Hangzhou Diagens Biotechnology Diagens Tech (02526.HK): Model‑as‑a‑Service Revenue Surges 101.1% Year‑over‑Year; Commercial Rollout of Its Medical Imaging AI Platform Gathers Momentum

On August 7, Diagens Technology Co., Ltd. (02526.HK, Diagens‑B, “Diagens Tech”) released its interim results for the six‑month period ended June 30, 2026. During the reporting period, Diagens Tech (the “Company”) posted revenue of RMB 108.7 million, up 21.0% year‑over‑year; gross profit reached RMB 80.504 million, representing a 14.0% year‑over‑year increase, with an overall gross margin of 74.1%. Among revenue streams, model‑as‑a‑service revenue hit RMB 94.541 million, jumping 101.1% year‑over‑year and accounting for roughly 86.9% of total revenue, emerging as the key driver of topline growth for the period. Scaling revenue from model‑as‑a‑service underscores the steady commercial momentum of the Company’s platform-based business model built around its foundational large models for medical imaging.

As a tech company revolutionizing medical diagnostics with AI, Diagens Tech has built a world‑leading “R&D and production acceleration platform” for medical imaging AI anchored on its self‑developed foundational medical imaging model iMedImage®. It combines massive hospital imaging datasets with clinician expertise to rapidly incubate AI‑powered diagnosis assistance models. Its core competitive edge lies in the reusability of its foundational model: one underlying framework works across imaging modalities, organs, diseases and clinical use cases, enabling efficient iteration of specialty‑specific models without building solutions from scratch. Launched after the reporting period, iMedLoop™ links critical workflows including data acquisition, professional annotation, model training and evaluation, release and deployment, and real‑world application feedback, creating a sustainable loop: foundational model – specialty‑specific model – services and products – real‑world feedback – model iteration.

From a business model perspective, the Company operates two core business lines: model‑as‑a‑service and intelligent medical imaging products. The former empowers clients to translate data and expertise into functional AI models. The latter turns proven models into regulatory‑compliant medical software and devices integrated into routine clinical workflows, delivering a full value chain from technology enablement to real‑world implementation. Diagens Tech is accelerating its transition from a development stage focused on isolated technical capabilities and project‑level validation toward a new commercial phase defined by reusability, scalability and continuous iteration.

During the reporting period, Diagens Tech further shifted its business focus toward model services. Revenue from model‑as‑a‑service rose from RMB 46.96 million in the same period last year to RMB 94.541 million. Its offerings cover model and technology licensing, cloud‑hosted iMedMaaS® services, and on‑premises SCTI all‑in‑one appliances for storage, computing, training and inference, catering to diverse hospital requirements for cloud access, data security and on-premises deployment. The Company redefined this revenue stream from “technology licensing” to “model‑as‑a‑service” in the current period. There have been no material changes to underlying business activities or revenue recognition methodologies. The updated label better reflects the current service portfolio and business development.

Aligned with the full end‑to‑end workflow of medical imaging AI – spanning data processing, model R&D and real‑world adoption – Diagens Tech keeps strengthening its medical imaging AI R&D and production acceleration platform. The iMedImage® foundational medical imaging model delivers reusable image comprehension and reasoning capacity. iMedStudio™ handles data processing, professional annotation, manual revision and QC review. iMedMaaS® enables training, publishing and deployment of specialty‑specific models, while DoctorBench® assesses model performance, safety and operational boundaries. Collectively, these modules form the core technical backbone of the R&D and production platform. After the reporting period, these components have been orchestrated by iMedLoop™ into one cohesive end‑to‑end system. As of the announcement release date, more than 3,000 professionals have participated in iMedLoop™, with roughly 28.95 million annotated samples accumulated.

As of June 30, 2026, the Company had completed 158 model-related projects in partnership with 99 hospitals, including 65 Grade‑3 Class‑A hospitals, covering 43 human organs/anatomical sites and 61 disease categories. These collaborative projects continuously build up real‑world assets: specialty task definitions, evaluation methodologies, deployment know‑how and clinical practitioner feedback. This practical foundation supports ongoing model iteration and future project delivery, while broadening the scope of medical imaging tasks addressable via the platform.

Within the medical imaging AI sector, commercialization strategies are diverging sharply. One group builds businesses around the sales of imaging hardware, rolling out AI as a value‑adding feature bundled with physical devices. Another group focuses on disease‑specific vertical models, pursuing commercialization through disease‑by‑disease R&D, regulatory registration and product sales. Adding new disease indications typically requires fresh data preparation, model development and regulatory validation. Diagens Tech has adopted a platform‑based foundational model strategy. The shared technical foundation of iMedImage® enables multi‑task reuse. iMedLoop™ links data governance, model training, evaluation, deployment, feedback and iteration. This strategy steadily brings down development and delivery costs for additional specialty‑specific clinical tasks.

As more Class III medical device products gain regulatory approval, standalone algorithms are no longer the primary scarce asset. Competitive moats are shifting away from the accuracy of isolated models toward full‑stack platform capabilities – compliant data use, industrialized model production, regulatory‑compliant commercialization and real‑world clinical delivery. Looking at industry evolution, companies with closed‑loop capabilities spanning data governance, model R&D and clinical deployment are best positioned to capture opportunities arising from medical imaging AI industrialization and market‑driven deployment of healthcare data assets. Accounting for roughly 86.9% of total first‑half revenue, Diagens Tech’s model‑as‑a‑service business demonstrates that the Company has built a proven end‑to‑end “data‑model‑clinic” commercial value chain, with its platform‑centric capabilities now undergoing scalability validation.

On the regulatory front, on May 19, 2026, Diagens Tech’s AI AutoVision® karyotyping image‑aided diagnosis software secured the Class III medical device registration certificate issued by the National Medical Products Administration (NMPA). This product assists segmentation, counting, identification, rearrangement and suspected‑anomaly flagging for G‑banded karyotype images from peripheral blood and amniotic fluid samples, with all outputs subject to review by qualified professionals. This regulatory clearance further validates Diagens Tech’s capacity to translate foundational medical imaging model capabilities into regulatory‑grade medical device offerings, moving its flagship product past regulatory review and into the commercial launch phase.

To strengthen its long‑term technology and product foundations, Diagens Tech recorded R&D expenses of RMB 64.118 million in the first half, representing a 67.4% year‑over‑year increase. Investment priorities included upgrades to its foundational medical imaging model, specialized workflows and high quality data governance, model performance evaluation, full lifecycle management for core products, and R&D for pipeline products. At period‑end, the Company held roughly RMB 655 million in cash and cash equivalents, net current assets of approximately RMB 701 million, and a debt‑to‑asset ratio of around 13.0%, providing solid backing for future R&D spending and commercialization initiatives.

During the results briefing, SONG Ning, Founder and Chairman of the Board of Diagens Tech, outlined three evolutionary phases for global medical imaging AI. The 1st phase is broad adoption of AI‑assisted diagnostics. Approximately 14,000 existing medical imaging diagnostic workflows across 3,000 categories are gradually shifting expert‑only interpretation toward AI-assisted workflows, lifting diagnostic efficiency and accuracy while extending high quality diagnostic capacity to grassroots medical institutions. The 2nd phase has already commenced and is expected to accelerate over the next two years. AI will enable entirely new imaging‑based clinical interventions. Use cases include ultrasound‑driven prediction of fetal preterm birth risk, AI-aided delivery mode assessment, and early directional prognosis judgment for tumor drug treatment derived from imaging analysis. The total number of relevant clinical workflows is projected to exceed 16,000. The 3rd phase is set to mature over the next four to six years, marked by deep integration between large imaging models and large language models. This will take public health management systems to the next level, delivering more efficient, precise and inclusive health services for all.

On capital market developments: per Hang Seng Indexes Company’s quarterly review schedule and market forecasts, Diagens Tech is viewed as a potential candidate for inclusion in the Hang Seng Composite Index in this round of adjustments. Review outcomes are expected to be announced in late August. Should the Company secure inclusion, it may qualify for Stock Connect trading upon the September index re‑constitution. Successful index inclusion would likely draw broader mainstream investor attention and improved liquidity for the Company’s platform‑driven commercialization strategy.

Diagens Tech stated it will keep driving technical iterations centered on iMedImage®, enhance iMedLoop™ capabilities in data generation, model training, evaluation, deployment services and real‑world feedback. It will push ahead with productization and large‑scale delivery of its model‑as‑a‑service offerings. It will further accelerate commercial rollout, overseas registration and indication expansion for AI AutoVision®, alongside progress across its in‑development product pipeline. Leveraging its platform strengths, Diagens Tech will capitalize on opportunities arising from medical imaging AI industrialization and market‑driven deployment of healthcare data assets, fueling the global medical imaging sector’s shift to intelligent clinical workflows.