Exclusive | 6 billion, Jishi Vision rings the bell at the Hong Kong Stock Exchange! 34-year-old Macau youth brings the "First AI Visual Large Model Stock"

Ask AI · How did Chen Zhenjie transform from a campus student into a leading figure in AI computer vision?

The “IPO Full Observation” column focuses on companies conducting their initial public offerings, covering entrepreneurs’ startup experiences and success stories, analyzing the company’s business model and operating performance, and revealing how capital forces such as VCs and CVCs empower the company through their investments.

Author | Man Di

Editor | Guan Ju

Image source | Jishi Jue (Extreme View)

An AI computer vision algorithm platform jointly created by three classmates is making its debut today with an IPO on the Hong Kong Stock Exchange.

Today (March 30), Shandong Jishijue Technology Co., Ltd. (hereinafter “Jishijue”) listed on the HKEX, with an issue price of HKD 40 per share. It surged nearly 50% at the open, and its market capitalization exceeded HKD 6.8 billion (about RMB 6 billion. As of the time of publication, market capitalization exceeded HKD 9 billion).

March 30, 2026 — Chen Zhenjie, founder of Jishijue, at the listing ceremony on the HKEX main board of Jishijue

The 34-year-old founder, Chen Zhenjie, is from Macau. He earned a graduate degree from Peking University Guanghua School of Management and an undergraduate degree from Lingnan College of Sun Yat-sen University. Ten years ago, together with two fellow undergraduate alumni from Zhongshan University, he co-founded a business, targeting the AI computer vision algorithm track, and established the company “Jishijue.”

As an AI computer vision solutions provider, Jishijue provides end-to-end enterprise-grade solutions covering development, deployment, and management for enterprises across industries. Currently, Jishijue has built a global community consisting of hundreds of thousands of AI algorithm developers. Its AI computer vision algorithm marketplace showcases 1,517 algorithms and has already served more than 3,000 government and enterprise customers in aggregate.

In 2024, Jishijue began offering large-model solutions to enterprises, which became the company’s second growth curve.

In recent years, Jishijue’s revenue has grown rapidly. Operating revenue increased from RMB 102 million in 2022 to RMB 128 million in 2023, and further to RMB 257 million in 2024. For the first three quarters of 2025, revenue was RMB 136 million. At the same time, the company’s gross margin rose from 30.6% in 2022 to 44.9% for the first three quarters of 2025.

In the shareholder structure after Jishijue’s listing, Chen Zhenjie, Luo Yun, and Hengqin Jili are acting in concert, holding a combined stake of about 26.54%, making them the single largest shareholder group. Among them, Chen Zhenjie holds 14.27%, Luo Yun holds 3.90%, and Hengqin Jili holds 8.37%.

China US Chuangxing Fund is the largest institutional shareholder, holding 9.53%. In addition, Qingdao Economic and Technological Development Zone Financial Investment Group Co., Ltd. holds 4.84%, Qualcomm (China) Holdings Co., Ltd. holds 4.42%, Qingdao Tianqi Frontier Technology Investment Fund Partnership (Limited Partnership) holds 4.30%, and Shantou CR Innovation Equity Investment Fund Partnership (Limited Partnership) holds 4.27%.

01 Anchoring the AI vision track ten years ago

Chen Zhenjie’s interest in business first became apparent during his undergraduate years. In his first year, while studying in the School of Life Sciences at Sun Yat-sen University, Chen Zhenjie found that he was always more interested in business-related content. So not long after entering the university, he transferred his major to the Lingnan College of SYSU to study economics. After graduating, he chose to go to Peking University Guanghua School of Management to pursue further studies.

During his time in the master’s program in enterprise management at Peking University Guanghua, Chen Zhenjie, like other classmates, also chose internships at large companies. He tried consulting-related work at consulting institutions such as Bain&Company and KPMG Advisory, and also interned in Tencent’s mobile game strategy department. But this fixed-content work model inside companies always made Chen Zhenjie feel constrained. “Back then, when I was interning at a consulting firm, it felt like I would go to work and wait for the day to end.” Chen Zhenjie said. He considered himself someone who preferred making top-level strategic plans, and validating those ideas through execution so they could truly take root.

So, during his graduate studies, he reached out to two fellow undergraduate classmates from the School of Life Sciences at Sun Yat-sen University—Luo Yun and Chen Shuo—deciding to start a business together.

A photo of Chen Zhenjie (middle) and Luo Yun (left) and Chen Shuo (right) in the early days of entrepreneurship

Ultimately, around 2015, they chose the computer vision track and founded the company “Jishijue,” the one that exists today.

In the startup atmosphere when the internet was all the rage, the AI field was not receiving much attention. Chen Zhenjie had his own considerations. “We chose AI not based on complex technical judgments, but on a simple logic: human labor is getting more expensive, and replacing humans with machines is inevitable. In the past, machines replaced part of manual labor. In the future, intelligent applications will definitely gradually replace mental labor.” Chen Zhenjie made a basic judgment about industry trends. “Therefore, the team decided to enter the long-term race track of artificial intelligence and accept the reality that the industry might need a few years or even ten years to break out.”

After deciding on the startup direction, the three founding team members each took on their own responsibilities. Chen Zhenjie, who was studying at Peking University Guanghua, handled commercialization and investment/financing. Luo Yun, who was studying for a PhD in artificial intelligence at The Hong Kong University of Science and Technology, was responsible for technical development. Chen Shuo, who was good at design, was in charge of the company’s product design.

Not long after it was founded, and still in its early stage, Jishijue faced tests from all directions.

Jishijue’s projects mainly serve enterprises. However, without sufficient market resources, it was very difficult to establish cooperation with companies. Therefore, before securing angel investment, they had almost no commercialization business. Meanwhile, the company needed to first conduct fundamental technical exploration and prototype validation, so it could better serve enterprise customers.

In 2015, AI was still a very new track, and many investment institutions did not have investment plans. “Back then, we basically had just one demo, so getting investment was quite difficult. We could only win by having more conversations and talking to investors.” During the Spring Festival that year, through introductions from a senior, Chen Zhenjie ultimately obtained RMB 2 million in angel investment from Hu Langtiao, founding partner of China US Ventures.

02 “Make algorithms as easy to use as an app store”

Jishijue’s most core business is to meet enterprises’ needs for AI computer vision recognition and to provide large-model solutions. This includes R&D, deployment, and end-to-end management services for AI computer vision solutions. It has already provided end-to-end services for over 100 industry scenarios for government and enterprise customers, including industrial, energy, retail, and transportation.

Jishijue’s business model

In 2015, the overall artificial intelligence environment was still in a phase of weak AI. At the early stage, Jishijue’s team chose to develop algorithms in-house. During experiments applying to real-world scenarios, Jishijue first selected a niche area—“customer flow analysis”—mainly serving the retail industry and helping malls and brand stores analyze offline foot traffic and conversion rates. This also became Jishijue’s first business entry point and source of cash flow.

But soon, Chen Zhenjie noticed that algorithms developed under this model were often only able to solve single problems (such as face recognition) and could not address diverse vision recognition needs across industries such as industrial and retail. Therefore, inspired by the internet platform model they had experienced in previous internships, Chen Zhenjie came to the idea that, by leveraging the scale effects brought by a platform model, they could build an algorithm platform that connects intelligent supply and demand—linking enterprise customers with algorithm needs to individual algorithm developers.

This is the early blueprint of the open algorithm development platform that Jishijue first built for AI algorithm developers, which is now Jishijue’s “Jishi (Jishiming)”. Jishi focuses on computer vision algorithms, providing infrastructure support for algorithm development.

As the developer community gradually matured, Chen Zhenjie and the team began to try taking on more diverse customer needs. However, acting purely as a “middleman” is not easy; the initial simple matchmaking model often ran into difficulties in delivery and operations/maintenance.

For this reason, Jishijue began investing in in-house R&D of AI underlying infrastructure and building an algorithm development platform as early as 2016. To ensure the stability and maintainability of algorithm delivery, Jishijue standardized the development process and transformed from a “middleman” into a genuine provider of AI computer vision solutions and large-model solutions.

Chen Zhenjie described the pairing of Jishijue and developers this way: “Previously, developers might have had to complete the end-to-end closed loop themselves. Now we keep the last-mile customized algorithm module for developers to do, and our most important job is to build the foundational infrastructure.”

As a result, Jishijue’s business focus shifted to delivering integrated AI solutions that include standardized AI computer vision solutions, customized AI computer vision solutions, and software-defined AI solutions. As of September 30, 2025, Jishijue’s AI computer vision solution algorithm marketplace has displayed more than 1,500 algorithms covering more than 100 industries. Meanwhile, after a decade of accumulation, the number of AI algorithm developers worldwide exceeded 100,000, and the company has cumulatively provided robust infrastructure platforms and broad AI solutions to 3,000 customers.

03 “Keep the direction right, and keep iterating with flexible paths”

Based on its R&D infrastructure (i.e., the AI vision language model and Jishi), Jishijue then provides solutions for enterprises through two delivery platforms: Jixing and Jizhan.

In 2019 and 2021, Jishijue launched its two platforms, Jixing and Jizhan. Jixing is an AI algorithm inference and deployment platform, mainly helping enterprises quickly build and deploy AI solutions. Jizhan is a private AI platform designed specifically for large enterprises, government agencies, and academic research institutions.

In Chen Zhenjie’s view, as a technology-centered enterprise, it needs the ability to quickly adjust as technology changes with the times—especially in the current AI wave, where the speed of iteration and adjustment must be fast enough to stay at the industry forefront.

Earlier, in the AI 1.0 era, if Jishijue wanted to recognize a new scenario task, developers needed to rewrite a new algorithm to make it work. But when the industry’s overall large-model technical capabilities began to grow rapidly, Chen Zhenjie became keenly aware that vertical industry models would face new opportunities.

Therefore, in-house R&D of AI vision language large models obviously has important value for a company focusing on computer vision. Leveraging more than ten years of algorithm engineering experience and data in monitoring camera recognition, Jishijue built its own AI vision language large model.

“Even our own evaluation results on the customer side confirm this. Compared with open-source models that currently focus more on general language capabilities, the vision-language models we developed—focused on recognition tasks under perspectives such as monitoring cameras and drones—show better recognition performance in specific scenarios than general-purpose large models.” Chen Zhenjie said.

In addition to in-house R&D of vision large models, Jishijue developed an agent application development platform called “Jizhi,” using existing general-purpose large models. For government and enterprise customers, building a large-model agent development and management platform can meet deployment needs for key scenarios such as knowledge Q&A, process automation, system coordination, and human-machine interaction.

“Our positioning on this platform is to hope to provide system support for future large enterprises to transform their business management operating flows based on Agent-native [process] changes,” Chen Zhenjie said. Large enterprises’ adoption of AI is an obvious demand. Therefore, the old set of enterprise ERP systems built for the pure digital era will gradually become less suitable. In the future, inside enterprises, people will need to coordinate with many intelligent agents to complete many tasks. Therefore, Chen Zhenjie believes that a secure operating system that is suitable for collaboration between humans and Agents will definitely be a future trend in enterprise needs.

As technology keeps iterating, in such a rapidly developing era of artificial intelligence, Jishijue’s only choice is to “keep up.”

According to the prospectus, Jishijue’s operating capabilities have strengthened in recent years. Revenues in 2022, 2023, and 2024 were RMB 102 million, RMB 128 million, and RMB 257 million, respectively. Gross profit was RMB 31.08 million, RMB 33.08 million, and RMB 100 million, respectively. Profit during the period was -RMB 60.72 million, -RMB 56.25 million, and RMB 8.71 million. In the first three quarters of 2025, Jishijue’s revenue increased year over year by nearly 72%.

Specifically, as Jishijue’s business has diversified, the proportion of revenue contributed by each business line has also been changing. In 2022 and 2023, revenue from AI computer vision solution business was the source of all of Jishijue’s revenue. In 2024, the newly launched large-model solution business began generating revenue; AI computer vision solutions accounted for 75.9% of revenue, while the new large-model solutions accounted for 24.1% of total revenue.

Jishijue’s revenue

In the view of Hu Langtiao, an investor at China US Chuangtou, “Jishijue has shown obvious growth every year, and every year has surprised us.” Besides the angel round investment, China US Chuangtou also increased its follow-on investment in subsequent rounds.

04 Why bet on Jishijue?

From when Jishijue received its first round of financing from China US Chuangtou in 2015 to today, it has already been 11 years. Almost every one to two years, Jishijue will conduct a new round of financing.

Looking back on why they were optimistic about Jishijue even ten years ago, Hu Langtiao said, “When we first met, Zhenjie was only 23 years old, yet he already had ‘a leader’s presence’—full of passion and strongly proactive. What we wanted to find were exactly such restless young people with a strong desire for achievement. This is also a basic model we use to identify talent.”

From the team perspective, the early founding team of Jishijue at that time included Luo Yun and Chen Shuo, among others. The three also happened to be undergraduate classmates. Hu Langtiao believed that they had good mutual understanding and complementary strengths, so he was optimistic about such a vibrant startup team.

In addition to the “people” factor, the track is undoubtedly another element in capital’s investment judgment. At that time, Jishijue’s computer vision customer flow analysis project was highly aligned with the AI track that China US Chuangtou favored, making the initial financing seem like a natural outcome.

Chen Yong, Managing Director of China Resources Innovation Fund, recalled his A+ round investment in Jishijue and said it was based on a fundamental logic rooted in technology: in the end, it is those innovative technologies that can drive changes in production relationships and significantly improve production efficiency that can become industry leaders and grow bigger.

“Over the more than 10 years since we came to know Jishijue, we have seen that, by leveraging their industry practices in traditional industries and their deep insights, they can combine the development of AI technology with solving the pain points and difficulties of traditional industries. This is a cross-border, cross-industry composite capability. This is the capability we believe has the strongest differentiation. That’s also why, after years of refinement and transformation, Jishijue has leapt to become a leading player in China’s AI vision algorithm marketplace.” Chen Yong said.

Subsequently in 2018, Mao Song, Managing Director of Qualcomm Venture Capital China, met Chen Zhenjie through an event organized by Chuangye Bang called “Entering Qualcomm,” and his unique business model also left a deep impression on the Qualcomm team.

That same month of October, at a startup competition jointly organized by Qualcomm Venture Capital, Sequoia China, and Chuangye Bang, Jishijue qualified into the finals as a TOP 10 company. After further discussions with Chen Zhenjie, Mao Song ultimately decided in the second half of 2019 to invest in Jishijue’s B round, and continued to make additional investments in later rounds.

When describing his impression of Chen Zhenjie, Mao Song used the word “steady.” “He is an entrepreneur who thinks deeply and makes independent judgments. For example, this can be seen in how he manages the company’s cash flow, keeping the company in a ‘ready to advance and able to retreat’ state.” Mao Song believes it is indeed not easy for Chen Zhenjie to withstand the industry boom over a decade and persist in exploring a unique path that combines developer community operations with commercial success.

The AI computer vision solutions industry and the large-model solutions industry in which Jishijue operates are highly competitive. On one hand, industry technology develops rapidly; customers’ needs and preferences change quickly; new solutions and services are frequently launched. Meanwhile, new industry standards and conventions keep emerging.

According to Frost & Sullivan, the market size for China’s emerging enterprise-grade computer vision solutions grew from RMB 2.2 billion in 2020 to RMB 11.1 billion in 2024, with a compound annual growth rate of 49.9%. It is expected to reach RMB 97 billion by 2029, with a compound annual growth rate of 54.3%.

Currently, measured by sales revenue in China’s emerging enterprise-grade computer vision solutions market in 2024, Jishijue ranks eighth among all market participants, with a market share of 1.6%.

Regarding judgments on Jishijue’s business model, several investors share similar views.

Mao Song believes that, by category, Jishijue can be classified as a “platform-type company.” Under such a business framework, through a platform model, it can meet the dispersed long-tail AI demands across industries at low cost and high efficiency—this is Jishijue’s unique advantage and is fundamentally different from a pure project-based or product-based company. Specifically, this model greatly reduces the development cost of a single solution, enabling it to economically serve a massive number of scattered small customers. At the same time, as time goes on, its accumulated developer resources, industry solutions, and algorithm pool form a powerful flywheel effect and competitive barriers.

In Hu Langtiao’s view, the company positioned itself early on to build AI “infrastructure” and a “highway network,” rather than merely being a technology fortress. The company’s “Algorithm Marketplace,” the largest AI developer community in China that it built, connects a large number of developers, partners, and customers, forming an ecosystem barrier that is hard to replicate quickly. This is why, in the industry, Jishijue is viewed as a partner rather than a competitor.

Jishijue was originally founded in Shenzhen. In November 2021, Jishijue relocated its headquarters to Qingdao, Shandong, and introduced multiple state-owned capital entities in Shandong, such as Qingdao Economic and Technological Development Zone Financial Investment Group Co., Ltd., Shandong LuHai Linkage Investment Fund Partnership (Limited Partnership), and Qingdao Guotou Investment & Management Co., Ltd., among others.

05 Entrepreneurship should “return to the roots,” becoming the kind of person it should become

As a company mainly serving government and enterprise customers, when discussing Jishijue’s business positioning, Chen Zhenjie believes it often depends on the style of the founding team.

“People like us aren’t really suitable for the C-end, because the C-end is more about being opinion leaders, who need to explore and uncover the needs of the general consumers in advance. And our strength is more about providing professional services with specialized capabilities.” In Chen Zhenjie’s view, for entrepreneurs, whether it’s the content of starting a business or the style of the enterprise, ultimately it all comes back to the entrepreneur’s own roots.

Therefore, there is not just one template for successful entrepreneurs. Instead, one should recognize one’s own root attributes, build a corporate culture on that basis, and have the team compensate for the individual’s shortcomings. “The commonality in entrepreneurship is that you like to try things, then you need to continuously learn and adapt to changes. Once you solve those, I think the foundation below can be constructed in your own style. Ultimately, you can balance that shortcoming through the complementary pairing of the team.”

Regarding the development direction of Jishijue after listing, Chen Zhenjie candidly said that it will continue to strengthen the current AI vision language large model so that it can address more general scenario recognition needs. The other part is to continue helping enterprises carry out AI-based transformation of organizational processes.

This article is an original work by Chuangye Bang. Without authorization, it may not be reproduced. Otherwise, Chuangye Bang reserves the right to pursue legal responsibility. If you need to republish it or have any questions, please contact editor@cyzone.cn.