By 2026, Hong Kong will witness a significant surge in the number of listed AI companies.

Data from the Hong Kong Stock Exchange reveals that from early December 2025 to the end of May 2026, IPO fundraising for enterprises within the AI value chain reached HKD 97.9 billion, representing roughly 55% of the total new stock fundraising on the exchange during that period.

From large-scale models and computing power chips to enterprise software and medical technology, a growing number of AI companies are stepping into the limelight on the Hong Kong Stock Exchange.

For a market to undergo revaluation, the introduction of new, high-quality assets is a prerequisite.

Historically, discussions about technology on the Hong Kong Stock Exchange often centered around a handful of internet platforms. Today, however, models, chips, enterprise software, and medical AI are all emerging simultaneously, offering a more diverse range of samples for comparison. The concentrated emergence of AI companies at least signals a transformation in the corporate landscape of the Hong Kong Stock Exchange.

Meanwhile, new questions have surfaced. When there were fewer AI companies, it was relatively straightforward for the market to debate “whether they were truly AI” and categorize them based on their rarity.

However, as the number of AI companies grows, the focus has shifted to—how should these companies, all branded as “AI,” be compared and evaluated?

Recently, VB Data unveiled a list of the “Top 10 Cost-Effective AI + Large Models on the Hong Kong Stock Exchange.” This ranking selects enterprises from the main board of the Hong Kong Stock Exchange that directly offer AI models, platforms, industry applications, and AI computing products. It then assigns a comprehensive score based on factors such as current price increases, peak year-to-date increases, and discounts from peak prices.

Ultimately, Zhipu AI claimed the top spot, followed by Data Search AI-B in second place, with Iluvatar CoreX, Xunce, and SmartAds rounding out the top five.

To grasp the significance of this ranking, it's crucial to differentiate between two aspects: the score reflects price performance, but what the market truly values is the company's ability to translate future capabilities into revenue.

Specifically, Zhipu AI, which tops the list, represents a type of asset that is familiar to the market. Its foundational model capabilities, token consumption, API calls, developer ecosystem, and enterprise clients all provide a ready framework for analysis.

The stronger the model capabilities and the larger the scale of use, the easier it is to comprehend service revenue and ecosystem expansion. It's no surprise that such companies rank at the forefront.

The real intrigue, however, lies with Data Search AI: in a market where chips and computing power remain the most recognizable AI narratives, how did a medical imaging company rise to second place?

The answer doesn't lie in the so-called scarcity label of “medical AI.”

To understand Data Search AI, one must look beyond the lens of a single medical software. After medical imaging evolves from a single-point tool to a foundational model, the underlying business logic undergoes a fundamental shift.

In the past, medical imaging AI was primarily developed around individual diseases or tasks. Switching to a different department or clinical issue often necessitated recollecting and labeling data and retraining the model.

This approach could solve specific problems but had a strong project-based nature. An increase in the number of products didn't necessarily lead to a corresponding decrease in research and development and delivery costs.

What the foundational model changes is the approach to reuse. According to the company's prospectus, Data Search AI's iMedImage is a general-purpose medical imaging foundational model that encompasses nearly a hundred clinical imaging modalities, including chromosomes, ultrasounds, CT scans, and MRIs, and supports tasks such as classification, detection, and segmentation.

It's important to note that “general-purpose” doesn't imply handling all diagnostics but rather enabling the same set of underlying capabilities to be transferred across imaging modalities, diseases, and tasks. New products don't have to start from scratch each time.

The reason why the market often oversimplifies the value of medical imaging is that outsiders tend to view it as merely “diagnostic software.” In reality, what truly determines its upper value limit is whether the underlying model can continuously generate products.

Data Search AI has completed the first step of validation. The AI AutoVision developed based on iMedImage has obtained a Class III medical device registration certificate. The threshold for medical AI has thus entered the pricing stage. Professional data, clinical validation, and regulatory approvals lengthen the implementation cycle but also increase the difficulty of replicating similar products.

The second step is to transform a one-time productization into a repeatable production process.

Based on iMedImage, Data Search AI provides model training and deployment processes through iMedMaaS, while iMedLoop continues to connect annotation, evaluation, release, and application feedback. Hospitals can develop models based on their own data, and Data Search AI undertakes different projects on the same foundation.

If this system can continue to operate effectively, what Data Search AI delivers is a set of capabilities for producing and delivering medical imaging models.

The answer to the ranking thus becomes clear.

Data Search AI ranks high not only due to its already approved products but also because behind these products lies a reusable foundation and a delivery path for further expansion into more diseases, departments, and tasks.

However, the ranking on the list is merely a price result within a specific observation period and cannot replace operational performance. If product approvals, model deployments, and customer payments can occur consecutively, the high scores on the list will transition from stage-specific price performance into more solid operational results.

Of course, more importantly, what is being re-recognized is not just Data Search AI but also the scalability potential of medical imaging as it evolves from single-point products to a general-purpose foundation.

The business landscape of AI companies is starting to look different.

Looking back at this ranking, it provides more than just a snapshot of the AI company market.

In the past, some of the market's skepticism about the Hong Kong Stock Exchange stemmed from a lack of supply of new economic assets. Today, with the emergence of models, chips, enterprise services, AI-driven pharmaceuticals, and medical imaging, the Hong Kong Stock Exchange is undergoing a more comprehensive round of AI industry transformation.

However, as assets proliferate, labels lose their effectiveness, and comparisons become more nuanced. Beyond technology, revenue structure, delivery capabilities, regulatory thresholds, and productization efficiency will all influence pricing.

As the AI industry has developed to this point, the technological foundation continues to advance, but industry competition has begun to shift to the next level: what scenarios AI has entered and what products and capabilities it can deliver.

From enterprise services and industrial manufacturing to AI-driven pharmaceuticals and medical imaging, the answers are becoming increasingly tangible.

Source: Hong Kong Stock Research Society