How to choose AI applications in Hong Kong stocks? DIAGENS-B (02526): A rare vertical AI target with high growth and strong barriers.

Since September, more than a dozen global flagship large models, including GPT-6 Astra and Claude Fable 5.1, have been intensively released, marking a renewed focus on the AI sector within the market. Regarding the allocation logic for AI applications in Hong Kong stocks, Huatai clearly stated its view on September 4: that until the narrative of new demand is fully confirmed, the chip structure of the AI application track is significantly superior to that of hardware and holds a relative short-term advantage; moreover, the beta attributes of Hong Kong software stocks are stronger than those of A-shares, with the core focus of stock selection consistently being the acceleration of revenue growth quarter-on-quarter, where performance reigns supreme. Reviewing the evolution path of the current AI market, it appears that the market is shifting from speculating on model parameters to betting on the implementation of applications, leading to a noticeable differentiation among AI application targets in Hong Kong stocks. Although the general large model track has great imagination potential, it faces fierce competition and a pricing war, resulting in long and uncertain performance verification cycles; conversely, the vertical industry AI track, characterized by stringent regulation and high barriers, is becoming a more cost-effective option for investment in AI applications due to its clear commercialization path, stable competitive landscape, and certain performance delivery pace. In the vertical AI track, medical imaging AI stands out as a high-quality sub-sector, combining short-term performance with long-term space. Unlike consumer-grade AI applications, medical AI faces three core barriers: data compliance, clinical validation, and medical device regulation, making it difficult for new entrants to rapidly break through solely through computational power investments. The competitive landscape in this sector is much more favorable than in the general track. Additionally, with improved medical insurance payment mechanisms and the release of demand from hospitals, medical AI is transitioning from pilot demonstrations to scaled implementation, entering a performance realization phase. DIAGENS-B (02526) is a rare stock in the Hong Kong medical AI track that perfectly aligns with the selection criteria of "high growth + strong barriers + clear catalysts." Performance: Precisely hitting the core indicator of "quarter-on-quarter revenue growth acceleration" The core of stock selection is performance, and the growth quality of DIAGENS is in the first tier within the Hong Kong AI application sector. The mid-2026 report shows that the company achieved revenue of 108.7 million yuan in the first half of the year, representing a year-on-year increase of 21.0%; among these, model service revenue reached 94.541 million yuan, significantly increasing by 101.1% year-on-year, accounting for 86.9% of total revenue. Compared to the 51.4% share of technology licensing revenue in the whole year of 2025, the revenue growth rate and share of the company's model business have both accelerated quarter-on-quarter, with the revenue structure fully completing the transition from "equipment sales" to "model services." More importantly, this portion of revenue is not a one-time project income but rather encompasses service revenue from model training, system integration, process adaptation, and ongoing operations, which enhances customer stickiness and income predictability. While most Hong Kong AI application targets remain in the "concept validation, minimal income" stage, DIAGENS has validated the feasibility of its business model with nearly 90% share of revenue, leading the sector in performance realization. This aligns perfectly with the core stock selection framework of "quarter-on-quarter revenue growth acceleration." According to an initial coverage report by KGY Securities, the companys operating revenue is forecasted to reach 490 million yuan, 930 million yuan, and 1.72 billion yuan in 2026, 2027, and 2028, respectively, maintaining nearly double-digit growth for three consecutive years, and given a "buy" investment rating; the current stock price corresponds to predicted revenue PS (price-to-sales ratio) valuations of 82.6 times, 43.7 times, and 23.6 times for 2026-2028. In horizontal valuation comparisons, popular general large model companies like Z.AI and MiniMax have dynamic PS ratios exceeding one hundred, while vertical specialized track large model targets possess higher industry barriers and commercialization certainty, thus offering a valuation premium compared to general large models. If calculated against a median industry PS of 100 times, the corresponding company's 2026 projected revenue could reach a reasonable market value of nearly 60 billion HKD, leaving ample room compared to the current valuation; if the expected revenue for 2027 is achieved, the market value will likely exceed 100 billion. Barriers: Deep moats in vertical tracks, avoiding the hyper-competition of general models Unlike the internal competition logic of large general models focusing on parameters and prices, DIAGENS' core barriers are built upon the depth of medical expertise and regulatory compliance, which are thresholds that general model manufacturers find challenging to quickly navigate through computational power alone. The companys self-developed iMedImage medical imaging foundation model is specifically designed for training across all modalities of medical data, including 2D images, 3D CT/MRI, and pathological slices, rather than being a fine-tuned version of a general model; it holds generational advantages in key metrics such as preservation of lesion details and clinical evidence linkage. Based on this, the iMedLoop full-process platform seamlessly integrates data annotation, model training, evaluation release, and clinical feedback, reducing the data requirements for single disease model development to one-two-hundredth of traditional demands and compressing the cycle to 2-3 months, thus achieving industrial-scale production of specialist models. More critically, on the regulatory barrier aspect, the companys AI AutoVision secured the worlds first NMPA Class III medical device registration certificate based on large model technology. This certificate is not only an approval for a single product but also validates the replicability of the foundation model specialized model compliant device pathway, paving the regulatory road for the commercialization of more models in the future, forming a long-term competitive moat. Catalysts: Dual events to unlock value reassessment windows Recently, the company has welcomed two significant catalytic events that jointly promote value restoration. First, its inclusion in the Stock Connect program on September 7 is now officially in effect. After the Stock Connect opens, the influx of incremental funds from the mainland will not only enhance trading activity but will also drive the valuation system to transition from niche Hong Kong biotechnology stocks to vertical AI platform companies, rectifying the previous liquidity discount. Second, the unveiling of the Joint Laboratory for General Artificial Intelligence and Medical Applications in collaboration with The Hong Kong Polytechnic University. As global large models upgrade towards intelligent agents, the company is proactively aligning with the trend of next-generation technologies: leveraging world-class cutting-edge research capabilities and integrating its industrial implementation capabilities, it is exploring a new paradigm where intelligent agents participate in the full process of medical research and development, potentially upgrading its technological moat from model mass production capability to research paradigm advantage, thus opening up a long-term growth ceiling. Returning to the stock selection logic for AI applications in Hong Kong, as the market has reached its current state, the phase of pure concept speculation is over, with funds increasingly concentrating on stocks that demonstrate performance, barriers, and catalysts. DIAGENS not only has performance validation with its model service revenue doubling but also possesses deep barriers in the medical sector, along with the dual catalysts of Stock Connect and technological upgrades, making it a rare certainty configuration target in the current Hong Kong AI application sector.