Four Registrational Wins, a Higher Estimate and a Lower Price: Upgrading Bristol Myers to Outperform

Results vs. Consensus

Q3 2026 Scorecard

Consensus figures are the pre-print sell-side average for revenue and adjusted diluted EPS. Trackers disagreed on EPS, with estimates running from $1.10 to $1.21 depending on how much of the quarter's acquired IPRD each one had absorbed, so the size of the headline EPS beat says more about the estimates than about the quarter. We grade the underlying number against the company's own July trajectory instead: that guide implied roughly $1.63 a quarter for the second half before the item, and the third quarter delivered $1.93. Gross margin, operating margin and the cash-flow lines are not consistently consensus-tracked for this name, so no beat or miss is claimed on them.

Year-Over-Year Comparison

Operating margin follows the company's own definition: gross profit less SG&A and R&D, as a percentage of revenues, excluding specified items. Pre-tax profit before acquired IPRD adds the acquired IPRD line back to non-GAAP pre-tax earnings in both years. The EPS line excluding the item uses the company's disclosed net per-share impact of acquired IPRD charges and licensing income, $(0.58) this year and $(0.20) last year. Percentage changes are computed from the dollar columns shown.

Quarter-Over-Quarter Comparison

Third-quarter cash-flow figures are the nine-month columns of the 10-Q less the six-month columns of the second-quarter filing. Operating cash flow of $4,181M matches the $4.2B bar on the company's capital-allocation slide.

Quality of Beat

Revenue assessment. This is the first quarter in 2026 where the beat was not a guidance reset in disguise. The July raise had already moved the Street onto the right Eliquis trajectory, and the third quarter still came in $346M above it, with growth spread across Eliquis, Breyanzi, Reblozyl and Yervoy and a Revlimid tail that is eroding more slowly than full generic availability would suggest. The constant in all three quarters is that revenue excluding Eliquis has not grown: down 1.1% in the second quarter and down 1.1% again here. The Growth Portfolio is now large enough to roughly offset the generic erosion of everything except Eliquis, which is progress, and it is not yet large enough to do more than that.

Margin assessment. We were wrong in July. Gross margin had printed above the 69% to 70% guide twice with the range left unchanged, and we read the implied second-half step-down to roughly 68.1% as cushion. It was not cushion. The 10-Q's explanation also changes how the number should be read. A large part of the compression is the Pfizer profit share on a product whose U.S. net price rose sharply this year. Payments to alliance partners in cost of products sold were $2,323M against $4,795M of alliance revenues, so roughly half of each alliance revenue dollar is paid away before the company's own manufacturing cost, and Eliquis growth arrives at a much lower gross margin than the rest of the portfolio. That makes the gross-margin percentage a poor guide to earnings power here: it falls when Eliquis grows and will rise mechanically when Eliquis declines, while gross profit dollars move the other way.

EPS assessment. Two numbers should travel together. EPS before the acquired IPRD item rose 5.5% and beat the company's own July trajectory by about $0.30, which is a genuine upgrade to the 2026 earnings base. Pre-tax profit before acquired IPRD fell 2.3%, which means the operating business did not grow its pre-tax earnings once the royalty hole is counted. The royalty comparison laps after the fourth quarter; the tax rate is not a repeatable source of growth. We take the first number into our estimate and the second into our view of earnings quality.

Segment Performance

Bristol Myers reports as a single operating segment and discloses performance through the Growth Portfolio and Legacy Portfolio split. The tables below present the worldwide product detail as filed.

Portfolio Mix

Growth Portfolio by Product (worldwide)

Change and Ex-FX columns are the percentages as filed, which the company calculates from unrounded amounts. The Q2 2026 column is shown for the products discussed below. Other Growth Products comprises Abecma, Augtyro, Empliciti, Inrebic, Nulojix, Onureg, Zenbexus and royalty revenues, including royalties received from Merck on Winrevair, which the 10-Q puts at $151M for the quarter against $86M a year ago.

Legacy Portfolio by Product (worldwide)

Key KPIs

Eliquis: net price did the work, and Europe goes generic in November

Eliquis worldwide revenue of $4,690M grew 25% reported and 26% Ex-FX. The U.S. line rose 39% to $3,653M while international fell 7% to $1,037M. The product is now 35.9% of company revenue, against 30.6% a year ago, and its $944M of year-over-year growth was larger than the company's entire $847M increase. Nine-month Eliquis growth of 21.1% sits inside the unchanged 20% to 25% full-year guide.

"Eliquis revenue was $4.7 billion up 26% driven by net price along with strong demand and continued market share gains. Eliquis sales in Q4 will be impacted by the upcoming LOE in Europe, which will be partially offset by the continued growth in the U.S." — David Elkins, EVP and Chief Financial Officer

The 10-Q shows the mechanism directly. Company-wide gross-to-net adjustments fell to 30% of gross sales from 47% a year ago, and in the U.S. to 32% from 54%, with Medicaid and Medicare rebates down 66%. The filing attributes the change primarily to the 2026 Eliquis list-price reduction, which cut rebates by more than it cut net price. Average U.S. net selling prices rose 7% year to date. In the European Union, the apixaban composition-of-matter patents and their supplementary protection certificates expire in November 2026.

Assessment: The third quarter was the high-water mark for Eliquis. The guide brackets a fourth quarter between $4,025M and $4,747M, and the Legacy guide of a 3% to 5% decline only reconciles with the lower half of that range, which implies Eliquis falls roughly 5% to 14% sequentially as Europe goes generic. That is not a surprise and it is already in the full-year guide. It is a reminder that more than a third of company revenue is now a product with weeks of European exclusivity left, a U.S. business that runs until generic entry in April 2028, and a reiterated step-down of $1.5B to $2.0B in 2027.

Growth Portfolio: broad, decelerating slightly, and still only an offset

The Growth Portfolio reached $7,662M, up 11.7% reported and 12% Ex-FX, after 14% Ex-FX growth in the second quarter. Its share of revenue edged up to 58.6% from 58.3%. Management said eight Growth Portfolio products are now annualizing above $1B on third-quarter sales.

"We now have 8 growth portfolio medicines, each annualizing over $1 billion based on Q3 sales, giving us a broad base of growth drivers." — Christopher Boerner, Board Chair and Chief Executive Officer

The coverage ratio, Growth Portfolio dollar gains against the dollar erosion of the Legacy book outside Eliquis, was 0.90, against 0.92 in July and 0.71 in April. Sequentially the Growth Portfolio added only $102M, or 1.3%, with inventory drawdowns cited on Camzyos and Cobenfy.

Assessment: The breadth claim is fair: no single product drives the bucket, and the three largest growth contributors below delivered just over half of the dollar gain. The pace is the watch item. A 1.3% sequential increase in a seasonally soft quarter is not a problem on its own, but the ratio that matters for 2027 has stopped improving at 0.9, and the Legacy book it is offsetting has shrunk to $725M a quarter. From here the Growth Portfolio's job changes from offsetting Revlimid and Pomalyst to offsetting Eliquis, which is more than six times larger.

Opdivo and Qvantig: conversion is protecting the franchise

Opdivo fell 6% to $2,374M and the subcutaneous Opdivo Qvantig reached $334M, up 28% sequentially. The combined franchise was $2,708M, up 4.2% year over year and down 1.4% from a second quarter that included roughly $200M of disclosed wholesaler build. Qvantig is now 12.3% of the franchise.

"Qvantig continued its strong launch trajectory with revenue of $334 million representing sequential growth of 28%. And is now annualizing at over $1.3 billion." — David Elkins, EVP and Chief Financial Officer

In Q&A management put conversion at roughly 17% of the IV business to date and said it is tracking toward the higher end of its 30% to 40% target.

Assessment: The franchise grew 4.2% year over year with no inventory effect called out for the quarter, which is better than the flat underlying read we took in July. Qvantig's value is in holding the franchise together as IV volume converts, not in expanding it. It does not make the franchise a growth driver, and we model it as a stable contributor.

Reblozyl, Breyanzi and Camzyos: one of the three slowed

Reblozyl grew 22% to $750M, Breyanzi 46% to $526M and Camzyos 42% to $420M. Together they added $426M, or 52.9% of the Growth Portfolio's $805M increase. Breyanzi grew 8.7% sequentially despite the summer moderation management had flagged in July. Camzyos grew 1.0% sequentially.

"Reblozyl grew 22% to $750 million reflecting continued strength in first-line MDS associated anemia. Breyanzi posted 48% growth resulting in revenue of $526 million driven by its best in class profile and strong demand in both the U.S. and in international markets." — David Elkins, EVP and Chief Financial Officer

Management attributed the flat Camzyos quarter to a typical third-quarter inventory drawdown that it expects to normalize in the fourth quarter. The FDA approved Camzyos for pediatric patients with obstructive hypertrophic cardiomyopathy on September 30.

Assessment: Breyanzi was the best-performing large product of the quarter and the clearest sign that the cell-therapy franchise is scaling into the community setting. Camzyos is the one to watch. A 42% year-over-year rate looks healthy, but a flat sequential quarter from a product that grew 60% in the second quarter is a deceleration, and the inventory explanation can be checked against the fourth-quarter print.

Cobenfy: inventory took back the quarter

Cobenfy grew 33% year over year to $57M and fell 9.5% sequentially from $63M. The company attributed the sequential decline to destocking after a second-quarter build, against total prescriptions that management said grew 3% to 4% a month.

"With neuroscience, the Cobenfy revenue grew 33% to $57 million in the quarter. On a sequential basis, inventory destocking offset increased demand. We anticipate that revenue will more closely track prescription volume going forward." — David Elkins, EVP and Chief Financial Officer

Assessment: A sequential revenue decline two years into the launch of an asset once framed as category-defining is a poor optic, even with a credible inventory explanation. Our view has not changed since July: the schizophrenia business is not the investment case, and the 2027 Alzheimer's disease psychosis readouts are. Management's own description of the remaining work, accelerating new patient starts, improving repeat prescribing and moving patients to the effective dose, is the same list it gave in the second quarter.

Legacy Portfolio excluding Eliquis: down 55%, and almost gone

Outside Eliquis the Legacy Portfolio generated $725M against $1,619M a year ago and $941M in the second quarter. Pomalyst and Imnovid fell 81% to $130M on U.S. generic entry in March, and Revlimid fell 47% to $303M.

"Our legacy portfolio revenue increased 1% with growth in Eliquis more than offsetting the declines from generic entry across the remainder of the legacy portfolio." — David Elkins, EVP and Chief Financial Officer

Assessment: The Revlimid and Pomalyst cliff has largely played out: those two products are now $433M a quarter against $1,250M a year ago. That is why the Legacy guide improved to a 3% to 5% decline. It also means the Legacy book is now Eliquis plus a small tail, and every Legacy forecast from here is an Eliquis forecast.

Key Topics & Management Commentary

Overall Management Tone: More assured than in July and almost entirely pipeline-facing, with readout wins replacing the readout delays that dominated the last call. The financial detail got less airtime than at any point this year: the gross-margin step-down received one sentence, the contract-termination charge was not sized, and no analyst asked about margins, the EPS guide revision or the implied fourth quarter. Where management was pressed, on the 120mg admilparant dose and on business-development pace, the answers were candid but did not add new information.

1. The guidance revision: a revenue raise, a charge, and about $0.15 of real improvement

Management raised full-year revenue guidance to approximately $49.75B to $50.25B from $49.0B to $50.0B and cut the non-GAAP EPS range to $6.35 to $6.55 from $6.75 to $7.00. The EPS change is a definitional one: the July range carried $(0.02) of acquired IPRD and licensing income, and the October range carries $(0.60) after the third quarter's $(0.58). Gross margin, operating expenses and the tax rate were unchanged, other income and expense improved to about $(650)M, and the Legacy decline improved to 3% to 5%.

"Excluding the third quarter acquired in process R&D charges and licensing income of $0.58 per share, the midpoint of our 2026 non GAAP EPS guidance would have increased by approximately $0.15 per share." — David Elkins, EVP and Chief Financial Officer

On a like-for-like basis, the midpoint before the item moved from $6.895 to $7.05. Against FY2025 non-GAAP EPS of $6.15, which included $(1.40) of the same item, the headline 2026 midpoint now implies 4.9% growth; on the ex-item basis, FY2025 was $7.55 and 2026 is guided down 6.6%.

Assessment: The raise is real and modest, about 2% on the ex-item midpoint, and it came entirely from the quarter already reported. Two framings will circulate and both are incomplete. The headline cut is an accounting effect of business development, not a deterioration. But the ex-item 2026 figure is also lower than 2025's on the same basis, which is the clearest single statement in the guide that 2026 is a year of earnings decline before the 2027 step-down begins.

2. The implied fourth quarter: everything steps down at once

The guide's arithmetic for the remaining quarter steps down on every line. Revenue at the $50.0B midpoint implies $12,468M for the fourth quarter, down 4.6% from the third and down 0.3% from the fourth quarter of 2025. EPS at the $6.45 midpoint implies $1.48 against $1.93 before the item in the third quarter. The 69.5% gross-margin midpoint implies about 67.9%, operating expenses of about $16.5B imply $4,441M, up 9.3% sequentially, and other expense of about $650M implies $258M against $93M in the third quarter.

"We are maintaining our projections for gross margin, operating expenses, and tax rate while slightly lowering the expense level on our other income and expense line." — David Elkins, EVP and Chief Financial Officer

Some of the step-down is genuine. European Eliquis generics arrive in November, the fourth quarter is seasonally heavy for operating expenses (the fourth quarter of 2025 ran $4,648M), and the nine-month tax rate of 16.4% leaves room for a higher fourth-quarter rate inside a year guided at about 18%. Some of it is the same pattern as the first half: three quarters running above the ex-item EPS trajectory the guide implied.

Assessment: We model a fourth quarter of about $1.59 and full-year EPS of $6.55, the top of the range, on revenue of $50.2B and a full-year gross margin of 69.7%. We do not assume the guide is sandbagged on gross margin; we got that wrong in July. We do assume the other-income line and the tax rate land better than the guide's arithmetic, because both have run better all year. The fourth quarter is the first test of whether the third quarter's margin is a new level or a trough.

3. Gross margin: the step-down arrived, and it is Eliquis-shaped

Non-GAAP gross margin was 68.4% against 72.9% a year ago and 71.4% in the second quarter. Management's explanation was one sentence. The 10-Q adds the two numbers that matter: the one-time contract-termination charge was $140M, and the cost increase was "primarily due to alliance profit sharing for Eliquis." Payments to alliance partners booked in cost of products sold were $2,323M against $1,851M a year ago.

"Gross margin in the third quarter was 68.4% reflecting product mix and to a lesser extent a 1 time expense for a contract termination." — David Elkins, EVP and Chief Financial Officer

Before the $140M charge, gross margin was 69.4%, inside the 69% to 70% full-year range after two quarters above it. The non-GAAP cost increase of $821M breaks down as $472M of higher alliance payments, $140M of the one-time charge and $209M of everything else.

Assessment: This settles the July question against us: the implied second-half step-down was real, and the guide was not conservative on this line. The useful insight is the mechanism. When the Eliquis net price rises, a large share of the gain is paid away to the alliance partner through cost of products sold, so Eliquis growth dilutes the percentage while still adding dollars. That changes Bear 2. Margin compression is confirmed for 2026, but it is mostly a function of Eliquis mix, and it reverses in percentage terms when Eliquis declines. The gross profit dollars are what fall in 2027 and 2028.

4. Earnings quality: the tax rate did the work this quarter

The non-GAAP effective tax rate was 14.1% against 22.3% a year ago and an approximately 18% full-year guide. Management attributed it to jurisdictional earnings mix and to the acquired IPRD charge itself.

"Our effective tax rate in the quarter was 14.1%, reflecting jurisdictional earnings mix and the impact of an acquired in process R&D charge. And we had a $0.58 per share impact from acquired in process R&D and licensing income in the quarter." — David Elkins, EVP and Chief Financial Officer

The bridge is short. Core operating profit, gross profit less SG&A and R&D, rose 3.9% to $4,873M. Other income and expense turned $299M worse as the $286M diabetes royalty stream from the prior year disappeared. Pre-tax profit before acquired IPRD therefore fell 2.3% to $4,781M. EPS before the item still rose 5.5%, and the difference is the lower tax rate.

Assessment: Bear 3 is confirmed for a third consecutive quarter, with a better shape. The operating business grew its profit, and the drag sits in one expired royalty that stops distorting comparisons after the fourth quarter. The risk is that the next distortion is already scheduled: see the following topic.

5. 2027: the pushes and pulls management named, and one it did not

Asked directly about 2027, management named three items: the Growth Portfolio's momentum, the reiterated $1.5B to $2.0B Eliquis step-down, and a strategic productivity initiative that the 10-Q says should deliver about $2.0B of annual savings by the end of 2027. The CEO added that the company would be aligning its operating-expense base with the opportunity ahead.

"As we think about the business today, we feel good about the momentum we have going into Q4. We hope that and anticipate that will continue into 2027." — Christopher Boerner, Board Chair and Chief Executive Officer

The 10-Q discloses a fourth item. Merck's 2.5% royalty on global Keytruda sales and Roche's royalty on Tecentriq both run through December 31, 2026. The two streams contributed $166M to other income in the third quarter and $499M over nine months, roughly $665M a year at the current run-rate. The filing also notes that the Medicare maximum fair price for Pomalyst takes effect on January 1, 2027, and that Orencia, a $983M-a-quarter product, was selected in January for government-set prices from 2028.

Assessment: We had not modelled the royalty roll-off in July and we should have. At an 18% tax rate it is worth about $0.27 of EPS, close to 4% of our 2026 ex-item estimate, and it lands in the same year as the Eliquis step-down. It does not change the direction of 2027, which management has already signalled, but it makes the first 2027 guide in February more likely to show a decline than a flat year. We size it explicitly in the valuation below rather than leave it implicit.

6. Acquired IPRD and business development: $1.56B of charges, mostly early-stage

The $1,558M of acquired IPRD comprised $775M for the Hengrui portfolio of 13 early-stage assets (a $600M upfront payment plus a $175M noncancellable 2027 anniversary payment), the $500M first anniversary payment to BioNTech on pumitamig, $250M for a SystImmune iza-bren development milestone, and $33M of other items. Hengrui is also eligible for up to $14.3B of contingent milestones.

"In terms of size, we continue to be agnostic on the size of the deal. Far more critical to us is does it make sense for us to be the rightful owner of the opportunity?" — Christopher Boerner, Board Chair and Chief Executive Officer

The posture was otherwise unchanged from July: business development is a top capital-allocation priority, the company is evaluating partnerships, licensing and acquisitions, and the near-term growth profile is a stated criterion. A question at the end of the call noted the gap between the company's business-development spending and its peers'.

Assessment: The spending is real cash, about $1.1B paid in the quarter, and it is buying early-stage optionality rather than near-term revenue, which is the opposite of what the 2027 to 2028 gap needs. We do not capitalize these charges into the multiple, because they are investment outlays rather than operating costs, but we note that each charge is one-time while the line itself recurs: acquired IPRD was $3,721M in 2025 and $1,651M in the first nine months of 2026. The stated criterion of strengthening the near-term growth portfolio has not yet been met by a transaction.

7. Admilparant: the primary endpoint hit, the higher dose did not

ALOFT-IPF read out on October 12. The trial met its primary endpoint at the 60mg dose, the 120mg dose missed statistical significance, and both arms showed discontinuations for adverse events similar to placebo. The stock fell 2.6% that day and another 1.6% the next.

"The trial met its primary endpoint at the 60mg dose, with admilparant demonstrating a statistically significant improvement compared to placebo in absolute change in forced vital capacity (FVC) from baseline at Week 52. At the 120mg dose, admilparant slowed the decline of FVC but was not statistically significant." — Company, Q3 2026 earnings release

On the call the CEO said the company will discuss the results with regulators and expects Phase 3 results in progressive pulmonary fibrosis, the larger commercial opportunity, in the first quarter of 2027. The Chief Medical Officer disclosed that the 120mg dose was never tested clinically in Phase 2 and was selected by modelling. Secondary endpoints, including time to event and mortality, were not disclosed.

"We will discuss the IPF results with regulators and expect phase 3 results in progressive pulmonary fibrosis in the first quarter of next year." — Christopher Boerner, Board Chair and Chief Executive Officer

Assessment: This was the readout we said in July was carrying the year, and it delivered the primary endpoint at the dose that emerged from Phase 2. That is a positive result and it satisfies the condition we set. The higher dose missing is a genuine qualifier for two reasons: a dose-response story that does not show a response at the higher dose invites regulatory questions, and the progressive pulmonary fibrosis trial tests both doses too. We count the readout as delivered and hold the qualifier against conviction rather than against the rating.

8. Zenbexus: approved, positive PFS, revenue undisclosed

The FDA granted accelerated approval to Zenbexus (iberdomide) on August 13, ahead of the August 17 PDUFA date, as the first approved CELMoD. On October 8 the EXCALIBER-RRMM trial met its progression-free survival endpoint with a 51% reduction in the risk of progression or death; the data go to ASH in December. Mezigdomide has a PDUFA date of May 13, 2027.

"Mezigdomide offers the potential to introduce another highly efficacious and convenient oral option for a different segment of multiple myeloma patients." — Christopher Boerner, Board Chair and Chief Executive Officer

The 10-Q shows $2.3B of IPRD reclassified to marketed product rights on the Zenbexus approval, which begins an amortization charge that falls outside non-GAAP earnings. Launch revenue sits inside Other Growth Products, which grew 12% to $731M, and was not broken out.

Assessment: The regulatory and clinical risk on the first CELMoD is now largely retired, and the PFS result arrived within the few months of approval that management had promised. The open question from July is unchanged: the commercial plan is to replace Revlimid and Pomalyst regimens at branded pricing against generics, and the company has still not offered a pricing or gross-to-net framework. Mezigdomide in May gives the franchise a second product in the same setting.

9. RYZ101 and arlo-cel: platform validation, modest dollars

RYZ101, the actinium-based radiopharmaceutical from the RayzeBio acquisition, met its interim progression-free survival endpoint in gastroenteropancreatic neuroendocrine tumors after Lutetium-177 therapy, which management described as the first Phase 3 results for an alpha-emitting targeted radiopharmaceutical therapy. Arlo-cel, a GPRC5D-directed CAR T, met its primary endpoint in the registrational QUINTESSENTIAL study in quadruple-class-exposed myeloma.

"Commercially, it is a relative modest opportunity. However, bigger picture, we see RYZ as a true IND engine with multiple INDs expected in the next 5 years that will ultimately enhance our oncology portfolio with unique platform and pipeline." — Adam Lenkowsky, EVP and Chief Commercialization Officer

Management sized the U.S. incidence at roughly 18,000 patients a year, with about 70% of those progressing on Lutathera needing a subsequent therapy.

Assessment: Credit to management for sizing the first RYZ101 indication honestly. Both readouts matter more as platform evidence, for the radiopharmaceutical engine bought with RayzeBio and for the CAR T franchise beyond Breyanzi, than as revenue. Neither moves the 2027 to 2028 arithmetic, and we assign neither explicit value.

10. Milvexian: the stroke readout by year end, atrial fibrillation in the first quarter

The secondary stroke prevention readout, which we said in July would break Bull 2 if it slipped, is still guided for 2026. The atrial fibrillation study remains on the first-quarter 2027 timeline set in July.

"Before year end, we expect Phase 3 data for Milvexian in secondary stroke prevention and Sotyktu in lupus." — Christopher Boerner, Board Chair and Chief Executive Officer

Two questions pressed on milvexian. One asked whether the stroke study's randomization window, relative to a competitor's stroke trial in the same class, puts the result at risk; the Chief Medical Officer defended the window as selected from the company's own Phase 2 and prior Phase 3 data. The other asked whether the atrial fibrillation topline will carry numbers; the CEO said more specifics in the press release will likely be important, balanced against presenting the full data at a medical meeting.

Assessment: Timing discipline held, which was the July test. The stroke readout in the next two months is now the largest single binary in our 12-month window, ahead of atrial fibrillation in the first quarter. Our base case still carries no value for milvexian, so a negative stroke readout costs the thesis catalyst value rather than estimate value; a positive one is upside we do not model.

11. Cobenfy: a 2027 story with a November preview

The ADEPT timelines set in July were reiterated. The company will present open-label Alzheimer's disease psychosis data from the ADEPT-1 lead-in and the ADEPT-3 extension at CTAD in November, has a potential ADEPT-1 interim analysis later this year, and expects all randomized ADEPT readouts and the BALSAM bipolar I data during 2027.

"Everything is tracking as we guided. We have a potential interim analysis for ADEPT-1 later this year. And we anticipate the top line results readouts for all the ADEPT program throughout 2027." — Cristian Massacesi, EVP, Chief Medical Officer and Head of Global Drug Development

Assessment: No slip is the best available news on this programme, and management's case for the Alzheimer's psychosis market, larger, underserved and dominated by off-label antipsychotics with boxed warnings, is coherent. The November data are open-label and cannot resolve anything. We continue to carry no value for the label expansion.

12. Zola-cel: a pause, a death, and a stated root cause

The autoimmune CAR T programme has been in a voluntary pause since June. In October the company disclosed a Grade 5 event, a patient in the Phase 1 multiple sclerosis cohort who died of septic pneumonia; management described it as the only Grade 5 event across roughly 190 patients treated. The stock fell 2.4% on the day of the disclosure.

"When you are doing drug development, as Christian alluded to, in very difficult to treat settings with a new modality like this, you are always gonna be evaluating the benefit risk profile." — Christopher Boerner, Board Chair and Chief Executive Officer

The Chief Medical Officer said the root cause of the issue behind the pause was a change in a reagent used in cell selection and activation, that the adjustment has been made, and that the company is working with regulators to restart.

Assessment: This is an early-stage programme with no value in our numbers, so the financial consequence is nil. The governance consequence is not. A manufacturing-process change that required a programme-wide pause is the kind of execution issue the company's emphasis on trial rigour is meant to prevent, and it sits alongside the ADEPT conduct remediation we discussed in July. We will watch the restart timeline.

13. Cash flow and capital allocation: the best cash quarter of the year, and still no buyback

Operating cash flow was $4,181M, the highest of 2026, and free cash flow of $3,837M covered the $1,287M dividend 3.0 times. Net debt fell $1,738M to $29,918M, mostly through cash accumulation: total debt fell only $243M while cash and marketable securities rose $1,495M to $12,959M.

"We remain in a solid financial position, with approximately $13 billion in cash equivalents and marketable securities as of September 30. And we generated approximately $4.2 billion in operating cash flow during the quarter." — David Elkins, EVP and Chief Financial Officer

Against the year-ago quarter, operating cash flow was still down 33.8%, and over nine months it is down 28.8% to $8,678M. The rebate liability drained $734M of cash in the third quarter against $3,135M in the first half, and its balance has fallen to $4,915M from $8,844M at the end of 2025. The approximately $5B repurchase authorization remains unused, and the diluted share count rose 0.7% year over year to 2,053M.

Assessment: The rebate offset promised in April is arriving gradually rather than all at once: the working-capital drain has narrowed sharply, and the liability is now a little over half its year-end level. Bear 4 stays contained. Capital allocation remains coherent and conservative: dividend first, debt down $4.1B year to date, cash building toward a transaction. The unused authorization is now a choice the company is clearly making, and the share count drifts higher in the meantime.

Guidance & Outlook

The guidance excludes acquired IPRD charges and licensing income incurred after September 30, 2026, and assumes 2,051M diluted shares. July guidance used mid-July exchange rates and October guidance mid-October rates. The "before the item" row adds back the company's disclosed acquired IPRD and licensing amounts and is our arithmetic.

Implied fourth quarter. Nine-month revenue was $37,532M, so the $50.0B midpoint implies $12,468M for the fourth quarter, 4.6% below the third. Nine-month non-GAAP EPS was $4.97, so the $6.45 midpoint implies $1.48, 23.3% below the third quarter's $1.93 before the item and 20.4% below the fourth quarter of 2025 on the same basis ($1.86). The Eliquis and Legacy ranges together imply a fourth-quarter Eliquis figure in the lower half of its $4,025M to $4,747M range as European generics arrive.

Street at: The pre-print revenue consensus of about $12.72B already sat above the roughly $12.52B a quarter that the July midpoint implied for the second half, so the Street was leaning above the guide and the quarter still beat it. EPS estimates were scattered by the treatment of acquired IPRD, so the more useful comparison is the company's own: three quarters running above the trajectory its guide implied, before the item.

Guidance style: Mixed, and more informative than in July. On revenue and EPS before the item, the pattern of setting a guide and beating it held for a third quarter. On gross margin the guide was right and we were wrong. We treat the fourth-quarter revenue and EPS arithmetic as conservative and the margin arithmetic as roughly accurate.

Analyst Q&A Highlights

Whether the 120mg admilparant dose still belongs in the package

The first question of the call went straight to the higher dose that missed in ALOFT-IPF and asked whether management still expects it to work. The answer led with the positive 60mg result and the placebo-like tolerability in both arms, then made a disclosure that reframes the miss: the 120mg dose was never tested clinically in Phase 2 and entered Phase 3 on the strength of modelling.

Q: "Recently, had results for the admilparant drug as you just mentioned and there were some mixed results in the IPF study. Could you just comment a little bit about your confidence at the 1 hundred and 20 mg high dose which had some more mixed results and whether you still expect that dose to be working ..."
— Michael Yee, UBS

A: "But you may recall that we did not formally test the 1 hundred and 20 milligram in Phase 2 setting clinically. It was a modeling. ... We cannot predict what will be the outcome at according to the dose is, but we are testing both of them and we will see the outcome."
— Cristian Massacesi, EVP, Chief Medical Officer and Head of Global Drug Development

Assessment: A candid answer that lowers the significance of the miss and raises the stakes for the next trial. If 120mg was a modelled extrapolation rather than a Phase 2 dose, its failure is less damaging to the mechanism than a dose-response failure would be. But the progressive pulmonary fibrosis study, the larger commercial opportunity, tests the same two doses, and management declined to predict which will work there.

What the IPF trial showed beyond the primary endpoint

A follow-up asked whether admilparant showed any benefit on time-to-event measures or mortality, the secondary endpoints that would distinguish it in a market where a competitor is launching well on a modest lung-function effect. Management declined to disclose them ahead of a medical meeting next year, and separately declined to discuss the monotherapy and add-on subgroups.

Q: "And did admilparant show a benefit on time to event or mortality in your IPF trial?"
— Akash Tewari, Jefferies

A: "Did not disclose this. We mentioned we talked in the press release about the primary and some of the safety profile. I can tell you that we are pleased the overall outcome of the trial, and we will present this data in a medical meeting next year."
— Cristian Massacesi, EVP, Chief Medical Officer and Head of Global Drug Development

Assessment: Withholding secondary data until a congress is standard practice and not in itself a negative signal. It does leave the commercial case resting on tolerability and an FVC effect at one dose, which is a narrower foundation than the market-expansion narrative management offered alongside it. The full dataset next year is the event that sizes this asset.

What the first weeks of the Zenbexus launch look like

Asked for early launch read-outs, management described positive physician feedback, use in both community and hospital settings, and an expectation that most use will eventually come from the community, where 70% to 80% of myeloma patients are treated. No prescription or revenue figure was offered.

Q: "I guess what are you seeing on the ground in the early weeks of launch And what feedback are you hearing from the community and academic myeloma physicians on uptake, tolerability, and where they are using it."
— Alexandria Hammond, Wolfe Research

A: "Now it is still early in the launch, but our teams are executing well. We are pleased with the adoption that we are seeing over the first few weeks on the market, and physician feedback has been positive."
— Adam Lenkowsky, EVP and Chief Commercialization Officer

Assessment: Qualitative and fair for a launch about eleven weeks old. The figure that matters, how quickly Zenbexus displaces generic lenalidomide and pomalidomide in second-line regimens and at what net price, will not be visible until the fourth-quarter print at the earliest, and the company has not committed to breaking it out.

Whether the stroke trial's design puts the readout at risk

A recurring line of investor concern was put directly: whether milvexian's secondary stroke prevention trial, given its randomization window relative to a competitor's trial in the same class, can still succeed. Management defended the design as grounded in its own Phase 2 and earlier Phase 3 work.

Q: "And then focusing on the upcoming readout for Milvexian in SSP, can, based on the baseline characteristics and assuming the asymptom time from index event data are reliable, how confident are you that SSP can still be successful?"
— Zach Dunn for Seamus Fernandez, Guggenheim

A: "This study was designed based on a very solid phase 2 trial. That we run showing 30 percent risk reduction in strokes. ... We based all the criteria when the patients are randomized based on our own Phase II and based on prior Phase III studies."
— Cristian Massacesi, EVP, Chief Medical Officer and Head of Global Drug Development

Assessment: The answer was the right one for a blinded trial and gave investors nothing to trade on. The question matters more than the answer: the market is already handicapping the stroke readout against a competitor's design, which means the result will be read comparatively rather than on its own terms. That raises the bar for a positive readout to move the stock.

How much the atrial fibrillation topline will tell investors

Questioning turned to disclosure: whether the first-quarter atrial fibrillation press release will carry numbers, given that the non-inferiority margin on the primary endpoint leaves a wide range of outcomes that would all count as success. Management acknowledged the point and committed to balance rather than to specifics.

Q: "Are we likely to see the details, the numerical details in that press release or are we just going to get a positive or not positive trial details there?"
— Courtney Breen, Bernstein

A: "We understand that in that context, providing more specifics in the press release will likely be important. But we are gonna need to balance this course, with ensuring that we can communicate the data in a more fulsome way at a future medical conference."
— Christopher Boerner, Board Chair and Chief Executive Officer

Assessment: A reasonable answer that implies a two-step resolution of the largest binary in the story: a topline in the first quarter, and the data that determine commercial uptake later at a meeting. Investors should expect a positive topline without numbers to move the stock less than they think, for the same reason the question was asked.

The pushes and pulls in 2027

The only financial question of the call asked for an early framing of 2027 and whether the Street has the next few years right. Management listed Growth Portfolio momentum, the pipeline readouts, the reiterated Eliquis step-down and the productivity programme, and declined to comment on consensus.

Q: "kind of the core question I had was just as we think about 2027, just any initial pushes and pulls we need to think about in the business?"
— Christopher Schott, JPMorgan

A: "And then, you know, I would say on the other side as we already talked about with Eliquis, we anticipate that $1.5 billion to $2 billion step down in 2027. And lastly, you know, our strategic productivity initiative, that $2 billion worth of efficiencies that we are well on track to achieve that by the end of 27."
— David Elkins, EVP and Chief Financial Officer

Assessment: Consistent with July, and incomplete. The Eliquis step-down was reiterated for a third consecutive quarter without revision, which is credible. But the list of 2027 headwinds omitted the expiry of the Keytruda and Tecentriq royalties on December 31, a $665M-a-year income stream on the current run-rate that the company's own 10-Q discloses. An answer to a question about 2027 pushes and pulls is the natural place to mention it.

What is holding Cobenfy back

A question pressed on why Cobenfy remains small against early expectations, and whether the constraint is patient identification, prescriber willingness or tolerability. The response conceded the same three problems as in July and then made the case that Alzheimer's disease psychosis is a structurally different market.

Q: "So it does look like this quarter, you saw 30% year over year growth. But the numbers still appear modest relative to what we were all thinking about, how big this asset could be."
— Mohit Bansal, Wells Fargo

A: "We need to accelerate new patient starts. We need to improve repeat prescribing. ... And there is still far too many patients that we see remaining on the 50 or 100 milligram dose."
— Adam Lenkowsky, EVP and Chief Commercialization Officer

Assessment: Honest, and unchanged. The same diagnosis has now been offered for several consecutive quarters, which tells us the schizophrenia launch has stabilized at a level the company is managing rather than solving. The Alzheimer's psychosis argument is the better one and it remains a 2027 event.

The zola-cel death and the path back for autoimmune CAR T

The question asked how the company now weighs the risk and reward of CAR T in autoimmune disease after the Grade 5 event in the multiple sclerosis cohort. Management separated the death from the earlier programme pause, gave a root cause for the pause and committed to restarting.

Q: "Can you perhaps walk us through your updated thoughts on the path forward for Zolacel in autoimmune and the risk reward for CAR T in autoimmune?"
— Trung Huynh, RBC

A: "But after an extensive analysis of all our data, we have identified that the root cause was a change that we made in the selection and activation of the reagent. We have now made the proper adjustment of it and we are working with authority to restart the clinical program as soon as possible."
— Cristian Massacesi, EVP, Chief Medical Officer and Head of Global Drug Development

Assessment: A specific root cause is better than a general reassurance, and the safety record cited, one Grade 5 event across about 190 patients, is the right frame for a modality being tried in severe disease. No restart date was given. Financially immaterial; operationally, a second process issue in a year for a company that has made trial rigour a central message.

Why business-development spending has trailed peers

The final question asked whether the gap between the company's business-development spending and its peers' reflects valuation discipline, confidence in the internal pipeline, or confidence that the Growth Portfolio can fill the exclusivity gap. Management repeated that business development is a top priority and that it has the capacity to act.

Q: "Know, there is been a bit of a disconnect in the amount that you have spent in BD over the past 12 months versus some of your peers. But is this difference primarily reflect valuation discipline confidence in your internal pipeline or your ability to grow the growth portfolio to fill the LOE gap?"
— Evan Seigerman, BMO Capital Markets

A: "We are certainly not waiting for anything If there is an opportunity that makes sense financially for us, improve our ability to deliver sustained growth. We have the capacity to be in the mix."
— Christopher Boerner, Board Chair and Chief Executive Officer

Assessment: The question offered three explanations and the answer chose none of them. With $13B of cash, net debt down $4.1B this year and an unused $5B authorization, capacity is not the constraint, and the company's own criterion is a deal that strengthens the near-term growth profile. The quarter's $1.56B of acquired IPRD bought mostly early-stage assets. We read the answer as valuation discipline, and we think that is a defensible choice that the market is not rewarding.

What They're NOT Saying

1. The second royalty roll-off. The 10-Q discloses that the Keytruda and Tecentriq royalty agreements run through December 31, 2026; the two streams contributed $166M in the third quarter and $499M over nine months. Neither the prepared remarks nor the answer to a direct question about 2027 pushes and pulls mentioned them. At the current run-rate the expiry removes about $665M of pre-tax income in 2027, in the same year as the Eliquis step-down.
2. Zenbexus launch revenue. The first CELMoD is reported inside Other Growth Products, which grew 12% to $731M. Neither the release, the 10-Q nor the call broke out its contribution, and the launch commentary stayed qualitative. The product's commercial thesis is substitution for Revlimid and Pomalyst at branded pricing, and still no pricing or gross-to-net framework has been offered.
3. Why the implied fourth quarter is so weak. The guide implies fourth-quarter EPS of $1.48 at the midpoint against $1.93 before the item in the third quarter, with revenue down 4.6%, operating expenses up 9.3% and gross margin near 67.9%. Some of that is European Eliquis generics and fourth-quarter expense seasonality. None of it was explained, and no analyst asked.
4. The size of the contract-termination charge. On the call, gross margin reflected product mix "and to a lesser extent a 1 time expense." The 10-Q filed the same day puts the charge at $140M, or about 100bps of the quarter's gross margin, without naming the contract or the counterparty. The more important disclosure in the same 10-Q sentence, that the cost increase was primarily Eliquis alliance profit sharing, was described on the call as product mix.
5. Admilparant's secondary endpoints. Asked directly about time-to-event and mortality data, the Chief Medical Officer confirmed the company had not disclosed them and would present at a medical meeting next year. The monotherapy and add-on subgroups were also withheld. With the higher dose having missed, the secondary data carry most of the commercial argument.
6. The status of the April rebate promise. The CFO said in April that lower rebate payments would more than offset lower net collections later in the year. The cash-flow statement shows progress, with the rebate drain narrowing to $734M in the third quarter from $3,135M in the first half, but nine-month operating cash flow is still down 28.8%. The 10-Q describes the reduction in gross-to-net payments as a partial offset. No update was offered on the call.
7. The unused authorization. About $5B of repurchase capacity remains, the nine-month cash-flow statement shows no repurchases, and the diluted share count rose 0.7% year over year. The capital-allocation slide lists the authorization; the call discussed only the dividend and business development.
8. IRA exposure beyond Eliquis. The 10-Q notes that Pomalyst's Medicare maximum fair price takes effect on January 1, 2027 and that Orencia, at $983M a quarter, was selected in January 2026 for government-set prices from 2028. Neither was discussed on the call, although Orencia is the company's third-largest product by quarterly revenue.

Market Reaction

• Pre-print setup: BMY closed at $60.57 on October 28, up 12.3% year to date against the S&P 500's 13.1%, up 42.2% over the trailing twelve months and down 3.6% over the trailing thirty days. The 52-week closing range was $42.60 to $68.09, and the stock arrived 11.0% below its September 3 high after a run of pipeline news that included a sell-side downgrade on October 5 (down 3.8%), the mixed admilparant readout on October 12 (down 2.6%) and the zola-cel disclosure on October 23 (down 2.4%).
• Reaction session: The company reported before the open. The stock opened at $61.00, a 0.7% gap higher, traded between $60.00 and $61.56, and closed at $61.39, up 1.4% or $0.82.
• Volume: 14.8M shares against a 13.5M thirty-day average, or 1.1 times normal. A routine session.
• Relative: The S&P 500 fell 0.3% on the day, so the move beat the index by about 170bps. The healthcare sector ETF rose 2.4% on a day of large-cap pharma prints, so the stock lagged its sector by 100bps.
• Follow-through session: On October 30 the stock rose another 1.8% to $62.48 while the S&P 500 fell 1.8% and the sector ETF rose 0.5%. Across the two sessions the stock gained 3.2%.

The reaction was muted for a reason that has little to do with the quarter. Every major pipeline event in the release had been announced between August and October, and the market had already priced them, net negative. Between our July note and the day before the print, the stock fell from $64.86 to $60.57 while the healthcare sector rallied on a rotation into large-cap pharma. The print added a revenue raise and a $0.15 improvement in EPS before the item, and against a sector up 2.4% that bought a 1.4% gain.

The second session was more informative than the first. With the S&P 500 down 1.8%, the stock rose 1.8%, a relative move of more than three and a half points, on volume close to the first day's. That pattern fits investors working through a headline EPS cut that turned out to be an accounting effect of business development, and a revenue guide that rose for the second consecutive quarter.

The bigger picture is the gap since July. From the July 30 close to October 30, BMY fell 3.7% while the healthcare sector ETF rose 13.1%, Merck rose 15.2%, AbbVie 14.6%, Pfizer 22.5% and Lilly 22.7%. In the same period our 2026 estimate before the item rose 2.7% and four registrational readouts came in positive. A stock can lag a sector rotation for good reasons, and the 2027 to 2028 exclusivity exposure is one. A 16.8-point underperformance while the fundamentals improved is also what an entry point looks like.

Street Perspective

Debate: Is the pipeline finally being priced, or is it being ignored?

Bull view: The constructive camp points to four positive registrational readouts in two months, an approved first-in-class CELMoD with positive survival data, and a second CELMoD with a May PDUFA date, all while the stock fell. On this reading the market is pricing BMY as an Eliquis cliff and nothing else, and the late-stage pipeline is now de-risked enough to deserve some value before the milvexian data.

Bear view: The skeptics answer that the wins are real but small against the gap. RYZ101 is a modest opportunity by management's own description, arlo-cel addresses a late-line population, admilparant missed at its higher dose, and Zenbexus mostly recaptures revenue already lost to generics. The asset that can actually replace Eliquis is milvexian, and its two readouts are still ahead.

Our take: The bears are right about scale and the bulls are right about the price. None of the four readouts closes the 2028 gap, and we carry none of them in our numbers. The point is that a stock trading at 8.7 times 2026 earnings before the item, with a 4% yield, now has fewer ways to disappoint on the pipeline than it had in July, and has de-rated anyway.

Debate: How steep is 2027?

Bull view: The optimistic case is that the company has guided the Eliquis step-down three times without revising it, the Revlimid and Pomalyst cliff is nearly complete at $433M a quarter, and a $2B productivity programme lands by the end of 2027. With the Growth Portfolio adding close to $3B a year, 2027 can look like a modest decline rather than a cliff.

Bear view: The bear camp notes that 2026 earnings before the item are already guided down 6.6% from 2025, the implied fourth quarter is 23% below the third, and 2027 adds the Eliquis step-down, the European generic year, the Pomalyst Medicare price and the end of the Keytruda and Tecentriq royalties. On this reading the multiple looks low only because the earnings base is about to fall again.

Our take: The bears have the direction and the bulls have the size roughly right. We expect the February guide to show a decline in 2027 earnings, and the royalty expiry is the item most likely to surprise. Our valuation absorbs it: removing the full royalty from our base still leaves an expected total return of about 11%, above a market return.

Debate: Is business-development restraint discipline or drift?

Bull view: One argument circulating is that restraint at peak deal valuations is exactly right for a company with a strong balance sheet. Net debt is down $4.1B this year, cash is $13B, and the dividend is covered twice over by free cash flow, so a transaction can be done when prices are better.

Bear view: The counterargument is that the window is closing. Every quarter without a near-term growth asset makes 2028 harder, peers have been spending, and the company has booked $1.65B of acquired IPRD this year, much of it for early-stage assets that cannot contribute before the next decade.

Our take: We side with discipline, with a time limit. The balance sheet now supports a meaningful deal without stress, and management has named strengthening the near-term growth profile as a criterion. If 2027 passes without such a transaction or a buyback, the unused capacity becomes a cost rather than an option.

Model Update

The table shows what this quarter changed against the working assumptions we set in July.

Valuation basis. We now apply our multiple to EPS before acquired IPRD and licensing income. On July's numbers the distinction was immaterial, $6.94 against $6.96. On this quarter's it is $0.58, and capitalizing a one-time licensing payment at nine to ten times earnings would subtract about $5.70 a share of value at the midpoint multiple for roughly $1.1B of cash paid in the quarter, about $0.54 a share. The cash cost is already reflected in net debt. On the July basis, the stock traded at 9.3 times $6.96 at the July reaction close; it now trades at 8.7 times $7.15.

Valuation. At the October 30 close of $62.48, on 2,053M diluted shares, the market capitalization is $128.3B and enterprise value is $158.2B including $29.92B of net debt. That is 8.7 times our $7.15 estimate before the item, 9.5 times our $6.55 headline estimate and 3.15 times enterprise value to our revenue estimate. The $0.63 quarterly dividend, $2.52 annualized, yields 4.0%.

Valuation impact: Applying an unchanged 9.0 to 10.5 times to $7.15 gives a fair-value range of $64.35 to $75.10 with a $69.70 midpoint, against $62.50 to $72.90 and a $67.70 midpoint in July. The midpoint implies about 11.6% of price appreciation plus the 4.0% yield, or about 15.6% of expected twelve-month total return, against about 8.3% in July. Deducting the full Keytruda and Tecentriq royalty from the earnings base, about $2.60 a share at the midpoint multiple, lowers the midpoint to roughly $67.10 and the expected return to about 11.3%. Both figures sit above what we would call a market return. The fair-value midpoint falls to the current price only if earnings before the item fall to about $6.41, a decline of more than 10% from our 2026 estimate.

Thesis Scorecard Post-Earnings

The Q2 checklist, graded

Our July rating action set out the conditions for adding: weakness toward the low $60s, and either the admilparant readout in hand or a 2027 guide that does not step down. The stock closed at $62.48 on October 30, after trading as low as a $56.67 close in mid-October. The admilparant readout is in hand and met its primary endpoint at the Phase 2 dose. The third condition, a 2027 guide without a step-down, has not been tested and is not needed for the first two to apply. We are honouring the conditions as written.

Standing thesis

Overall: The standing thesis is unchanged in direction and stronger in its catalyst pillar. Bull 2, which the July slips had put at risk, recovered on four registrational wins and held timing on the one readout we said would break it. The bear points are all intact: margin compression arrived as the guide said it would, earnings quality below the operating line remains weak, and the 2027 headwinds now include a royalty expiry we had not modelled. What changed most is the price. The stock fell 3.7% while our estimate rose 2.7% and the sector rose 13.1%, so the same thesis now sits on a cheaper entry point.

Action: Upgrade to Outperform from Hold, with conviction at 6 from 5. The two conditions we set in July for adding have been met, and we see about 15.6% of expected twelve-month total return, or about 11% after deducting the full royalty roll-off, against a market return in the high single digits. This is a valuation call on a business whose 2027 earnings are likely to decline, not a call that the decline is avoidable, and we carry no pipeline value in it. We would revisit the rating if the milvexian stroke readout fails or slips past year end, or if the February 2027 guide implies earnings before the item below about $6.40, the level at which our fair-value midpoint meets the current price. If the stroke readout is positive and the 2027 guide lands above that level, nothing in this framework argues for a lower rating.