← Back to the report

Akeso, Inc.

Healthcare · Generated 16 November 2026

Akeso, Inc. (9926.HK): Deep Dive Research Report

Healthcare / Biopharmaceuticals. Report date: 16 November 2026.

Reporting-period check. Akeso's fiscal year ends 31 December, and it reports half-yearly on the Hong Kong Stock Exchange. HKEX rules require interim results within two months of 30 June, so the latest period was due by about 31 August 2026. It was released on 27 August 2026 (HKEX interim announcement), and the results call followed on 28 August. That is within the last 90 days and anchors this report. The six results events used are:

  • FY2023 (announced 18 March 2024)
  • H1 2024 (late August 2024)
  • FY2024 (30 March 2025, call 31 March 2025)
  • H1 2025 (call 28 August 2025)
  • FY2025 (26 March 2026, call 27 March 2026)
  • H1 2026 (27 August 2026, call 28 August 2026)

One event since then changes the picture. On 15 November 2026, one day before this report, the US FDA rejected ivonescimab in the United States, the licence partner's first US filing, with a Complete Response Letter (CRL).


1. What the Company Does

Akeso is a Chinese drug developer that designs antibodies, makes them in its own plants in Guangdong, and sells them to Chinese hospitals through its own sales force. Most antibody drugs grab one target on a cell. Akeso's specialty is antibodies built to grab two targets at once (bispecifics). It has taken two of these from the lab to approval, and both were the first of their kind approved anywhere:

  • Cadonilimab blocks PD-1 and CTLA-4, two "brakes" on the immune system.
  • Ivonescimab blocks PD-1 and also neutralises VEGF, the signal tumours use to grow new blood vessels.

Selling these two drugs inside China is where nearly all of the company's revenue comes from today. Licensing them abroad is where most of its long-term option value sits.

History and milestones. Per its own history and its Wikipedia entry, Akeso was founded on 19 March 2012 in Zhongshan, Guangdong by four scientists:

  • Michelle Xia (Xia Yu), now Founder, Chairwoman, President and CEO
  • Baiyong Li (Li Baiyong), now Executive Vice President and Chief Scientific Officer
  • Max Wang (Wang Zhongmin Maxwell), now Senior Vice President
  • Peng Zhang (Zhang Peng), now Senior Vice President

The company's management-team page lists all four as co-founders. It says Dr Xia previously worked at PDL BioPharma, Bayer, Crown Bioscience and the University of Louisville School of Medicine, but gives no dates for any of those roles. Another senior vice president on the same page, Dennis Xia (regulatory, manufacturing and quality), shares her surname. He is not listed as a founder, and the page states no relationship between them.

The early company was a discovery shop. In 2015 it licensed a CTLA-4 antibody, now called quavonlimab, to Merck & Co. That deal is the first sign that Western pharma valued its antibody engineering. It listed in Hong Kong in April 2020.

Three decisions then turned it from a licensor into an operating drug company:

  1. Commercialising at home rather than licensing. Akeso kept China rights to its bispecifics and built its own manufacturing and sales force. Cadonilimab was approved in China in 2022.
  2. The December 2022 ivonescimab licence to Summit Therapeutics. Summit, a US-listed company, took rights to the US, Canada, Europe and Japan for US$500 million upfront and up to US$5 billion in total, plus low double-digit royalties (Akeso, December 2022). This deal is the reason Akeso booked its only profitable year (2023): the licence income was a one-off (FY2023 results).
  3. Choosing to test ivonescimab head-to-head against Merck's Keytruda (pembrolizumab), the world's dominant PD-1 drug, rather than against chemotherapy alone. When the HARMONi-2 trial beat Keytruda on progression-free survival (PFS: how long patients live before the disease worsens) in 2024, Akeso became a global name almost overnight. Ivonescimab had been approved in China on 24 May 2024.

The problem it solves. PD-1 drugs such as Keytruda transformed cancer care, but most patients still do not respond or eventually relapse, and many tumours are "cold" to the immune system. Akeso's bet is that one molecule hitting two complementary pathways does better than one pathway alone, and better than two separate drugs given together. For ivonescimab, the logic is that VEGF is concentrated in tumours. An antibody that binds VEGF and PD-1 together should concentrate its immune effect where it is needed and spare normal tissue. The buyers are oncologists treating lung, gastric, cervical and biliary cancers, together with China's state insurer, which decides whether patients can afford the drug.

Why it is hard to copy. Making a stable, manufacturable antibody with four binding arms (Akeso calls its format "Tetrabody") is hard engineering. Proving it works is harder still: that takes multiple randomised Phase III trials against the current best drug. Section 3 covers the detail.

An example. A newly diagnosed Chinese patient has squamous non-small-cell lung cancer (NSCLC) that has spread. Until recently, the standard in China was a PD-1 antibody such as BeOne's tislelizumab plus chemotherapy. In HARMONi-6, published in The Lancet and presented at ASCO 2026, ivonescimab plus chemotherapy instead cut the risk of death by 34% against tislelizumab plus chemotherapy (ASCO Post, July 2026). The NMPA, China's drug regulator, approved that use in August 2026.

From there:

  • The hospital's oncology department adds the drug to its formulary.
  • China's clinical-oncology society (CSCO) lists it in its guidelines.
  • Insurance coverage determines what the patient pays.
  • Akeso's medical representatives educate the prescribing doctors.
  • The drug is made in Zhongshan and shipped to the hospital through distributors.

2. Business Segments

Akeso is effectively a single business: discovering, making and selling innovative antibody drugs. The franchises worth separating economically are:

  • Oncology (the engine). Two first-in-class bispecifics, cadonilimab and ivonescimab, plus the PD-1 antibody penpulimab. Oncology is almost all of product sales. The last period with a disclosed product split was H1 2024:

    • cadonilimab: about three-quarters of sales;
    • ivonescimab: about a tenth, after only one month on the market;
    • everything else: the remainder (H1 2024 results).

    Since then, ivonescimab has gained the large first-line lung indications. Management said its sales rose more than tenfold after the first-line lung indication entered national reimbursement (H1 2026 call coverage). The mix is now shifting towards ivonescimab, but no current split has been published.

  • Non-oncology specialty (the diversifier). Treatments for psoriasis, cholesterol and atopic dermatitis:

    • ebdarokimab (IL-12/IL-23)
    • ebronucimab (PCSK9)
    • gumokimab (IL-17), approved June 2026
    • manfidokimab (IL-4Rα)

    These sell in crowded, price-sensitive classes against rivals from Johnson & Johnson, Novartis, Amgen and Sanofi/Regeneron, plus domestic biosimilars. Its core capability is reusing Akeso's antibody engineering and manufacturing in markets the oncology sales force does not reach. Management built a dedicated metabolic/autoimmune sales team for it (FY2024 results). It is a small share of sales.

  • Global partnering (the option). Licence and collaboration income, mainly from Summit, plus Akeso's equity stake in Summit. It received 10 million Summit shares as part of the 2022 deal. This line is lumpy: large in 2023, negligible since. In 2025, revenue other than product sales was under 1% of the total (FY2025 results, AASTOCKS summary).

Geography. Essentially all product revenue comes from mainland China. Ivonescimab and cadonilimab are sold through Chinese hospitals and the "dual-channel" designated pharmacies. Akeso has no commercial presence of its own in the West. Its overseas exposure runs through three channels:

  • Summit's ivonescimab programme;
  • the April 2025 FDA approval of penpulimab, Akeso's first US approval;
  • international trials it runs itself (for example, cadonilimab in gastric cancer against Bristol Myers Squibb's nivolumab, and in liver cancer that has stopped responding to immunotherapy).

Akeso keeps ivonescimab rights outside Summit's territories: China and the rest of Asia (excluding Japan), plus South America, the Middle East and Africa. Summit's own filings describe Summit's territory more broadly, as North America, South America, Europe, the Middle East, Africa and Japan (AstraZeneca/Summit, September 2026), which suggests the territory was enlarged after the original 2022 deal. Either way, Akeso's international economics outside Asia are milestones and royalties, not product sales.


3. Products and Operations

The commercial portfolio. Akeso says it has eight commercially available drugs. Two more drugs, covering three indications, are under regulatory review, and five self-commercialised drugs covering 12 indications are on China's National Reimbursement Drug List (NRDL) (H1 2026 results). The ones that matter are below.

Ivonescimab (AK112; China brand Yidafang; Summit code SMT112), PD-1 x VEGF

  • Approved in China for:
    • EGFR-mutant non-squamous NSCLC after EGFR-TKI targeted pills fail (May 2024);
    • first-line PD-L1-positive NSCLC as a single agent (2025);
    • first-line squamous NSCLC with chemotherapy (August 2026).
  • Under priority review: first-line biliary tract cancer, granted on 4 November 2026 (PR Newswire).
  • Pivotal evidence:
    • HARMONi-2 beat Keytruda on PFS, then on overall survival (OS) at the September 2026 lung-cancer world conference (WCLC). Median OS was 30.75 months against 22.57 months (HR 0.73), the first OS win for any drug over Keytruda monotherapy (IASLC).
    • HARMONi-A showed OS and PFS benefit in EGFR-resistant disease.
    • HARMONi-6 is described in Section 1.
    • HARMONi-GI1 beat AstraZeneca's durvalumab regimen on OS in biliary cancer.
  • Guidelines: CSCO's 2026 lung guideline upgraded first-line monotherapy to its top (Class I) recommendation (Akeso, April 2026).
  • Reach: about 70,000 patients treated by end-2025 (FY2025 results).
  • Development: more than 17 registrational Phase II/III studies, seven of them global.

Every positive Phase III above was run only in China. The one global trial that has read out, Summit's HARMONi, met PFS but missed statistical significance on OS. That gap matters in Sections 8 and 9.

Cadonilimab (AK104; China brand Kaitanni), PD-1 x CTLA-4. Approved for:

  • second/third-line cervical cancer;
  • first-line cervical cancer in all patients, including PD-L1-negative ones (2025);
  • first-line gastric cancer (2024).

All of these are reimbursed. About 120,000 patients had been treated by end-2025. Its pitch is that it brings CTLA-4's extra immune activation without the toxicity that limits combining two separate antibodies, and that it works in PD-L1-negative tumours, where single PD-1 drugs struggle. It has 13 or more registrational/Phase III studies, including international ones in gastric cancer (against nivolumab) and liver cancer.

Penpulimab (PD-1). A conventional single-target antibody, approved in China and, since April 2025, by the FDA for nasopharyngeal carcinoma. It matters mainly as Akeso's first US regulatory precedent.

The next wave ("IO 2.0 + ADC 2.0")

  • Antibody-drug conjugates (ADCs), which carry a chemotherapy payload into the tumour. Four are in the clinic:

    • AK138D1 (HER3)
    • AK146D1 (a TROP2/Nectin-4 bispecific ADC)
    • AK157D1 (B7-H3)
    • AK158D1

    AK138D1 and AK146D1 are in Phase II combinations with the two bispecifics.

  • Combinations with partners' ADCs. Pfizer's (agreed February 2025, with Akeso supplying ivonescimab), GSK's, Biokin's and MediLink's.

  • Immunology and neurology:

    • AK139, an IL-4R/ST2 bispecific in seven Phase II trials;
    • AK152, an anti-amyloid bispecific with a brain-shuttle arm for Alzheimer's;
    • AK150, an ILT2/ILT4/CSF1R trispecific;
    • AK154, a personalised mRNA cancer vaccine.

The company counts more than 50 assets, around 30 of them in clinical trials (H1 2026 results).

What makes the products hard to make

  • Engineering. Bispecifics must hold two different binding sites in a stable shape and resist clumping (aggregation). Ivonescimab is designed to bind PD-1 more tightly when VEGF is present, which is what concentrates its effect in tumours. That property comes from the molecule's design and cannot be reproduced simply by co-dosing two drugs.
  • In-house process development. The company says it does 100% of process development itself and has taken more than 50 clinical candidates through it (management page). That speed is why Akeso can start so many trials at once.

Manufacturing. All biologics are made in Guangdong using single-use (Cytiva/GE FlexFactory) and stainless-steel bioreactor lines (Akeso manufacturing). The value chain runs: discovery (Zhongshan/Guangzhou, plus a Shanghai R&D centre being built for about RMB 850 million), cell-line and process development, bioreactor production, fill-finish, release testing, distributors, then hospitals and pharmacies.

SiteRoleCapacity (company figures)
National Health Technology Park, ZhongshanOriginal FlexFactory site3,000 L; being upgraded to make ADCs
Sino-Singapore Guangzhou Knowledge CityThree units (Akeso Pharma, AD Pharmaceuticals, Akeso Longyue)Started with 36,000 L; Akeso Pharma planned to 60,000 L; total investment over RMB 3 billion
Akeso Greater Bay Area Technology Park, ZhongshanMain commercial plant, fully running since March 2024Four 10,000 L stainless-steel reactors; planned capacity above 100,000 L; RMB 2.5 billion investment

The company says its quality system is designed to NMPA, FDA and EMA GMP standards. Under the Pfizer collaboration, Akeso is responsible for producing ivonescimab (FY2024 results), so the Zhongshan plants are a strategic asset for partnered global trials as well as for China. No capacity constraint has been reported. Capacity is ahead of demand, and the binding constraints are reimbursement and clinical evidence, not plant output.


4. Customers

The buyers. Physically, Akeso sells to pharmaceutical distributors, who deliver to hospitals and "dual-channel" designated pharmacies. Economically, three parties make the decision:

  1. The National Healthcare Security Administration (NHSA), China's state insurer. Through the annual NRDL negotiation it decides whether a drug and each indication is reimbursed, and at what price. It is the most powerful single customer. Its criteria are clinical value per yuan, comparative evidence, and budget impact. The cycle is annual: drugs approved by mid-year are negotiated in autumn, and the list takes effect on 1 January. Akeso's December 2025 outcome put every approved indication of its five self-marketed drugs on the list from 1 January 2026 (Akeso, December 2025).
  2. Hospital pharmacy and therapeutics committees. Each hospital must "list" a drug before its doctors can prescribe it routinely. Committees weigh guideline status, budget caps under China's diagnosis-based payment reforms (DRG/DIP), and doctor demand. Listing takes months to years per hospital, which is why management treats "hospital access" as a key indicator. At the FY2024 call, Xia said 80% of core hospitals were covered and set a goal of more than 2,000 hospitals by end-2025 (NBD, 31 March 2025). By the FY2025 results, the two bispecifics were listed in more than 3,000 hospitals, with about 90% of them having in-hospital or dual-channel access.
  3. Oncologists, especially the opinion leaders who write the CSCO guidelines. Their criterion is survival evidence against the current standard. Akeso's head-to-head trials against Keytruda, tislelizumab and durvalumab are aimed squarely at them. The commercial team grew from 844 to 1,221 in the year to mid-2025 (H1 2025 call coverage) and to more than 1,600 by mid-2026 (H1 2026 coverage).

Patients themselves were the payers before 2025. Both bispecifics then sold mainly outside hospitals at full cash price, which capped volume. NRDL inclusion from January 2025 moved the business into hospitals, swapping lower prices for much larger volume.

Why doctors choose Akeso

  • In several settings, its drugs are the only ones with randomised survival data against the incumbent PD-1 standard.
  • Cadonilimab works in PD-L1-negative patients, where single PD-1 drugs struggle.
  • Ivonescimab can replace "PD-1 plus bevacizumab" combinations with one infusion.
  • Their top guideline ratings make them the default choice in their indications.

Switching costs. These are low at the level of the individual patient: a doctor can prescribe a rival PD-1 the next day. They are high at the level of the system:

  • a rival needs its own NRDL listing for the same indication;
  • it then needs its own approvals at each hospital, which take years;
  • above all, it needs its own Phase III evidence, which takes years and hundreds of millions of dollars.

Once a drug is guideline-endorsed and listed in a hospital, inertia favours it until a better-proven alternative arrives. The lock-in, in other words, belongs to the evidence base, not to the customer.

Concentration. Akeso has not disclosed individual distributor concentration in the sources reviewed. The real concentration is structural: one payer (NHSA) and two products. A single NRDL renegotiation or price cut flows straight into sales. Management blamed the cadonilimab price cut that came with the 2025 renegotiation for slowing reported growth in H1 2026 (Section 9).

Contract structures

  • China product sales: recurring and volume-driven, with prices reset at each NRDL renewal or indication expansion.
  • Partner income: milestone-based and lumpy. Summit's upfront payment was the 2023 windfall. Regulatory and commercial milestones, worth up to US$4.5 billion in total, plus royalties are contingent on approvals and sales outside China, and the 15 November 2026 CRL has pushed them back.

As a result, revenue is fairly predictable within a year but resets at each NRDL cycle, and partner income is close to binary.


5. Competitive Landscape and Moat

The competitive structure differs by layer.

Layer 1: China PD-1 incumbents (the share Akeso is taking). China's PD-(L)1 market is crowded with homegrown antibodies that have been through repeated NRDL price cuts:

  • BeOne's tislelizumab, the market leader by sales;
  • Innovent's sintilimab;
  • Hengrui's camrelizumab;
  • Junshi's toripalimab;
  • Merck's Keytruda.

Akeso wins here on evidence. HARMONi-6 beat tislelizumab plus chemotherapy, and HARMONi-2 beat Keytruda, in randomised China trials. Its bispecifics compete as an upgrade to a PD-1 drug, not as another PD-1 drug. The incumbents are fighting back with their own next-generation assets, for example Innovent's IBI363 (PD-1/IL-2) and many China-origin PD-1/VEGF candidates. Akeso's lead in China is a head start of years, not a permanent position.

Layer 2: The global PD-(L)1 x VEGF race (where the long-term value is decided). Ivonescimab created this class, and big pharma has since paid heavily for fast followers:

  • Pfizer licensed 3SBio's SSGJ-707 (PF-08634404) in 2025: US$1.25 billion upfront, up to US$4.8 billion in milestones. It registered two global Phase IIIs in October 2026, in NSCLC and colorectal cancer (Pfizer).
  • BioNTech and Bristol Myers Squibb are running global Phase IIIs of pumitamig (BNT327, PD-L1 x VEGF).
  • Others reviewed at ASCO 2026 include Huabo's HB0025 and SCTB-14 (PMC review).

These competitors have far bigger development budgets than Summit and global sales forces of their own. Ivonescimab's advantage is its lead: the most patients treated, the most Phase III data, and the only OS wins against PD-1 standards. The CRL has, however, narrowed its lead in time on the US market.

Two developments in 2026 changed the shape of this race:

  • AstraZeneca bought a US$2 billion equity position in Summit in September 2026 and agreed to combine its ADCs with ivonescimab (AstraZeneca). Ivonescimab now has a large-pharma ally, though not a buyer.
  • The FDA's demand for statistically significant OS (Section 8.1) now applies to every member of the class. That favours sponsors whose trials are powered for OS, and Summit's ongoing trials are not obviously behind on that.

Layer 3: PD-1 x CTLA-4. Cadonilimab competes with the two-drug combination of nivolumab and ipilimumab and with newer bispecifics. In 2026, AstraZeneca's volrustomig failed a Phase III, which led some commentators to declare the PD-1/CTLA-4 bispecific class dead. Xia rejected that at the H1 2026 call, arguing that cadonilimab's four-armed Tetrabody structure makes it a different product (Section 9). Cadonilimab still has no approved bispecific competitor anywhere, and it has a Merck connection of its own: its CTLA-4 heritage relates to quavonlimab, the antibody Akeso licensed to Merck in 2015.

CompetitorCountryListingApprox Market CapProduct OverlapRelative Strength
Merck & Co.USANYSE: MRK~US$364bn (Nov 2026) (source)Keytruda, the PD-1 standard ivonescimab is tested againstGlobal incumbent; lost HARMONi-2 on PFS and OS in a China-only trial
BeOne MedicinesSwitzerland (China-rooted operations)Nasdaq: ONC; HKEX: 6160~US$47.6bn (Nov 2026) (source)Tislelizumab, China PD-1 leaderLost HARMONi-6 on OS; deep hospital reach
Jiangsu HengruiChinaSSE: 600276; HKEX: 1276~US$45-50bn (Nov 2026) (source)Camrelizumab; broad oncologyLargest domestic sales machine
Innovent BiologicsChinaHKEX: 1801~HK$148-154bn (Nov 2026) (source)Sintilimab; next-generation IO assetsStrong commercial platform; fast follower
3SBio (partnered with Pfizer)ChinaHKEX: 1530~HK$34-40bn (Oct 2026) (source)SSGJ-707 PD-1 x VEGF, in Pfizer's global Phase IIIsPfizer's budget and reach outside China
BioNTech (partnered with BMS)GermanyNasdaq: BNTX~US$21.6bn (Nov 2026) (source)Pumitamig PD-L1 x VEGFBMS's global oncology reach

Barriers to entry

  • Clinical: Phase III programmes powered for OS take years and hundreds of millions of dollars.
  • Manufacturing: building GMP bispecific capacity at 100,000 L scale.
  • Regulatory and payer: NMPA approval, NRDL negotiation and hospital listings.

In China these barriers are high for start-ups, but moderate for the six or so large domestic companies that already have PD-1 drugs and sales forces. Globally, the barrier is the speed of evidence, and money buys speed.

Market share. No audited class shares were found in the sources reviewed. Akeso owns 100% of the approved PD-1 x VEGF and PD-1 x CTLA-4 bispecific classes in China. That is because it is first to market, not because it has a structural monopoly.

Moat: Narrow - intangible assets (patents and a proprietary randomised evidence base), regulatory/qualification barriers

Akeso's advantage is that its drugs have beaten the incumbent PD-1 standards on survival in randomised trials and are embedded in guidelines, the NRDL and more than 3,000 hospital formularies (Section 4). A rival needs years and Phase III money to match that. It is not Wide. The bispecifics have not yet held their position through a well-funded attack, the evidence is still almost entirely Chinese, and the main pricing customer (NHSA) can cut prices at every renewal. The single biggest threat to the moat is a big-pharma PD-(L)1 x VEGF rival (Pfizer/3SBio or BMS/BioNTech) producing global OS data before ivonescimab does. That would turn Akeso's head start into a regional story.


6. Industry

Demand drivers. China has one of the largest cancer burdens in the world, with lung cancer the biggest killer. Three things drive demand for immuno-oncology drugs:

  • Patients: an ageing population and earlier diagnosis.
  • Payers: expansion of NRDL coverage, which turns cash-pay drugs into reimbursed ones.
  • Evidence: guideline upgrades as trials read out.

Industry estimates put China's PD-(L)1 market at roughly US$5-7.5 billion in 2025, and the estimates vary widely by source (PharmaSources). Globally, the market is dominated by Keytruda and growing at mid-teens rates. Next-generation bispecifics aim to become the new backbone that other drugs, especially ADCs, are combined with. That is why Pfizer, BMS, Merck and AstraZeneca have all paid for exposure to PD-(L)1 x VEGF.

Position in the supply chain. China has moved from making generics and copies of Western drugs to inventing them. In 2025, Chinese companies were the source of a large share of global licensing deals for new molecules, and the most expensive ones were in oncology bispecifics. Akeso sits at the top of that value chain, as originator, and keeps China rights while licensing out the West. That model pays well in milestones but leaves ex-China execution to partners.

Import substitution. In China's PD-1 market this is complete: domestic antibodies displaced imported ones once they entered the NRDL at lower prices. The next substitution is happening inside the domestic market, as bispecifics displace single-target PD-1s on survival evidence.

Regulation

  • China. The NMPA now offers priority and conditional pathways, and in 2024-2026 regularly approved drugs on China-only Phase III data. The NHSA runs the annual NRDL negotiation, which grants national coverage in exchange for steep price cuts and resets prices at renewal or when an indication is added. The 2026 list was published in mid-November 2026 and takes effect on 1 January 2027 (Innovent's announcement). Hospital spending is increasingly governed by diagnosis-based payment (DRG/DIP), which pressures expensive drugs that lack clear value.
  • United States. The FDA has toughened its stance on two points. It is sceptical of applications that rest heavily on single-country foreign data. And, as shown by the November 2026 ivonescimab CRL, it holds that a PFS gain without statistically significant OS may not be enough in first-line and later-line NSCLC. That raises the evidence bar for the whole PD-(L)1 x VEGF class.

Cyclicality. Cancer drug demand is not economically cyclical. The "cycle" that matters is policy: NRDL price-cut intensity, hospital budget tightening, and the funding climate for Hong Kong biotech, which decides how cheaply companies can raise equity to pay for trials.

Tailwinds:

  • big pharma's appetite for China-origin molecules;
  • broadening NRDL coverage;
  • combining bispecifics with ADCs as the next standard.

Headwinds:

  • price compression at every NRDL renewal;
  • dozens of me-too PD-(L)1 x VEGF candidates;
  • tougher US evidence standards;
  • US-China political friction over biotech supply chains and data.

7. Growth Triggers

Ranked by potential impact. All are management statements or partner and company announcements.

  • Global OS readouts in first-line NSCLC (HARMONi-3 and HARMONi-7), the only way back into the US after the CRL.

    • HARMONi-3 squamous cohort: the PFS events for the primary analysis are expected in Q4 2026, with an early interim OS analysis at the same time.
    • An interim OS analysis independent of PFS is planned for H1 2027.
    • Non-squamous cohort: PFS events expected in H1 2027.
    • HARMONi-7 (single-agent ivonescimab against Keytruda in PD-L1-high NSCLC): enrolment completes in 2027.

    Summit's Q3 2026 release (November 2026) gives these timings (Summit Q3 2026). Akeso has repeated the global-programme framing at every results event since FY2024. The CRL release says development will continue across five further Phase III trials (Summit, 15 November 2026).

  • First-line squamous NSCLC launch in China. Approved in August 2026, with HARMONi-6's OS win behind it. The H1 2026 results call (28 August 2026) presented it as the new first-line lung indication added to ivonescimab's reimbursed base. It will need its own NRDL listing, and Akeso's outcome in the 2026 round was not found in the sources reviewed.

  • First-line biliary tract cancer approval in China. HARMONi-GI1 met its OS endpoint against durvalumab plus chemotherapy (ESMO, October 2026), and the NMPA granted priority review on 4 November 2026. This would be ivonescimab's first approval outside lung cancer.

  • "IO 2.0 + ADC 2.0" combinations. Four of Akeso's own ADCs are in the clinic, and AK138D1 and AK146D1 are in Phase II combinations with the bispecifics. Further trispecifics and T-cell engagers are expected to enter the clinic "next year" (H1 2026 call, 28 August 2026; H1 2026 results). Partner ADC combinations add to this: Pfizer, GSK, Biokin and MediLink, and through Summit, AstraZeneca's sonesitatug vedotin. The strategy has been repeated at every results event since H1 2025.

  • International development of cadonilimab. Two international registrational studies are under way: first-line gastric cancer head-to-head against nivolumab, and liver cancer after immunotherapy failure. Management said it plans to accelerate cadonilimab's global development (FY2025 results, 26 March 2026; H1 2026 call, 28 August 2026).

  • Non-oncology launches. Gumokimab (IL-17) was approved for psoriasis in June 2026. Manfidokimab (IL-4Rα) had positive Phase III atopic dermatitis results, and its approval filing (NDA) was being prepared (H1 2025 results). AK139 is in seven Phase II trials across respiratory and autoimmune diseases (H1 2026 results). These broaden the business beyond oncology and its NRDL exposure.

  • Manufacturing build-out. The Greater Bay Area park is planned to exceed 100,000 L, and the Zhongshan site is being converted for ADCs, so Akeso can supply both its own and partnered global trials (Akeso manufacturing). This supports the two triggers above rather than being a trigger in its own right.


8. Key Risks

8.1 The US path for ivonescimab now depends on overall survival, and the first test failed. The FDA's CRL of 15 November 2026 said substantial evidence of effectiveness was not established. Its reasoning was that HARMONi's PFS benefit, without a statistically significant OS improvement, did not support a positive benefit-risk profile (Summit). The later HARMONi analysis (OS HR 0.76, nominal p=0.0151) did not change the outcome.

The mechanism: regulatory milestones (up to US$1.05 billion in the original deal) and royalties now depend on HARMONi-3 and HARMONi-7 delivering statistically significant OS in global populations. HARMONi-3's squamous cohort had already missed a high-bar PFS interim in spring 2026 (ApexOnco, 1 May 2026), although the analysis used a stringent alpha and the trial continues as planned. Akeso's own announcement of 16 November 2026 named the CRL as the likely cause of unusual share-price moves (TipRanks).

Calibration: highly likely to delay the US opportunity and moderately likely to impair it. China-only OS wins (HARMONi-2, -6, -A, -GI1) show the drug works in Chinese patients. They do not guarantee the effect carries over to Western patients.

8.2 Better-funded fast followers take the global class. As Section 5 explains, Pfizer/3SBio and BMS/BioNTech are running global Phase IIIs with budgets far above Summit's. Summit raised money through its at-the-market (ATM) share sales in Q3 2026 even before the CRL. If a competitor's global OS data arrive first, or look cleaner, the first-mover value of ivonescimab outside China shrinks, and so do Akeso's royalties. Calibration: moderate probability, large effect on the long-term value of the licence. It does not affect the China business.

8.3 NRDL repricing compresses China economics. Each new indication brings the whole drug back to the negotiating table. Management attributed H1 2026's slower reported growth to the cadonilimab price cut from the 2025 renegotiation. It said internal growth excluding that cut exceeded 45% (H1 2026 call). Every new ivonescimab indication (squamous, biliary) carries the same trade of volume for price, and DRG/DIP budgets limit expensive regimens. Calibration: very likely, moderate in effect, and recurring every year.

8.4 Two products carry the company, and the PD-1/CTLA-4 class is under suspicion. Almost all sales come from cadonilimab and ivonescimab (Section 2). A safety signal, a guideline downgrade, or a rival's better data in one big indication would hit the whole company. AstraZeneca's volrustomig Phase III failure in 2026 led some commentators to declare the PD-1/CTLA-4 class dead. Xia argued at the H1 2026 call:

"We cannot deny an entire product, an entire class, because of AstraZeneca's failure in a single Phase III indication... cadonilimab is a tetravalent Tetrabody structure; it is not the same product." - Michelle Xia, H1 2026 results call, 28 August 2026 (translated from 21st Century Business Herald coverage)

Calibration: low probability that it fails outright, but the damage would be severe.

8.5 Funding needs and dilution. Akeso runs about a dozen Phase III trials while still losing money. Xia acknowledged in August 2025 that R&D spending had reached about 50% of sales, against a pharma norm of 20-30% (H1 2025 call coverage). The company has raised equity through placings three times since 2024, the latest at a discount in October 2026 (Section 10). The CRL makes milestone income less likely to fill the gap. If the Hong Kong biotech funding window shuts, Akeso will have to choose between cutting trials and issuing equity at lower prices. Calibration: likely to continue in the near term, moderate dilution risk.

8.6 The partner as a single point of failure, and the Summit stake. Akeso's Western economics depend on Summit's execution and balance sheet, now partly supported by AstraZeneca. Akeso also holds Summit equity. Losses on that investment contributed to Akeso's wider 2025 loss (AASTOCKS). A weaker Summit would delay trials and could lead to renegotiation. Calibration: moderate probability, moderate effect.

8.7 US-China policy friction. US scrutiny of China-originated drugs, and of trials that rely mainly on Chinese data, affects both FDA reviews and partners' appetite to license. Media framed the CRL as a rejection of a "China-originated" drug (Spokesman-Review, 15 November 2026). Calibration: low to moderate probability of direct barriers, and an ongoing drag on how the evidence is received.


9. Walk the Talk

The six results events used:

  • FY2023 results (18 March 2024)
  • H1 2024 (late August 2024)
  • FY2024 (call 31 March 2025)
  • H1 2025 (call 28 August 2025)
  • FY2025 (results 26 March 2026; call and slides 27 March 2026)
  • H1 2026 (results 27 August 2026; call 28 August 2026, which falls within 90 days of this report)

Full call transcripts are not published. Management statements come from the results announcements and Chinese press coverage of the calls, and FY2023 and H1 2024 are covered by announcements only.

What was guidedWhenWhat happenedVerdict
Ebdarokimab and ebronucimab to be approved in 2024-25H1 2024 resultsBoth approved and now NRDL-listedKept
Bispecifics in 2,000+ hospitals by end-2025FY2024 call, 31 March 2025Passed by mid-2025; 3,000+ by the FY2025 slidesKept, early
First-line cervical cancer (cadonilimab) and first-line PD-L1+ NSCLC (ivonescimab) approved in 2025FY2024 callBoth approved in 2025 and NRDL-listed from January 2026Kept
HARMONi global Phase III topline by mid-2025FY2024 resultsDelivered on time; PFS positive, OS not statistically significantKept on timing; missed on substance
HARMONi-7 global Phase III to start in 2025FY2024 resultsOngoing; enrolment completing 2027Kept
First-line squamous NSCLC approval in China on HARMONi-6H1 2025 (sNDA filed)Approved August 2026, with OS benefitKept
Overseas launch of ivonescimab in 2026 "if data are positive"FY2024 callFDA CRL, 15 November 2026Missed
Stable prices for the core first-line indications in 2026FY2025 results, 27 March 2026Five months later, the cadonilimab NRDL price cut was blamed for slower H1 growthPartly missed

The guidance it refused to give. At the FY2024 call, investors asked for sales guidance on the two bispecifics in their first year on the NRDL. Xia declined:

"Now is not the right time to give guidance, because the ramp-up has no ceiling. The numbers will definitely look good." - Michelle Xia, FY2024 results call, 31 March 2025 (translated from Tencent News coverage)

The hospital-access target she did give was beaten early, and sales growth in 2025 was the fastest since launch. Refusing to give numbers while hitting the operating milestones she did name is the strongest evidence in Akeso's favour.

The overseas promise. The same FY2024 call pointed to an overseas launch in 2026 if the data were positive. HARMONi delivered half of what was needed: PFS yes, OS no. Akeso management then repeatedly argued that the global data were consistent with China's. At the H1 2025 call, Xia stressed "high consistency between global and China data" (Tencent). The H1 2026 call presented a November FDA decision as the next milestone. The FDA did not accept consistency of effect in place of statistical significance on OS. Management's reading of the science may be right, since the later OS hazard ratio was 0.76, but its reading of the regulator was too optimistic.

The pricing message. In March 2026, management said prices for the core first-line indications had held stable. In August, it explained slower reported growth by pointing to the cadonilimab price cut. The two statements can both be true, because "core first-line" excludes some cadonilimab indications. But the second statement was needed to explain a result the first one did not prepare investors for.

Assessment. Akeso management delivers what it controls (approvals in China, hospital access, trial starts and Chinese readouts) on time and often early. It has been consistently optimistic about what it does not control, chiefly Western regulators and NRDL pricing, and it frames setbacks in the best light rather than flagging them in advance. Investors should trust its China operating targets and discount its global timelines.


10. Ownership, Governance and Shareholder Friendliness

Ownership and control. Akeso is incorporated in the Cayman Islands and has one class of ordinary shares. There is no dual-class structure or golden share. An HKEX filing showed founder Dr Xia Yu entitled to exercise voting rights over 225,013,453 shares, about 24.4% of the issued shares at that date (HKEX filing). Placings since then have diluted that percentage. The other three co-founders are also executive directors holding shares.

No parent company or state holder controls Akeso. With a roughly one-quarter founder voting bloc and a large institutional free float, the founder has decisive influence in ordinary general-meeting votes (which are often decided on low turnout) but cannot pass special resolutions alone. A minority shareholder's main exposure is to the board's freedom to issue new shares under the general mandate. It has used that freedom three times since 2024.

Governance. One person holds the roles of Chair, President and CEO: the founder (per the management page). All four founders sit as executive directors, so independent oversight rests on the independent non-executive directors.

Four points stand out:

  • Executive pay. It relies heavily on share incentives. On 24 May 2025, Akeso granted 7.115 million share options at an exercise price of HK$84.60 to 114 grantees, plus 3.42 million restricted share units (RSUs) to 107 grantees (Leju Finance). Whether vesting is tied to measurable performance targets could not be confirmed from the sources reviewed.
  • Related parties. No material related-party transactions with the founders were found.
  • Pledges and auditor. No share pledges by directors were found, nor any qualified or emphasis-of-matter audit opinion.
  • Investment in the partner. Akeso holds equity in its licensing partner Summit. This is a disclosed investment, not a related-party arrangement, but it ties the balance sheet to Summit's share price (Section 8.6).

Capital returns: dividends. Akeso has never paid a dividend. The DPS was nil for FY2023, FY2024 and FY2025, and the board recommended no final dividend for 2025 (FY2025 results). That fits a company that still loses money and has accumulated losses. FY2023's profit came from one-off licence income and was not distributed.

Capital returns: buybacks and dilution. Last 90 days (since mid-August 2026): MoatMap records no buybacks. Separately, the company announced that on 16 November 2026 the trustee of its 2021 RSU scheme bought 336,000 shares on market for about HK$21 million (TipRanks). That purchase used the June 2026 authority described next.

On 11 June 2026, the board approved spending up to HK$200 million to buy shares on market through the RSU-scheme trustee, saying the share price did not reflect intrinsic value (TipRanks). The total spent so far under that authority is not disclosed in the sources reviewed. Shares bought by the trustee are held for employee awards rather than cancelled, so they offset dilution from share schemes rather than reducing the share count. The AGM also renewed the standard mandate to repurchase up to 10% of shares (92,114,317 shares). No cancellation buybacks under that mandate were found for 2024-2026.

Dilution over the three years far outweighs these purchases. Akeso issued:

  • 24.8 million shares at HK$47.65 in March 2024 (HKEX);
  • 23.55 million at HK$149.54 in August 2025 (MarketScreener);
  • 20 million at HK$97.76, a 6.9% discount, announced 13 October 2026 (TipRanks).

Together with option and RSU issuance, the share count has risen by roughly 11-12% since the start of 2024, from about 0.84 billion to about 0.94 billion. That estimate is derived from the placing announcements and the 2026 repurchase-mandate share count.

Verdict: Hoards Capital. This is expected at Akeso's stage. It pays no dividend, has issued equity three times in under three years, and its only buyback is a modest trustee programme that offsets share-scheme dilution.


11. Insider Activities

MoatMap's HKEX disclosure-of-interests record (data current to 16 November 2026) is the sole source here. It shows nine transactions by five insiders in the past 12 months: three on-market purchases, six "Other" entries and no sales. The record runs to 16 November 2026, after the CRL, and shows no transactions after 16 October.

DateInsider (Role)TypeSharesApprox ValueNotes
2026-10-16Xia Yu (Chair/CEO)Other: decrease in interestn/an/aNo share count or reason disclosed in the record (HKEX DI, 2026-10-16)
2026-07-24Xia Yu (Chair/CEO)Other900,000n/aSame-day entries for five insiders, scaled by seniority; consistent with a share-incentive award, but the type is not specified (HKEX DI, 2026-07-24)
2026-07-24Li Baiyong (Executive Director, CSO)Other450,000n/aAs above (HKEX DI, 2026-07-24)
2026-07-24Wang Zhongmin Maxwell (Executive Director)Other130,000n/aAs above (HKEX DI, 2026-07-24)
2026-07-24Zhang Peng (Executive Director)Other90,000n/aAs above (HKEX DI, 2026-07-24)
2026-07-24Zeng Junwen (Director/Officer)Other15,000n/aAs above (HKEX DI, 2026-07-24)
2026-06-15Xia Yu (Chair/CEO)On-market purchase100,000~HK$9.4m at HK$94.07Under the June 11 pledge (HKEX DI, 2026-06-15)
2026-06-12Zhang Peng (Executive Director)On-market purchase60,000~HK$5.2m at HK$87.43Under the June 11 pledge (HKEX DI, 2026-06-12)
2026-06-11Xia Yu (Chair/CEO)On-market purchase100,000~HK$8.8m at HK$87.56Same day as the pledge (HKEX DI, 2026-06-11)

Buys. The three purchases total about HK$23.4 million in personal money from the founder-CEO and a co-founder director. Two insiders bought within five days, which makes this a cluster. But the purchases followed a public pledge on 11 June 2026 by the chair, executive directors and CFO to buy up to HK$50 million combined. That pledge was announced alongside the company's HK$200 million trustee programme as an explicit confidence signal (Section 10). About half the pledge appears in the record.

Relative to Xia's roughly one-quarter holding, 200,000 shares is a rounding error: about 0.02% of shares outstanding. The buying is genuine and was made with personal funds, but it was pre-announced and modest in size. That makes it a weaker signal than unprompted buying would have been.

Sells. None were recorded. The 16 October "decrease in interest" for Xia Yu has no share count or reason in the record. It is not identified as an on-market sale, and its cause is not disclosed.

Net assessment. Insiders are net buyers, but the buying is concentrated in two founders, came in one week in June, and was publicly announced in advance. The July entries look like routine incentive awards. No insider has bought or sold since the 15 November CRL. Read: mildly bullish signal. It is a real cluster of purchases with personal money, but the size is small for holders who already own this much, and the purchases came before the CRL.


12. Scenarios

Bull case. HARMONi-3's squamous cohort delivers a clean PFS win with a supportive early OS read in Q4 2026, and the non-squamous cohort and HARMONi-7 follow in 2027. Summit refiles with the OS evidence the FDA asked for, and AstraZeneca's ADC combinations make ivonescimab the backbone for its global combination programme. In China, squamous and biliary indications enter the NRDL, more than 1,600 representatives carry ivonescimab into nearly every major hospital, and HARMONi-2's OS win over Keytruda makes it the default first-line lung therapy. Cadonilimab's international gastric trial against nivolumab succeeds, putting to rest the doubt that volrustomig's failure cast on the class. In two to three years, Akeso has a large self-funding China business and a stream of milestones and royalties from the West. Its ADC and immunology pipelines then decide whether it becomes a multi-franchise drugmaker, and further dilutive placings are no longer needed.

Base case. China carries the business. New squamous and biliary indications add volume, the NRDL takes back part of each gain in price, and growth stays solid but uneven, as H1 2026 showed. Management keeps hitting its China milestones on time, as its record suggests. The US becomes a 2027-2028 question that depends on HARMONi-3 OS. Summit and AstraZeneca keep the global programme funded, but Pfizer's and BMS's PD-(L)1 x VEGF programmes catch up in time, so ivonescimab's eventual Western launch competes in a crowded class rather than defining it. Akeso keeps spending heavily on R&D. It may issue equity once more before breaking even, and it pays no dividends.

Bear case. Risks 8.1 and 8.2 compound. HARMONi-3's OS is positive in direction but not significant, as in HARMONi, so the FDA's position hardens. A Pfizer/3SBio or BMS/BioNTech asset then posts global OS data first, and ivonescimab's Western value shrinks to a minor royalty. Summit, under pressure, cuts trials (8.6). In China, 8.3 and 8.4 work together: each new indication brings another price cut, and a rival's PD-1/VEGF or PD-1/IL-2 data erode first-line share. Sales growth slows while R&D commitments stay high, and 8.5 forces further discounted placings in a tougher funding market. Akeso ends up a respected Chinese oncology company with two ageing bispecifics, a pipeline it must keep funding, and no Western revenue.


Sources:

Generated by MoatMap on 16 November 2026. Read it online at moatmap.ai/deep-dive/9926.HK; today's score is at moatmap.ai/stock/9926.HK.
Akeso, Inc. (9926.HK) Deep Dive - Nov 2026 | MoatMap