as last quarter, but because the bridge no longer depends on it. Three-times-yearly treatment is in a started registrational trial for a 2028 approval, which is the front of the loss-of-exclusivity window rather than the back of it, and three-times-yearly prevention is a 2027 event. Meanwhile the commercial base is compounding faster than guided. The residual risk is entirely in the cannibalisation assumption, which is discussed below.

8. Bepirovirsen Approaches Its PDUFA as the Last Free Option

Bepirovirsen's pivotal B-Well 1 and B-Well 2 data were presented at EASL in May with simultaneous publication in the New England Journal of Medicine, meeting the primary functional-cure endpoint, with a higher cure rate as a key secondary in participants at or below 1,000 IU/ml surface antigen. The FDA decision is due by 26 October 2026, with reviews under way in Japan, China and the EU. If approved it would be, in the company's words, the first finite six-month therapeutic option for chronic hepatitis B.

Commercial framing on the call was more measured than the science. Management deliberately distanced the launch from the hepatitis C analogue, where a large prevalent cirrhotic population created immediate payer pressure. Here the treated population is not cirrhotic, testing for surface antigen is not routine in the US because it does not currently change the treatment pathway, and prior authorisation is expected to require it. On price, the chief commercial officer said only that "we do expect to be in the scale of hep C" while noting that reference point is now some years old, with payer negotiations running through the end of 2026 and into 2027.

Assessment: The asset is de-risked scientifically and still under-modelled commercially, which is the definition of an option. But the launch shape described on this call is slower than the one the bull case has been drawing: no bolus, a diagnostic gap to close in the largest market, and reimbursement conversations that start after approval rather than before. We continue to carry bepirovirsen at zero in our estimates. The 26 October decision is the single largest scheduled catalyst in the story.

9. Vaccines Guide Raised While Shingrix Runs Flat

GSK raised the Vaccines outlook from a low-single-digit decline-to-stable range to stable-to-low-single-digit growth, a genuine two-notch improvement, at the same print where Shingrix grew 3% CER including three points of prior-period rebate benefit. Both facts are in the same document and they are reconcilable: the raise is driven by the year-to-date position of +6% CER, by Meningitis outbreak demand, and by an Arexvy comparator that is easy for the rest of the year after a first half that more than doubled.

The Shingrix disclosure is the more informative one. US cumulative immunisation reached 45%, up three points in twelve months, while the ten largest ex-US opportunity markets average around 12%. A franchise that is 39% of Vaccines turnover and holds nearly half of its home market is arithmetically a low-growth asset in the US from here.

Assessment: The comp wall was real, it arrived on time, and it was absorbed. Our Bear-3 stays contained, but the reason has changed: it is contained because Meningitis and Arexvy filled the gap this quarter, not because Shingrix reaccelerated. Those two fillers are outbreak-driven and tender-driven respectively. Watch Q3, when the Arexvy comparator normalises.

10. Guidance Split in Two Directions at Once

The most unusual feature of this guide is that it moved up and down simultaneously. Turnover and Core operating profit went to the upper half of their ranges; Core EPS went to the lower half. The release is explicit that the divergence is financing:

"Core earnings per share is also expected to increase at the lower half of the range between 7 to 9 per cent at CER, reflecting higher interest charges of around £800 million, including the impact of the Nuvalent acquisition, and the tax rate which is expected to rise to around 17.5%, offset by the expected benefit from the share buyback programme." — GSK Q2 2026 results announcement

Underneath, the composition changed as well: SG&A is now expected broadly stable where it previously grew, royalty income was quantified at £850-900m, and R&D moved from ahead of sales to significantly ahead of sales. Those three together are the Accelerate Growth programme showing up in the current-year guide a year before its savings land.

Assessment: An operating upgrade paid for with a financing downgrade is a better outcome than the reverse, because operating changes compound and interest charges are refinanceable. But it does mean the headline "guidance reaffirmed" understates how much moved. On our estimates the two effects roughly cancel at the FY 2026 EPS line, which is why our number does not change this quarter.

11. Capital Allocation: The Buyback Is Finished and Nothing Replaced It

The £2bn buyback announced with the FY 2024 results completed in June at an average price of £16.13 over eighteen months. No successor programme was announced, and no analyst asked about one. The Q2 dividend was declared at 17p with 70p expected for the full year, unchanged, against 66p paid in 2025.

Cash generation was strong. Cash generated from operations of £2,906m was up 19%, and free cash inflow of £1,994m was up 77%, though roughly half of the half-year improvement came from one-off receipts on the linerixibat divestment and the ViiV shareholding restructure rather than from the business. Half-year contingent consideration payments were £757m, of which £710m went to Shionogi.

Assessment: This is the one leg of Bull-4 that weakened. At initiation the story was 1.4x leverage with a buyback running; it is now just under 2x with the buyback finished and £2.1bn of restructuring cash to spend over 2026 and 2027. The dividend is unaffected and remains covered roughly 2.6 times, and management repeated its capacity for further business development. But the near-term equity-return support that flattered Q1 and Q2 EPS by about a point is gone from here, and the guide already reflects that.

12. Cambridge, Stevenage, and Where the R&D Money Goes

GSK will establish a flagship R&D centre on the Cambridge Biomedical Campus in the UK and close Stevenage, with the move complete in early 2029. The stated logic is co-location with academic and clinical partners, in a pattern GSK is repeating across four hubs on the US East Coast, the UK, Europe and China. The CEO was unusually personal about it, describing the move as putting money where the company's mouth is on early-stage productivity, which he named as the area needing the most improvement.

Assessment: Site consolidation is part of how the £1.9bn gets delivered, so this is a cost item dressed as a science item, and both readings are true. For a shareholder the relevant number is that early-stage output is the one part of R&D management conceded is not yet working, and the fix has a 2029 completion date. That is well beyond any horizon on which this rating operates.

Guidance & Outlook

Implied second-half ramp. Weighting the 2025 halves by their actual split and solving for the second-half growth that produces each guided full-year rate gives a very uneven picture:

The turnover guide is undemanding: the upper half requires the second half to grow no faster than the first, and at the low end it can decelerate by two points. The operating profit guide is the stretch, requiring acceleration from 8% to as much as 10%, which is where the General Medicines recovery and the Accelerate Growth savings have to show up. And the EPS guide implies growth roughly halving, from 9% to somewhere near 5% to 7%, as the Nuvalent interest lands, the buyback support disappears and the tax comparator hardens.

Phasing. Management was specific and, unusually, front-loaded the warning:

"Now to support your modeling with respect to phasing, we expect operating profit growth to be significantly Q4 weighted. And a number of factors lead to this, including productivity charges taken in Q4 last year, whilst Q3 is impacted by the consolidation and phasing of Nuvalent costs and the acquisition-related interest together with a tough tax comparator." — Julie Brown, Chief Financial Officer

That is a change from the "predominantly half-two weighted" framing given in April, and it is a downgrade in disguise: half-two weighted became Q4 weighted, which means Q3 is now explicitly signposted as weak. An investor who marks to the H1 run rate in October will be disappointed by the Q3 print and should not be.

Street at: Consensus entering the print sat at £8.24bn of turnover, £2.68bn of Core operating profit and 47.1p of Core EPS, all of which the quarter cleared. The forward question is 2027, where management explicitly refused to guide margin and where the sell-side carries expansion.

Guidance style: Conservative and patterned. GSK narrowed within its standing full-year range at the half-year in 2025 as well, guiding Core operating profit "towards the top end" of a 6% to 8% range. The mid-year narrowing is a house habit, not a signal, and the company has a record of delivering the narrowed number.

Currency: The sensitivity is unchanged in magnitude from April. Holding the 20 July closing rates of $1.35, €1.18 and ¥219 to the pound for the rest of 2026 costs an estimated two points of sterling turnover growth and four points of sterling Core operating profit growth. That drag did not widen this quarter, which is itself a small positive after a first quarter in which it did.

Analyst Q&A Highlights

Whether the New Savings Programme Raises the Margin Target

The first question of the event went straight to the commercial point of the £1.9bn: does it lift the medium-term margin, or does it all get spent? Management's answer confirmed a change in the wording of the outlook while declining to attach a number to it, which is the whole of the ambiguity in one exchange.

Q: "So one of the most significant questions I've received today over and over is, does the new cost saving take that margin target higher? So am I right in interpreting Slide 24 that, yes, you are now aiming for margins up at the end of the 2031 period relative to where we are today."
— Matthew Weston, UBS

A: "We've not given a specific percentage, but what we have done is changed the margin guidance through dolutegravir now to say it will be stable to improving. So we're giving a range and recognizing the drop-through from the Accelerate growth program."
— GSK executive, results and strategy event

Assessment: The answer is a real upgrade and a real refusal at the same time. Stable to improving is better than stable and management chose the words deliberately. But without a split between reinvestment and drop-through, the market cannot underwrite more than the direction. Expect this question every quarter until a number appears.

What Deteriorated in General Medicines Within Three Months

The one guide that moved down drew the most pointed questioning, framed around whether something specific had emerged, including the June agreement with the US government on drug pricing. Management's answer separated the structural decline from the cyclical one and put the second-half recovery on two named mechanics.

Q: "Relative to the General Medicines guidance decrease, it was in part attributed to the soft environment. I'm curious what emerged in the last 3 months that led to this softness. Specifically, the company signed the agreement with CMS on June 15. So were there any details of that agreement, which were negative surprises?"
— Steve Scala, TD Cowen

A: "We've obviously seen Gen Med under some pressure in Q1 and the pressure increased slightly in Q2. So the full year view reflects that. I think very importantly, we knew Trelegy would be under pressure in the first half and Trelegy has been the asset that's been growing double digits. So it's taken away some of the strength in Gen Med that we had before."
— Julie Brown, Chief Financial Officer

Assessment: Notably, nobody attributed the deterioration to the CMS agreement, and management did not take the invitation to do so. The honest content of the answer is that Trelegy stopped being the offset that held the segment together, which is a mix problem rather than a policy shock. It is also the answer that makes the second half harder, since Trelegy has to stop falling for the group's operating guide to work.

Whether Nuvalent Is Inside the 2031 Target or On Top of It

With camlipixant gone and Nuvalent bought in the same six-week window, the composition of the more-than-£40bn 2031 outlook is now materially different from the one management described in February. The question asked directly whether the target had been re-based; the answer declined to itemise in either direction, while confirming the target requires no further business development.

Q: "First question is just on the 2031 target and beyond. I think Julie sort of ended with a comment on it in terms of 2031 now sounds like it's organic, but just to confirm that Nuvalent included in the 2031 are greater than $40 billion. And then the target or ambition for accelerating growth beyond 2031 ... how much of that is dependent on BD in terms of the ambition beyond 2031."
— Zain Ebrahim, JPMorgan

A: "We'd like to emphasize that we are saying we will deliver and we will deliver more than 40 but we didn't want to get into minor increments each way along. ... We obviously had the camlipixant news just over a week ago, but it is a portfolio, and we believe we will deliver more than 40% overall when you take it asset by asset. ... And then none of it at all is dependent on what you might call new BD."
— Julie Brown, Chief Financial Officer

Assessment: This is the most important non-answer of the event. A multi-billion acquisition described at announcement as incremental to the 2031 outlook has, seven weeks later, become part of a portfolio in which increments are not itemised, in the same breath as an acknowledgement of the camlipixant loss. The target survived; the bridge to it was quietly rebuilt. That is defensible portfolio management and it is also exactly why the Street's 2031 build sits several billion below the company's.

The 2027 Margin, Asked and Declined

The follow-up went to the year nobody has guided: 2027, where the Gilead royalty stream rolls off, US pricing policy bites and R&D is guided to outgrow sales significantly, yet consensus carries margin expansion. Management confirmed the framework and declined the year.

Q: "The second question is just a follow-up on 2027 margins ... next year, you've got the Gilead royalty going in Q4 IRA impact potentially and R&D, you've mentioned will significantly outgrow sales. So just how should we think about '27 in terms of margins? I think consensus has got margin expansion. So what are the other drivers that could get towards margin expansion?"
— Zain Ebrahim, JPMorgan

A: "So we have guided the margin to the end of 2026, which is more than 31% at 25 average exchange rates. We then guided the dolutegravir period because we knew and appreciated that investors were quite concerned about that period because of the profitability of HIV. ... But we did not want to get into guiding a margin every year."
— Julie Brown, Chief Financial Officer

Assessment: The gap between a 2026 guide and a 2028-2030 guide is 2027, and it is the year with the most identifiable headwinds. Declining to bridge it is a choice, and the consensus assumption of expansion is therefore unsupported by anything management has said. This is the single most likely source of a downward revision over the next two prints.

What Was Learned From the BELLUS Acquisition

With business development now explicitly central to the pipeline, the natural question was what a $2bn acquisition that produced a discontinued asset teaches about the next one. Management's answer was that the process was sound and the outcome was drug development.

Q: "Following the failure of camlipixant that we saw recently, are there any key learnings that you can now take from the BELLUS acquisition? Anything you would highlight here, particularly in the context of more recent business development decision-making post that deal with BELLUS ..."
— Kerry Holford, Berenberg

A: "I'm not sure what else we could have done different. ... And at the end of the day, this is drug development. We have a portfolio for a reason. Typically Phase III programs, 75% probability. So you are going to have situations like we had."
— Luke Miels, Chief Executive Officer

Assessment: Defensible, and also the answer that maximises future variance. If the process needs no change and phase III carries a 75% base rate, then twenty-plus starts in one year is a deliberate bet on the law of large numbers. The R&D chief's separate answer, that the group will now focus on hard endpoints where it can, was the more useful admission, because the Hawthorne effect that killed camlipixant is a subjective-endpoint problem.

Where Six-Monthly HIV Prevention Went

An earlier plan for a six-monthly prevention product was absent from the pipeline slides, and the questioning surfaced that it has been demoted to a line extension while the twice-yearly treatment regimen keeps its end-of-decade date.

Q: "So on HIV, I think the previous plan was to have a 6-monthly prevention product. I think it was in 2028, but I couldn't see it on the slide. So are you not doing 6 monthly prevention anymore? And on treatment, I think it was 6 month treatment, 28 to 30. ... when do you think that launch is that beyond '29 so post dolutegravir LOE is when you'd have a 6-monthly treatment?"
— James Gordon, Barclays

A: "So our market research and prevention says to us that really 3x yearly is the optimal. ... That said, we know there will be a subset of the market that we'll still seek for twice yearly. And so we remain committed to that as a line extension. ... the twice yearly treatment, you asked about potential time lines for that. We have a bold plan. We have a bold plan looking to get to approval in the latter part of 2030."
— Charlotte Allerton, Head of R&D and Chief Scientific Officer, ViiV Healthcare

Assessment: A deprioritisation presented as a market-research finding, and the finding may well be right, since a three-times-yearly cadence maps onto routine sexual-health visits in a way six-monthly does not. But the practical effect is that the prevention roadmap now has one fewer product in it, and the twice-yearly treatment approval date remains the latter part of 2030, at the very end of the loss-of-exclusivity window rather than inside it.

What Patent Life Is Actually in the Revenue Model

The most useful disclosure of the day came from a question about whether the long-dated HIV patents are being credited in the company's own forecasts. The answer was that they are not.

Q: "Just on HIV, the IP for cabotegravir up to 2040, maybe longer. What are you assuming in your revenue estimate? So when we look at your indicative revenue chart, what's in those assumptions?"
— Michael Leuchten, Jefferies

A: "So in terms of cabotegravir, the new chemical entity patent protection is into 2031. Obviously, 3-year patents, 3 times a year patents are then pending, and they would take you into the mid-2040s. I mean just in the way in terms of how we take these things into our forecast. Until we get an extension or until we get a new combination, we would always -- in the forecast we do, we would use the earlier date of expiry until we have extensions available to us, et cetera."
— Julie Brown, Chief Financial Officer

Assessment: This is a materially conservative disclosure and it cuts in GSK's favour. The company's own more-than-£40bn 2031 outlook credits cabotegravir only to its 2031 new chemical entity expiry, not to the 2040 patents already granted on the current presentation nor to anything pending into the 2040s. If those hold, the 2031 number is understated on that line rather than overstated. It also reframes the succession risk: the ViiV chief's stated model is rapid cannibalisation of the two-monthly into the three-times-yearly from 2028 and then into the twice-yearly, which is a franchise-management problem rather than a patent-cliff problem.

The Criteria for Accelerating an Asset

With seven assets accelerated across eighteen indications, the obvious question was what changed in the decision rule, and whether faster means riskier. The answer described an external benchmarking exercise followed by a funding constraint.

Q: "I wanted to check on the asset accelerations. So you've highlighted 7. What are the criteria for accelerating? Is there some new decision-making in that mix? And then I don't think Nelo is an accelerated asset. Is there a reason for that?"
— Luisa Hector, Berenberg

A: "Nuvalent wasn't in there because it was running on a similar track with the deal and the programs are more advanced. ... we wanted to sit down and look at the portfolio, benchmark it versus external parameters, not just marking your own homework ... And then we ended up -- we ran up against Julie, who said, that's lovely. And she's doing her job, she said, this needs to be paid for. And that's what then triggered the second round of the process."
— Luke Miels, Chief Executive Officer

Assessment: The sequencing in that answer is the honest bit and it explains the shape of the day. The R&D acceleration was designed first and the £1.9bn savings programme was constructed afterwards to pay for it. That is a healthier order than the reverse, in which a cost programme is announced and reinvestment is promised vaguely. It also means the savings target is calibrated to a spending plan rather than to a margin target, which is consistent with management's refusal to say how much drops through.

What They're NOT Saying

1. How much of the £1.9bn reaches the margin. The split between reinvestment and drop-through was asked twice and answered with a direction rather than a number. Without it, the upgrade from stable to stable-to-improving cannot be modelled, only believed.
2. The 2027 margin. Guided to 2026, guided for 2028-2030, and explicitly declined for the year in between, which happens to be the year the Gilead royalty rolls off, US pricing policy bites and R&D is guided to grow significantly ahead of sales. Consensus carries expansion on nothing management has said.
3. A re-bridge of the 2031 outlook. Camlipixant was removed and Nuvalent added within seven weeks, and the more-than-£40bn target did not move. Management declined to itemise increments in either direction. The asset-by-asset build behind the number remains undisclosed, which is why the Street's estimate of it sits several billion lower.
4. A successor buyback. The £2bn programme completed in June at an average £16.13 and nothing replaced it. Nobody on the call asked. With leverage at just under 2x and £2.1bn of restructuring cash to fund, the silence is informative.
5. Any peak-sales figure for the accelerated assets. Seven assets, eighteen indications, twenty-five studies, and no revenue ambition attached to any of them except in the aggregate 2031 number. For a strategy event whose stated purpose was to build confidence in the late-stage portfolio, that is a conspicuous absence.
6. What the Q3 print looks like. The CFO warned that operating profit growth is significantly Q4 weighted and named Nuvalent cost consolidation, acquisition interest and a tough tax comparator as Q3 headwinds, without quantifying any of them. The warning is real; the size of it is not disclosed.

Market Reaction

• Pre-print setup: The ADR closed at $51.98 on 27 July, up 6.0% year to date against the S&P 500's 8.3%, up 38.8% over twelve months and down 1.0% over the trailing thirty days. The London line closed at 1,961.0p, up 7.5% year to date and 40.3% over twelve months. The stock entered the print in the upper half but not at the top of a twelve-month closing range of $36.75 to $61.18.
• Reaction session: GSK reports before the London open, but held the results and the Accelerate Growth event at 14:00 BST, so 28 July is both the print day and the event day. The ADR opened at $54.24, a 4.3% gap, traded between $53.60 and $55.79 and closed at $53.71, up 3.3%. The London line opened at just 1,969.5p, up 0.4%, traded as low as 1,926.0p, which is 1.8% below the prior close, reached 2,096.0p and closed at 2,029.0p, up 3.5%.
• Volume: 8.3m ADRs against a 30-day average of 4.0m, or 2.1 times normal; 18.4m shares in London against 8.5m, or 2.2 times. Heavy but not extreme.
• Benchmark: The S&P 500 rose 0.2% on the session, so essentially all of the 3.3% ADR move was stock-specific.

The intraday shape is the informative part, and it is the mirror image of April. Last quarter the London line opened up 1.1% on a beat and closed down 5.4%, having sold off through the midday call. This quarter it opened up only 0.4% on a bigger beat, traded below the prior close at some point in the session, and finished up 3.5% with the day's high 6.9% above the prior close. The print did not move the stock; the event did. Whatever the market thought it was buying at 08:00 on 28 July, it thought better of it by the close.

That matters for how to read the move. A 3.3% ADR gain on a quarter that beat consensus by 7% on EPS is not an enthusiastic reception on the numbers alone. The plausible reading is that the beat was substantially anticipated, the Core EPS guide moving to the lower half of its range was an unwelcome surprise on the open, and the strategy content, specifically the £1.9bn programme and the margin-outlook upgrade through the cliff, is what re-rated the stock over the following six hours.

The camlipixant failure is not in this session. It was disclosed on 17 July, when the ADR fell 1.9% from $52.77 to $51.76 on 2.9 times average volume. A billion-dollar phase III asset dying for the cost of a 1.9% move tells you how little of it the market was carrying, which is consistent with the reset management applied to the success bar back in April.

Street Perspective

Debate: Does the £1.9bn Programme Change the Investment Case or Just the Slide?

Bull view: The market's structural discount on GSK has always been about the 2028-2030 earnings trough. A funded, dated cost programme that moves the margin outlook through that window from stable to improving attacks the discount at its source, and cost savings are more controllable than pipeline outcomes. Ninety percent of the annual benefit lands by 2028, precisely when it is needed.

Bear view: The savings are "primarily" reinvested and management would not say what fraction reaches the margin, twice. A restructuring that funds a doubling of phase III starts is a spending programme with a savings label, and it costs £2.4bn up front to find out. Pharmaceutical efficiency programmes have a long history of being announced, reinvested and then re-announced.

Our take: The bulls have the better of it on process and the bears on quantification. The sequencing management described, in which the R&D acceleration was designed first and the savings were constructed to pay for it, is a real answer to why the number is £1.9bn and not some rounder figure. But the case for a higher multiple rests on drop-through, and drop-through is the one thing not disclosed. We are underwriting the direction, not a specific margin.

Debate: Is the 2031 More-Than-£40bn Outlook Still Credible?

Bull view: It survived the loss of camlipixant without moving, which is what a genuinely diversified portfolio target should do, and it requires no new business development. It is also built on conservative patent assumptions: the CFO confirmed the model credits cabotegravir only to its 2031 new chemical entity expiry, ignoring patents already granted into 2040.

Bear view: The sell-side's build has sat several billion below the company's for some time, and camlipixant was one of the more commonly cited ways of closing that gap. It is now zero. The replacement, Nuvalent, was described at announcement as incremental to the target and seven weeks later as an unitemised part of a portfolio. That is a bridge being rebuilt in flight without being shown.

Our take: The conservative patent treatment is a genuine and underappreciated offset, and it is the single most useful thing disclosed at this event. But the bear framing is correct that the target's composition has changed materially in a quarter and the company chose not to re-bridge it. A 2031 number that is defended rather than rebuilt in public is worth less than one that is shown. We treat it as an ambition, not an estimate, and nothing in our valuation depends on it.

Debate: Does a Doubled Phase III Programme Create Value or Just Volatility?

Bull view: Twenty-plus starts across seven accelerated assets and eighteen indications, on validated targets with data already in hand, is the highest-quality shot-on-goal count GSK has had in a decade. The trials are 25% faster than they were, the portfolio has more than doubled its count of blockbuster-potential phase II and III assets since 2022, and this is what a company does when it believes its own data.

Bear view: At the CEO's own stated 75% phase III success rate, twenty-plus starts implies five or more failures, each of which arrives as an impairment headline like the one that just took Total EPS down 69%. The spending is certain, in 2026 and 2027 P&L; the returns are probabilistic and land in 2029 and beyond. And the same organisation just misjudged a subjective endpoint badly enough to write off £1.3bn.

Our take: Both are right, and the resolution is a matter of duration rather than argument. This is a decision to convert two years of earnings quality into a longer-dated revenue option. For a shareholder who owns GSK for a 3.45% covered yield and a mid-single-digit earnings algorithm, that is an acceptable trade at ten and a half times, and it would not be at fifteen. The failures will come and they will be noisy; investors should pre-commit to reading them as portfolio arithmetic.

Debate: Is Ten and a Half Times a Discount or a Verdict?

Bull view: The stock trades at roughly 10.4 times a 2027 estimate with net debt still investment grade, a covered dividend, and an October PDUFA on a first-in-class chronic hepatitis B asset that consensus barely carries. Peers on similar growth trade several turns higher.

Bear view: The multiple is the market's settled judgement on a company facing a patent cliff in three years, whose established-medicines base is falling 9%, whose 2027 margin management will not guide, and which just added £7bn of debt to buy back into a target it already claimed to be on track for.

Our take: The bear case is a fair description of the last three years and a weaker description of the next three, because the two things that would break the discount are both now scheduled: a funded margin bridge across the cliff and a first-in-class approval decision in October. The stock does not need a re-rating to sector leadership to work from here. It needs about one and a half turns, and it has two dated catalysts to earn them with.

Model Update & Valuation Framework

A note on basis. GSK guides Core EPS growth at constant currency, so our estimates above are stated on the same basis for comparability with the guide and with our prior quarter. The reported sterling figure will be lower while sterling is strong: half-year Core EPS grew 9% at constant currency and 6% as reported, and if that three-point gap persists the reported FY 2026 Core EPS lands nearer 180p than 185p. On reported earnings the multiple below is roughly 11.3 times rather than 11.0 times. We anchor the target on the constant-currency figure because that is the basis on which the business is guided and managed, and flag the translation as a risk to the target rather than to the thesis.

Valuation. At the 28 July close of $53.71 per ADS and 2,029.0p in London, which implies an exchange rate of $1.324 to the pound, GSK trades at 11.0 times our FY 2026 Core EPS estimate of 185p and 10.4 times our FY 2027 estimate of 195p, with a 3.45% dividend yield on the guided 70p and pro-forma leverage just under 2 times Core EBITDA. Applying a 12.0 times multiple to FY 2027 Core EPS produces a 12-month price target of $62 per ADS, or approximately 2,340p on the ordinary shares, implying 15.4% upside from the reaction-day close plus a 2.99% forward yield at the target price.

The target rises by a dollar despite a four-pence cut to the 2027 estimate, and the two moves have different causes. The estimate falls because the Nuvalent financing is a real, quantified, below-the-line charge. The multiple rises from 11.4 to 12.0 times because the reason for GSK's structural discount, an unfunded margin trough through the dolutegravir cliff, is now partly funded, dated and reflected in a changed outlook. Twelve times is not a re-rating to sector leadership; it is a smaller re-rating ask than the one embedded in our April target, and the stock has re-rated by less than a turn since.

What would change our mind. Three things, of which one has now half-happened. A bepirovirsen approval on 26 October restricted to the sub-1,000 surface antigen population would validate the bear framing of a niche asset and remove the story's largest free option. A Q3 print with General Medicines still at minus nine and no Trelegy stabilisation would break the second-half operating ramp that the upper-half guide requires. And any 2027 guidance, when it arrives, that puts the margin flat or down would confirm that the £1.9bn is entirely reinvestment. Any two of those together take us to Hold. The General Medicines trigger is closest: the segment printed worse than our threshold this quarter, but management named the mechanics of the recovery rather than asserting it, which is the difference between a warning and a trigger.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened. Two pillar upgrades, Bull-2 from at risk to on track and Bear-1 from emerging to contained, against one downgrade in Bull-4 from on track to at risk. The upgrades are structural, concerning the loss-of-exclusivity bridge that is the whole of the bear case on this stock; the downgrade is a financing decision that bought two late-stage oncology assets. Bear-2 is confirmed for a second consecutive quarter and is now the standing reason to discount reported growth rates rather than a reason to sell.

Action: Hold the position and add on weakness into the 26 October bepirovirsen decision. The next two prints carry a signposted weak Q3 and a Q4 on which the full-year operating guide depends, so the path to the target is not smooth. Nothing in this quarter argues for trimming.