An examination of Kingzhiwei's prospectus reveals a business that appears promising in terms of both market position and customer value. Yet a puzzling question emerges: why has the company seen no capital intervention since 2023? Is it that the market does not endorse the company's business model, or are there other considerations in the company's financing timeline? Notably, the company's cash flow remains persistently negative, customer stickiness is deteriorating, and growth lags behind industry rates - raising the question of whether its project-centric business model is fundamentally flawed. Additionally, the company carries 2.1 billion yuan in redeemable liabilities on its books while holding less than 100 million yuan in cash, suggesting an urgent need for an IPO to relieve the pressure.
Zhuhai Kingzhiwei Artificial Intelligence Co., Ltd. recently submitted its prospectus to the Hong Kong Stock Exchange, planning to list on the Main Board via H-shares, with Guotai Haitong and BOCI serving as joint sponsors. This marks a reapplication following its initial filing on December 15, 2025, which lapsed on June 15, 2026, after the six-month validity period expired. The proceeds from this IPO are intended for the following purposes: strengthening R&D capabilities and continuously optimizing the AI solution platform and AI products; investing in the development of industry-specific solutions; expanding sales and marketing network coverage to more business areas and regions; deepening Hong Kong market presence and actively exploring overseas markets; and the remainder for working capital and general corporate purposes.
Is the listing urgent? 2.1 billion yuan in redeemable liabilities loom overhead
According to the prospectus, Kingzhiwei was founded in 2016 as an AI enterprise focused on providing AI digital employee solutions and enterprise-level intelligent agent solutions. Through its self-developed AI solutions, the company assists enterprises in accelerating their digital and intelligent transformation. Zhuhai Kingzhiwei integrates AI algorithms, large language models, and robotic process automation (RPA) technology with industry scenarios, enabling various tasks to be handled by AI digital employees and intelligent agents, reshaping organizational productivity through human-machine interaction.
However, the company remains in a loss-making state. The prospectus shows that from 2023 to 2025 and the first half of 2026, the company's revenue was 217 million yuan, 243 million yuan, 256 million yuan, and 65 million yuan respectively, with corresponding net losses of 63 million yuan, 122 million yuan, 346 million yuan, and 109 million yuan. Adjusted net losses for the same periods were 49 million yuan, 3 million yuan, 27 million yuan, and 57 million yuan respectively.
Notably, the company's core business cash generation capacity remains weak, with operating cash flow consistently negative. Against this backdrop of weak core business cash generation, the company's book cash has been significantly depleted in recent years. As of the first half of 2026, the company's book cash stood at only 72 million yuan. It is worth noting that the company has not conducted any financing since its C3 round in 2023. According to public information, the company has completed five rounds of financing, raising approximately 686 million yuan in total, with investors including market-oriented institutions such as Qiming Venture Partners, Hillhouse Capital, and Shunwei Capital, as well as state-backed capital including China Development Bank Manufacturing, Guangdong Yuecai Investment, and CITIC China Capital, along with industrial investors such as Kingstar Technology and Wens Group. After completing the C3 round in June 2023, Kingzhiwei's post-investment valuation was approximately 3.075 billion yuan.
It should be added that during the financing process, the company signed betting agreements with certain investors, granting some previous investors special rights including liquidation preferences and redemption rights. These rights were terminated before the company submitted its application to the stock exchange, and all other special rights will terminate subsequently. However, if the listing fails, redemption clauses may be triggered. Specifically, if the company voluntarily withdraws its application, the application is rejected by the exchange, the company fails to pass the hearing within 18 months from the application date, or the company fails to complete the listing within six months from the date of passing the hearing, the liquidation preferences and related rights of previous investors will be reinstated. As of the first half of 2026, the company's redeemable liabilities reached 2.126 billion yuan.
Does the project-based business model have flaws?
According to the prospectus, the company's business appears to have growth potential in terms of both market position and customer value. Yet the puzzling question remains: why has there been no capital involvement since 2023? Is the market not endorsing the company's business model, or are there other considerations in the financing timeline? According to Frost & Sullivan data, by revenue in 2025, Kingzhiwei was the largest AI digital employee solution provider in China, ranking first in the financial services sector for three consecutive years from 2023 to 2025. The company's leading advantages are particularly notable in financial services and government affairs.
Kingzhiwei focuses on high-value industry-leading customers. As of June 30, 2026, the company served over 1,600 customers, covering key sectors including financial services (primarily banks and securities firms), government affairs, and manufacturing. In the financial services sector, the company's coverage as of June 30, 2026 included all six large state-owned banks and over 130 securities companies, covering more than 90% of China's securities companies. The prospectus shows that the company launched its K-APA platform in 2021. The K-APA solution focuses on addressing automation problems with clear rules that enterprises encounter during digital transformation, enabling users to automate and intelligently operate software systems, data, external systems, and websites without modifying existing business systems. In 2025, the company launched the Ki-Agent enterprise-level intelligent agent platform. The Ki-Agent solution delivers end-to-end value creation from efficiency improvement to strategic impact, addressing common problems of fragmented processes and low efficiency in traditional enterprises through multi-agent collaboration and process automation. From 2023 to 2025, the company's K-APA revenue was 212 million yuan, 232 million yuan, and 240 million yuan, accounting for 97.7%, 95.3%, and 93.6% of total revenue respectively. Ki-Agent revenue was 4.96 million yuan, 11.36 million yuan, and 16.45 million yuan respectively. In the first half of 2026, Ki-Agent revenue increased to 18.99 million yuan, accounting for 29.2% of total revenue.
Kingzhiwei adopts a business model combining project-based, subscription-based, and operations and maintenance services. For large customers with high customization requirements, complex system structures, or strict data confidentiality requirements, the company typically employs a project-based model, providing software licenses, system integration, process configuration, and custom development based on the customer's IT environment, charging according to milestones such as deployment and acceptance. From 2023 to 2025 and the first half of 2026, the company's project-based revenue was 160 million yuan, 170 million yuan, 197 million yuan, and 47 million yuan respectively, accounting for 74.1%, 69.7%, 76.9%, and 71.6% of total revenue. Clearly, project-based revenue forms the foundation of the company's business, accounting for as much as 70% of revenue, while subscription revenue accounts for less than 20%.
Industry insiders indicate that subscription-based projects may have a higher ceiling compared to project-based models, meaning the market is willing to give subscription-based models higher valuations. First, revenue predictability is stronger - traditional one-time licensing or project-based revenue is susceptible to contract signing pace, acceptance, and procurement cycles of large customers, making future revenue difficult to predict. Under a subscription model, customers pay continuously at regular intervals, allowing vendors to more clearly see future contract revenue and renewal potential, making PS/ARR valuations more viable. Second, customer stickiness is higher - under subscription models, customers do not leave after purchasing but continuously rely on product usage, updates, and services. If the product becomes truly embedded in business processes, customers are more inclined to renew, add purchases, and expand modules, forming high LTV. Third, marginal costs are lower - the core economics of highly standardized subscription software lie in front-loaded R&D investment, but once the product matures, the marginal delivery cost of adding a new customer is far lower than under a project-based model. Fourth, cash flow quality is better - overseas subscription software generally collects payments annually in advance, with deferred revenue and contract liabilities often growing first, while the income statement reflects business growth with a lag.
In fact, Kingzhiwei appears to need improvement in terms of cash flow quality, customer stickiness, and growth. The company's customer retention rate has been declining continuously, from 74% in 2023 to 66% in 2025, and further dropping to 60% in the first half of 2026. This raises questions about whether the company's customer stickiness is deteriorating. Kingzhiwei's revenue growth also appears to lag behind industry growth rates. In 2023, 2024, and 2025, Kingzhiwei's revenue was 216.7 million yuan, 243.5 million yuan, and 256.1 million yuan respectively, representing a compound annual growth rate of 8.7% from 2023 to 2025. According to Frost & Sullivan data, China's enterprise-level AI solution market size grew from 14.3 billion yuan in 2020 to 69.4 billion yuan in 2025, with a compound annual growth rate of 37.1% over the same period.
Why does the company still rely primarily on a project-based model? This may be related to the company's customer structure. As of June 30, 2026, the company served over 1,600 customers, covering key sectors including financial services (primarily banks and securities firms), government affairs, and manufacturing. On one hand, many financial, government, and state-owned enterprise customers have extremely high requirements for private deployment, data security, system integration, process adaptation, and compliance review, making it impossible to directly adopt standardized public cloud products. On the other hand, AI applications, especially intelligent agents and digital employees, require more than just installing software - they involve process decomposition, permission access, knowledge base governance, multi-system integration, result verification, and feedback loops, which inherently carry project-based characteristics.