On September 18th, ten government departments including the Ministry of Industry and Information Technology released the “15th Five-Year Plan” for the pharmaceutical industry. The policy framework simultaneously incorporated technologies such as artificial intelligence, quantum computing, precise molecular delivery, cell programming, and gene editing, and the market responded immediately: XTALPI (02228.HK) closed up 16.32% at HKD 8.02 with turnover of HKD 1.251 billion, while other AI drug developers saw gains of over 12%. XTALPI ultimately led the three AI pharma companies higher, reflecting that capital attention has shifted from single algorithms to the integrated capability of computation, experimentation, delivery, and frontier modalities. Policy acts as a catalyst, but industry trends determine the ceiling on valuations. Today, the core question for evaluating AI drug companies has become: who can consistently produce differentiated assets, whose technology can be reused across projects, and who can select drug modalities with a higher probability of success across different targets?
The value of clinical leadership must be confirmed by subsequent data. The first generation of AI drug developers has already undergone a round of clinical attrition and industry consolidation. When Recursion acquired Exscientia, it disclosed over 10 clinical and preclinical programs; months later, the company compressed its focus portfolio to five or more programs, pausing or discontinuing three clinical and one preclinical pipeline, with REC-2282 and REC-994 both terminated due to insufficient clinical data. This history illustrates the full meaning of “move fast, fail fast”: rapid clinical entry can enable early falsification, but it may also bring issues of target selection, patient stratification, and molecular quality into higher-cost stages.
After this wave of consolidation, the three listed AI drug companies in China have formed three clearer paths to leadership.
OCUMENSION-B (01477.HK) has moved fastest in registration progress. MTS-004 has completed Phase III clinical trials, with its licensee advancing production verification and NDA submission preparation; MTS-201 has completed Phase I Part B; MTS-105 has entered an IIT dose-escalation study for hepatocellular carcinoma; and MTS-109 has obtained early human data and is preparing for IND filings in China and the US. Its core moat lies in organ-targeting LNPs, RNA sequence design, and in vivo expression, improving drug development efficiency by co-designing both the “payload” and “delivery method.” Moderna’s mRESVIA took about three years and four months from the first elderly subject dosing in January 2021 to FDA approval in May 2024. This timeline shows that after a mature mRNA platform completes process and regulatory validation, subsequent products have the opportunity to reuse development groundwork. If OCUMENSION-B can build advantages through organ-targeted delivery, orphan drug designation, and early human data, MTS-105 may have the potential for accelerated development, though actual registration timelines remain subject to clinical outcomes.
XTALPI has turned its small molecule approach into a scaled matrix: over 40 programs disclosed on its website with 13 receiving IND approval, and its core pipeline Rentosertib has initiated a Phase III IPF study. The research plan enrolls 320 subjects across 47 centers in China, with 52 weeks of continuous dosing, and the registration completion date is October 2029; recruitment, last-patient follow-up, and data cleaning account for the time gap between “one year of dosing” and “three years to complete.” Phase III typically requires hundreds to thousands of subjects and timelines measured in years. BIO statistics show the success rate from Phase III to NDA/BLA across disease areas is approximately 57.8%. IPF is particularly challenging due to patient heterogeneity, background medications, and long-term pulmonary function endpoints: ziritaxestat and pamrevlumab both showed positive signals in Phase II but were later terminated in Phase III due to benefit-risk concerns and primary endpoint failures. If Rentosertib passes this major test, it will become a critical validation for the entire AI drug discovery industry; recruitment efficiency at Chinese clinical centers may also help reduce time costs.
XTALPI presents a third type of leadership: its pipeline's ultimate form spans small molecules, antibodies, molecular glues, peptides, small interfering RNA, and cell therapy. Its 2026 interim report disclosed 3 clinical pipelines, more than 10 IND-approved or IND-preparing pipelines, nearly 10 PCC-stage pipelines, and over 20 discovery-stage projects; its 2027 target is over 10 clinical, over 10 IND-stage, and approximately 20 PCC pipelines — meaning more than 40 pipelines reaching PCC or more mature stages, forming the broadest drug modality pipeline globally. What better demonstrates the platform’s advancement is that these pipelines already sit at key value nodes across different modalities: in small molecules, SIGX1094, RTX-117, and PEP08 have entered clinical trials, and SIGX2649 has received US IND approval; in antibodies, three wholly-owned pipelines are planned to enter the clinic in 2027; among six siRNA pipelines, more than half have completed in vivo efficacy evaluation, with the fastest program reaching PCC; a molecular glue program obtained picomolar-level degraders within one quarter; brain-delivered peptides and oral cyclic peptides are advancing toward PCC; and incubated cell therapy programs have obtained multiple IND approvals in China and the US. Simultaneously advancing six major drug modalities to high maturity on the same technological foundation is almost unimaginable in traditional biotech history. Previously constrained by single-platform limitations, the market now favors platforms with “systemic optionality,” with more AI companies even favoring biologic modalities like antibodies and small nucleic acids — enabling drug modalities to actively adapt to biology and precisely match the most efficient weapon for specific diseases.
Following this “optionality” logic, examining each company's officially disclosed pipeline forms reveals clear divergence in industry division of labor and moats. The following table only counts officially disclosed ultimate drug modalities. Algorithm or service capabilities alone are not counted as formal pipelines; △ represents disclosed platform or partnership capabilities without clear named assets yet.
Comparing officially announced modality pipelines across AI drug companies as of publicly available information through September 2026. The table counts each company’s officially disclosed self-developed, partnered, or incubated pipelines, with stage information based on the company’s latest public statements. Table notes: ✓ indicates clear assets or pipelines; △ indicates disclosed platform or partnership capabilities without clear named assets; — indicates no officially announced pipeline. Sources: company announcements, official websites, and public clinical progress, compiled September 21, 2026.
The division of labor revealed by the table is clear: OCUMENSION-B leads in registration progress and delivery technology; XTALPI leads in small molecule pipeline breadth and late-stage validation; AbCellera, Absci, and Generate focus on proteins and antibodies; and XTALPI holds the broadest disclosed modality portfolio.
Multi-modality is becoming the common choice for tech giants entering AI drug discovery. Google DeepMind has expanded AlphaFold to proteins, DNA, RNA, ligands, and antibody complexes; NVIDIA BioNeMo focuses on protein binder design; OpenAI has partnered with Retro Biosciences to develop protein engineering models. When large companies enter drug development, they generally start with biologics and novel modalities because these areas have scarce data and more complex structural space, better demonstrating the value of closed-loop computation and experimentation. Different targets correspond to different optimal solutions: intracellular pockets suit small molecules, cell surface signaling can be blocked by antibodies, pathogenic proteins can be degraded by molecular glues, gene expression can be silenced by siRNA, and immune reset can employ TCE or CAR-T. Multi-modality platforms obtain the option of “target-first, modality-second” selection.
Historical data also offers reference points. In BIO statistics for Phase I to approval probability, small molecules are approximately 7.5%, monoclonal antibodies approximately 12.1%, RNAi approximately 13.5%, and CAR-T approximately 17.3%. These cross-period figures cannot directly predict individual pipelines, but they demonstrate that drug modality itself is a risk allocation tool.
XTALPI's breadth stems from horizontal reuse of the same foundation. Each of XTALPI's modalities achieves such broad coverage because they share a base built on quantum mechanics and physics-based modeling, AI generation and prediction, automated synthesis, and experimental validation, with standardized data flowing back into models. The R&D workflow and “wet-dry combination” AI drug discovery practical experience share a common foundation, with expansion speed driven by platform reuse rates. Its current small molecule clinical portfolio includes SIGX1094, RTX-117, and PEP08; SIGX2649 has received US IND approval and its China IND has been submitted. The self-developed TRK/RET gut-restricted small molecule XTN004 has completed US pre-IND material submission, with mid-report guidance for China-US filings in the second half of 2026; according to recent company communications, the program is scheduled to submit the IND this week, subject to official announcements. In biologics and novel modalities, multiple programs are advancing simultaneously. XTALPI's subsidiary Ailux's ALX001, ALX002, and ALX005 are planned to enter clinical trials in 2027; Kodexia has established 6 siRNA pipelines, with more than half completing in vivo efficacy evaluation, including an IgA nephropathy program that obtained non-human primate data in approximately 7 months; a molecular glue program has achieved picomolar-level degraders within a quarter; the peptide platform's brain-delivery and oral cyclic peptide programs are advancing toward PCC; and the incubated company Leman Biotech's META 10-19 has obtained multiple IND approvals in China and the US.
SIGX1094 provides a sample for observing platform conversion efficiency. Targeting diffuse gastric cancer, it has shown preliminary safety and anti-tumor signals in Phase I and received FDA orphan drug and fast track designations; the Phase II/III application for combination with Innovent's KRAS-G12C inhibitor has been accepted by CDE. Fast track can improve regulatory communication and rolling review efficiency, and the subsequent registration path depends on Phase II design and efficacy strength. This pipeline may qualify for a rare disease Phase III exemption, potentially initiating market launch as early as 2027 after completing Phase II. SIGX Health oversees clinical advancement while XTALPI retains future commercialization revenue sharing, allowing platform value to be realized as assets mature.
In the first half of 2026, XTALPI's AI4S business revenue grew 136.4% year-over-year to RMB 193.5 million. The growth in pipeline count and multiple modalities crossing PCC and IND nodes simultaneously indicate that foundation reuse is beginning to convert into asset density.
Ailux and DoveTree: The platform begins actively capturing pipeline value. XTALPI's subsidiary Ailux can be seen as a vanguard for its business model evolution. The company first licensed its structure prediction platform to Johnson & Johnson and UCB, then reached a bispecific antibody partnership with Eli Lilly worth up to USD 345 million, including platform licensing options, co-development, and milestone revenue; it is now further concentrating resources into three wholly-owned autoimmune pipelines. This evolution has rapidly completed the value jump path of “platform licensing — co-development and revenue sharing — proprietary pipelines.” Ailux's AtlaX proprietary data foundation includes billions of protein sequences, approximately 140,000 antigen-antibody sequence pairs, and approximately 30 million synthetic structure data points, combined with a 30,000-square-foot wet lab forming a solid feedback loop. While Anthropic is still building biological labs through acquisitions to fill experimental gaps, Ailux has already accumulated real-world data directly serving model training. Maria Belvisi, who joined Ailux this April, previously served as Senior Vice President of Respiratory & Immunology R&D at AstraZeneca, managing approximately 500 scientists and advancing tozorakimab from preclinical to Phase III. Her arrival points to a clear goal: upgrading the AI antibody platform into a Biotech with global clinical development capabilities. As a private market valuation reference, Anew Labs recently raised USD 290 million at approximately USD 1.5 billion post-money valuation, with its fastest publicly disclosed pipeline still in preclinical stages. In contrast, Ailux already has multinational pharma clients, proprietary data, a wet lab, and three clear proprietary pipelines, with significantly higher asset maturity. The DoveTree partnership shows XTALPI began introducing external leverage for clinical and commercialization capabilities earlier. XTALPI has received USD 51 million in upfront payments and a USD 19 million second payment, totaling USD 70 million; up to USD 30 million remains in near-term payments under the original agreement, with total potential value up to USD 5.99 billion; the first oncology asset has entered IND-enabling stages. DoveTree founder Gregory Verdine has co-founded over a dozen biotech companies, with more than five having reached public markets; Fog Pharma, renamed Parabilis, went public this June. At signing, XTALPI disclosed that Verdine had co-developed 3 FDA-approved drugs; this year's approved pancreatic cancer near-blockbuster daraxonrasib, whose RAS(ON) tricomplex technology engine absorbed early work from Verdine and Warp Drive, brings the approved drug record of this technology lineage to 4. In this complementarity, XTALPI needs Verdine's seasoned judgment on targets and commercialization paths, while Verdine values XTALPI's hardcore capability of integrating AI, physical intelligence, and automated experimentation into a single platform. This synergy is converting pipeline counts on paper into high-value assets in real money.
The key to valuation: converting technological breadth into continuous value nodes. In speed, OCUMENSION-B is closest to product registration; XTALPI has the AI-native small molecule Phase III benchmark; in modality and business model breadth, XTALPI occupies a rare position among listed AI drug companies globally. This breadth can bring investors three layers of tangible defensive and offensive value: 1. Decentralized pipeline risk: more independent clinical and BD events reduce the dominance of any single pipeline's success or failure over the company's overall valuation. 2. Biology-driven “modality options”: comparing different modalities around the same target lets drug form truly serve biological problems rather than being constrained by technical limitations. 3. Multi-tier revenue structure: platform service income, milestones, sales royalties, proprietary pipelines, and incubated ecosystem equity create a substantial profit pool.
Traditional Biotech valuations are often held hostage by one or two core assets; XTALPI more closely resembles a composite valuation model of “R&D infrastructure revenue + risk-adjusted value of proprietary pipelines + partnered pipeline royalties + options on Ailux and incubation ecosystem + new materials platform.” The market has already assigned initial pricing to its service revenue, but the pricing of its multi-modality pipeline and substantial ecosystem interests remains in very early stages. The upcoming value realization nodes are clear: XTN004 formally submitting its IND, SIGX1094 entering its next clinical phase, three ALX antibody programs entering the clinic in 2027, and small nucleic acid and peptide programs continuously reaching PCC. Policy tailwinds have lifted industry risk appetite, but the scarcity of foundational technology determines how far the premium can extend. If secondary market investors are betting that AI drug discovery is upgrading from “single-point black-box tools” to “industrialized systems capable of reproducibly producing blockbuster drugs,” then XTALPI, with the broadest portfolio, undoubtedly holds the most complete value mapping in the current market.
Comments