Taking on multinational giants, gross margin tops 75% — “first stock of medical‑imaging large model” eyes Hong Kong listing

Fast rise built on licensing, not hardware

Hangzhou Deshi Biotechnology Co., Ltd. (德适生物) has passed a Hong Kong Stock Exchange hearing and is preparing to list, reportedly positioning itself as the market’s first public “medical‑imaging large model” company. The IPO filing shows revenue in the first nine months to 30 September 2025 of RMB 111.6 million, up 469.8% from RMB 19.6 million a year earlier, while overall gross margin jumped to 75.9% from 42.9% a year prior. How did a company whose flagship product is still awaiting a top‑tier regulatory certificate achieve that kind of leap? The short answer: it pivoted from selling boxes to selling model licences and MaaS.

Deshi’s prospectus breaks revenue into three parts: imaging software and devices, technology licensing, and consulting services. Technology licences — access to its iMedImage® base model via the iMed MaaS® platform — surged from RMB 0.475 million in the first nine months of 2024 to RMB 57.37 million in the same period of 2025, accounting for 51.4% of revenue. Those licence sales report gross margins above 96%, because the firm has expensed large parts of model development as R&D rather than capitalizing training and infrastructure, leaving near‑zero marginal delivery cost once the model is packaged as a service. Is that sustainable? Critics will point to the accounting effect: high upfront R&D spend today can yield very high reported margins tomorrow.

Clinical promise, big regulatory caveat

Deshi’s core clinical product, AI AutoVision®, targets chromosome karyotype analysis — a labor‑intensive, high‑volume niche in reproductive and hematological diagnostics. Its trial of 1,518 subjects claims large efficiency gains (analysis time cut from ~34.1 to ~11.3 minutes per sample; report turnaround from ~30 days to 4–7 days) and strong accuracy metrics. The company has also packaged a hardware‑software stack (scanners, automated sample prep, and analysis) and, by 2024 sales, claims a 30.6% market share in China’s karyotyping market — narrowly ahead of long‑dominant foreign incumbents such as Zeiss and Leica. If AI AutoVision® secures Class III registration, it could become the first certified third‑class AI software to automatically flag chromosomal abnormalities and alert clinicians, a regulatory milestone that would materially lift pricing and market access, it has been reported.

But the product is not yet approved. Deshi says the device was designated a Class III innovative medical device and moved through accelerated review; it has been reported that the company submitted all supplementary materials in January 2026 and expects approval in Q1 2026. That timeline is critical: until the certificate is granted, the high‑ticket AI AutoVision® will remain constrained by regulatory and procurement processes.

Growth runway — and clear risks

Deshi has established distribution across 31 provinces through 75 distributors and more than 400 hospitals, with penetration in 40% of China’s top‑10 hospitals, providing a ready commercial channel. But two risks loom. First, R&D spending spiked to RMB 68.67 million in the first nine months of 2025, underscoring the cash intensity of training and productising large medical models. Second, geopolitical and trade dynamics matter: constraints on advanced AI chips, sanctions or export controls could complicate overseas scaling or hardware procurement, especially for compute‑heavy medical AI and integrated devices. Investors should ask: are the margins structural or a function of accounting and a short‑term licensing boom? Deshi’s claim to be a pioneer in medical‑imaging large models is bold. The answer to whether it becomes a durable challenger to century‑old multinational incumbents will depend on regulatory approval, margin sustainability and how it navigates the geopolitics of AI and medical devices.