The "NVIDIA Moment" in Medical Imaging: Will Dazhi Become the Next AI Ten-Bagger?

In 2026, AI industrialization is shifting from general-purpose models to deep specialization in vertical industries.

OpenAI has launched ChatGPT Health, Google has released its medical large model MedGemma 1.5, and domestic tech companies are rolling out plans in AI for healthcare. This suggests an industry consensus is forming: AI that can create value doesn’t necessarily have to be “bigger,” but it definitely needs to be “more specialized.”

Against this backdrop, as “the first listed company for medical imaging large models,” Desim is set to list on the Hong Kong stock exchange on March 30, and the significance of this IPO has been underestimated.

Desim began its share offering on March 20 and closed applications on March 25. Judging from broker trading software, as of now the company’s margin subscription multiple has already exceeded 140x. Among multiple new listings in the same period, its performance has been particularly hot.

During the period when the Hong Kong IPO market sees a concentrated surge of offerings on the application side, Desim can attract a large amount of new-issue subscription capital. Behind this is investors’ early bet on the AI large-model industrialization narrative.

Desim’s self-developed trillion-parameter foundation model iMedImage®, covering medical imaging scenarios where intelligent demand is especially urgent, combined with a commercially validated closed loop, enables the company to tell a story about a healthcare AI infrastructure-level platform.

“Super brain” for medical imaging—an anchor for AI applications the market is eagerly searching for

At the start of 2026, the listings of the two “big model” champions, Zhipu and MiniMax, ignited market enthusiasm. But after the frenzy, capital has started thinking about new questions: where exactly has AI commercialization progress gotten to? Those large models trained with huge amounts of money—where can they truly start making profits?

At present, it seems that general-purpose models—whether enterprise API calls or consumer subscription memberships—are still in the exploration stage.

So, domestic and overseas large-model companies have all turned their attention to medical large models.

The unique thing about the medical track is that it simultaneously has three traits: a strong “must-have” demand, monetization capability, and policy-driven momentum—so the commercialization outlook is relatively clear. Especially in the medical imaging sub-segment, it is urgent to use AI to solve long-standing supply-side contradictions.

From the perspective of supply and demand: medical imaging data accounts for more than 70% of all medical diagnostic data, forming the cornerstone of modern medicine. Data from Frost & Sullivan shows that China’s medical imaging detection market is expected to reach a scale of RMB 159 billion by 2030. But on the other side of strong demand is severe shortage on the supply side. Training cycles for imaging physicians are long, and growth in talent supply is limited—especially in primary hospitals, where this scarcity is even more pronounced.

So, the policy side is accelerating the advancement of “Artificial Intelligence + Healthcare.” It is moving from encouraging exploration toward system building, enabling the supply side with AI. In October 2025, five departments including the National Health Commission jointly issued the “Implementation Opinions on Promoting and Regulating the Development of ‘Artificial Intelligence + Medical and Healthcare’ Applications,” clarifying that by 2030, the goal is to promote nationwide, in general, AI-assisted medical imaging services for clinical diagnosis in hospitals at the secondary level and above. The newly released outline of the “Fifteenth Five-Year Plan” also proposes that the use of digital intelligence technologies in scenarios such as assisted diagnosis should be promoted in an orderly manner.

A clear industrial signal is being released: the golden age of AI diagnosis has arrived.

In the new AI investment narrative, Desim’s scarcity lies in the fact that it has entered the highly technical barrier field of karyotype analysis and has become the “number-one player” in a sub-segment that establishes dominance.

In the microscopic imaging market, karyotype analysis is the gold standard for cytogenetic diagnosis. According to Frost & Sullivan, based on 2024 sales revenue, Desim holds a 30.6% share in China’s karyotype analysis market, ranking first in the industry. It has successfully broken the long-standing market dominance by overseas players such as ZEISS and Leica.

What truly supports this market position is Desim’s self-developed iMedImage® medical imaging foundation model.

This product, which Frost & Sullivan has recognized as a “general-purpose medical imaging foundation model with the largest global parameter scale,” supports 19 medical imaging modalities and covers more than 90% of clinical scenarios—like a “super brain” with deep medical imaging know-how.

This means Desim is building a technology foundation with platform characteristics. In the company’s plans, the iMedImage® foundation model will enhance computing power to process large-scale medical imaging data, strengthen cloud service capabilities, and expand to more medical imaging modalities, broadening applications across a wider range of clinical scenarios.

From a development strategy standpoint, Desim has landed at the intersection of two narratives that are simultaneously exploding: the “Artificial Intelligence + Healthcare” policy narrative and the large-model industrialization narrative. And the company’s IPO prospectus also answers the market’s core focus question: how medical AI can run through commercialization.

Seizing the “AI healthcare infrastructure” position—growth driven by a business closed loop

If you understand Desim only as a medical AI application company, you would underestimate the company’s core value: creating a new paradigm, solving the tough problems of traditional AI commercialization, and accelerating the process of intelligence across the healthcare industry.

The core issue with the commercialization of traditional medical AI is that it cannot monetize at scale. Hospitals often need customized development—each application is a separate redevelopment, with a long cycle, high cost, and difficulty to replicate. For example, developing a lung nodule detection model may take years of R&D, and it can only be used on CT imaging. If you switch to ultrasound or pathology slide samples, you need to start from scratch again.

This “project-based” model traps medical AI vendors into the outsourcing-of-labor pit. It is hard to achieve large-scale replication, let alone platform-based expansion.

Against this backdrop, the value of Desim’s self-developed foundation model lies in its generalization capability. When facing new tasks, iMedImage® only needs a few hundred samples of imaging data and a training cycle of several days to complete high-precision model transfer training.

This means Desim can transform medical AI from project-based development into platform-based supply—giving the medical industry, for the first time, a capacity akin to cloud computing scalability. This capability gives Desim the narrative space to position itself as “AI healthcare infrastructure.”

The narrative is supported by Desim’s dual-engine driven business model.

On one hand, its AI intelligent equipment and systems business serve as the anchor, providing stable cash flow and customer stickiness through a closed loop of “medical imaging software + medical equipment + reagent consumables + technology licensing.”

Enabled by the foundation large model, Desim has developed medical imaging software products AutoVision® and three medical devices, as well as four major reagents and consumables. All have been commercialized, entering more than 400 medical institutions. In 2023 and 2024, it recorded revenue of RMB 52.844 million and RMB 70.352 million respectively, representing year-over-year growth of 33.1%. Among them, the core product AI AutoVision® submitted a registration application for a Class III medical device to the National Medical Products Administration (NMPA) in May 2025, and in the same month was recognized by the NMPA as a “Class III innovative medical device.”

This anchor business proves that Desim’s technology can be implemented, that real demand exists, and that the business model is sustainable. In addition, a sub-segment market share of more than 30% is a critical threshold: it means the company’s narrative is transitioning from domestic substitution to domestic leadership, and it will benefit from the expansion dividend of this golden track.

On the other hand, the MaaS (Model as a Service) model opens a second growth curve. Desim’s iMed MaaS platform cloudifies model capabilities. Its business model extends from selling services to selling computing power and model licensing.

Hospitals do not need to purchase expensive hardware equipment; they can obtain top-tier AI diagnostic capabilities via cloud-based “use it right away” access. In 2024, new model service business quickly scaled up, driving an improvement in the company’s overall gross margin.

In the first three quarters of 2025, Desim Bio achieved revenue of RMB 112 million, up 470% year over year. Overall gross margin increased to 75.9%. With continued increases in R&D investment, the company’s losses narrowed significantly compared with the same period last year. In the AI healthcare sector—still widely in a “burning money validation period”—this looks especially scarce.

Looking ahead, as hospitals connect, data continues to accumulate, and models are continuously optimized, more applications and customers will be attracted to connect, forming a positive-feedback loop.

This is the “data and business dual flywheel” unique to the AI era. And considering the digital transformation of the primary healthcare market, Desim’s growth flywheel can spin even faster.

Under China’s tiered healthcare system, many primary-level hospitals lack imaging physicians of high caliber. However, cloud-based AI capabilities that are low-cost and high-efficiency can quickly fill this gap. Combined with AI AutoVision® improving doctors’ analysis efficiency by more than three times in clinical trials, shortening report turnaround time from 30 days to 4 to 7 days of data, Desim’s commercial closed loop has substantial advantages in bridging regional differences in medical resource availability.

AI can help make high-quality medical resources more accessible. This means a trillion-scale primary-care market is opening the door to Desim.

A “new paradigm” asset scarce on the Hong Kong market—anticipating an NVIDIA-like moment for medical AI

As “the first listed company for medical imaging large models,” Desim stands on the dividing line of the era.

In recent years, two types of medical companies have risen. One type is medical platform companies like intuitive surgical companies (the da Vinci surgical robot), with a “hardware entry + software soul + consumables flywheel.” In a few years, such companies have crossed into becoming billion-dollar-scale leaders. The other type is new forces driven by “data + AI-powered decision-making,” such as Tempus and Viz.ai. These companies have shown that medical AI itself can generate scalable revenues, enabling high valuation premiums typical of emerging technology.

Desim sits between the two: it has the anchor business of AI intelligent equipment and systems, as well as a second growth curve represented by the iMed MaaS platform. This composite model of “soft-and-hard integrated with platform output” has not appeared in Hong Kong’s AI healthcare sector before.

But if you zoom out a bit more, you’ll find Desim’s listing timing has perfectly matched the evolution laws of the AI industry: from breakthroughs in infrastructure to breakthroughs in applications.

In the United States, the general domain completed the underlying construction of the “compute layer (NVIDIA) – cloud infrastructure (Amazon) – model layer (OpenAI),” and only then did application companies in vertical domains see explosive growth. Platform-type companies like Intuitive Surgical crossed into trillion-dollar valuations, while many emerging AI application companies have become the focus of capital pursuit in the past couple of years.

China’s AI industry is going through a similar process.

In 2025, domestic AI chipmakers were the first to sound the re-valuation call. Earlier this year, general-purpose large model companies such as Zhipu and MiniMax took the baton and surged.

Now, it’s Desim—a medical platform company that simultaneously has both a hardware entry and the model core—that is stepping onto the stage. For the first time in Hong Kong, a medical company is being priced “based on AI infrastructure.” This is a signal worth paying attention to.

Desim’s listing at this moment objectively reveals that the Hong Kong market is following the industrial trajectory, and that it is brewing a Chinese “NVIDIA-like moment” in China’s AI healthcare sector.

Looking forward, the second half of “Artificial Intelligence + Healthcare” is beginning.

A wave of innovation from infrastructure to application is about to arrive. As the “first stock,” Desim is expected to become a key focus for capital allocation, often benefiting from liquidity premium and attention-tailwind gains.

During the long process of industrial evolution, companies that can continuously build ecosystem barriers and achieve a commercial closed loop will have the opportunity to cross from the tens of billions to the hundreds of billions—and even from billions to trillions. This is a combined pricing of scarcity and certainty. The market always favors the definers of underlying rule sets.

Source: Hong Kong Stock Research Society