Interest in Hong Kong's IPO market remains intense, with another offering proving exceptionally difficult to secure. The public offer for HQVT (01392), dubbed the "first physical AI stock," concluded successfully on June 16, 2026. By the end of the subscription period, the stock was oversubscribed by nearly 7,200 times, attracting over 300,000 applicants and total subscription funds exceeding HKD 440 billion, placing it among the top five Hong Kong IPOs for the year. The simultaneous influx of institutional and retail capital has made it one of the hottest AI and hard-tech investment opportunities in the recent Hong Kong market.
The overwhelming demand reflects strong market recognition of HQVT's core competitive strengths and growth potential. As the dual leader in China's multispectral AI industry and multispectral AI large model services, the company has built a complete closed-loop system from multispectral perception modules to end-cloud collaborative solutions, supported by its fully self-developed technology stack, creating a significant technical moat.
Financially, its performance has been robust. From 2023 to 2025, revenue grew from RMB 117 million to RMB 523 million and then to RMB 669 million, achieving a three-year compound annual growth rate of 138.9%. Gross margin improved consistently from 12.2% in 2023 to 22.3% in 2025, indicating sustained enhancement in operational quality. In 2025, revenue from multispectral AI large model services accounted for 53.1%, becoming the primary income source. Furthermore, adjusted net profit attributable to shareholders was RMB 43 million in 2024 and RMB 55 million in 2025, with the company achieving profitability in 2024, demonstrating a clear growth trajectory that underpins its appeal to investors.
Designated as a national-level "Specialized, Refined, Distinctive, and Innovative" key enterprise, the company's self-developed "Zhiyuan Qiyuan" multispectral large model has been selected as one of Guangdong Province's first AI industry large models, completed algorithm filing with the national cyberspace authority, and achieved scaled commercial deployment in 2024. Benefiting from multiple policy supports, the company holds 158 patents and 46 software copyrights, and has led or participated in formulating nearly ten national and industry standards, highlighting its technical authority.
Its end-side inference speed reaches as fast as 50 milliseconds, leading the industry average. The technology has achieved scaled implementation in high-value scenarios such as data centers, power systems, new energy, and energy storage, with a client base covering several leading state-owned enterprises and large telecom operators.
From an industry perspective, data from Frost & Sullivan shows that HQVT currently holds a 3.3% market share in China's multispectral AI sector, ranking first. The combined market share of the top five companies is only 8%, indicating that HQVT, leveraging its first-mover advantage, is firmly positioned as the industry leader with substantial room for future market consolidation and share growth.
Building on its leading technology, HQVT's business comprehensively covers data centers, power, new energy, energy storage, smart cities, industrial and commercial security, and IoT facility management. It is also gradually expanding into emerging areas such as food safety, intelligent driving systems, and embodied AI. Its products act as AI "super eyes" for the physical world, automatically and comprehensively capturing various safety risks, with clear and tangible application value.
With the subscription period now closed, market sentiment is generally optimistic regarding HQVT's grey market and listing performance. Considering the recent recovery trend in Hong Kong tech stocks, the strong performance of similar hard-tech listings, and the positive market sentiment fueled by this high oversubscription rate, several institutions anticipate the stock may see favorable activity in the grey market, positioning it as a focal point in the recent Hong Kong IPO landscape.
Comments