Summary of Key Points
As a leading AI company in China, Zhipu's stock price experienced extreme fluctuations after its listing on the Hong Kong Stock Exchange in 2026 (soaring from an issue price of HK$116 to HK$2980 before rebounding following news about A-share listing preparations), reaching a peak market value of over HK$1.27 trillion, second only to Tencent and Alibaba. Its business model focuses on B-side services and an open-source ecosystem, with code capabilities comparable to those of the American company Anthropic at a more affordable price, making it a potential alternative in China. However, the company's heavy investment in research and development (R&D) has led to continuous losses. The current market value of HK$1 trillion, which corresponds to a price-to-sales ratio of over 1,000 times, raises questions about the rationality of its valuation, and there is uncertainty as to whether it will meet market expectations in the future.
Detailed Analysis
1. Stock Price Volatility: A Mirror of Market Sentiment
Zhipu's stock price has been highly volatile, rising from HK$116 to HK$2980 within six months (a 25-fold increase) after its listing, only to plummet in July due to news of the withdrawal of its A-share listing application. The subsequent rebound was driven by market concerns about changes in its financing plans. This volatility reflects the extreme sensitivity of the market to the AI industry:
- Reasons for the Rise: The AI sector was booming, and Zhipu, as a prominent player in China with innovative models (GLM series) and a strong commercial strategy (B-side services and open-source), was seen as a potential Chinese version of Anthropic.
- Reasons for the Fall: The withdrawal of the A-share application caused investor panic, fearing changes to its funding plans.
- Rebound: Once the situation was clarified, market confidence in Zhipu's core capabilities returned.
In short, stock price movements reflect the balance between market expectations for AI’s future and short-term news developments.
2. B-side Services + Open-source: The Dual Engines of Profitability
Zhipu generates revenue through two main channels: serving enterprises (B-side) and providing open-source models:
- B-side Services: It offers customized model deployments on clients' servers and cloud-based services via APIs. In 2025, its B-side revenue amounted to HK$724 million, with 50% coming from large-scale enterprise deployments; nine leading internet companies, including ByteDance and Alibaba, are using its models.
- Open-source Strategy: Zhipu released the world's first Chinese-English bilingual model, GLM-130B, in 2022, and made its flagship model fully open-source in 2025. Using a permissive MIT license that allows free commercial use, it attracted 4 million developers worldwide. These developers later opted for paid API services, driving cloud revenue from 4.5% in 2022 to 26.3% in 2025. This closed-loop model of "developers → paying users → enterprise customers" is key to its profitability.
3. Cost-effectiveness and Competitiveness
Zhipu's success can be attributed to its user-friendly and affordable approach, as well as its ability to capitalize on Anthropic’s shortcomings:
- Competitive Pricing: Its GLM-5.2 model ranks second globally in code generation capabilities (after Anthropics Claude Fable5) but costs only $0.9 per million tokens, compared to Claude's $7.7.
- Market Access: Anthropic discontinued services for Chinese users in 2024 and was criticized for monitoring and banning them in 2026, raising security concerns from the Chinese Ministry of Industry and Information Technology. Many users turned to Zhipu, effectively shifting business to this Chinese-based alternative.
4. The Quest for AGI: A Long-term Investment
Zhipu's vision is to develop AGI (artificial general intelligence), a goal that requires substantial investment:
- The Challenge of AGI: Building an AGI involves solving problems across multiple domains, which is costly.
- R&D Expenses: In 2025, Zhipu spent HK$3.18 billion on R&D, a 45% increase from the previous year, resulting in a annual loss of HK$4.7 billion, mainly due to R&D expenses.
- Progress: The company is still in the third stage of its AGI roadmap (self-learning) and has a long way to go before achieving true AGI.
In summary, while Zhipu’s vision is ambitious, the high cost of R&D means it faces significant challenges in the short term.
5. Is the Valuation Excessively High?
With a peak market value of HK$1.27 trillion and annual revenue of only HK$724 million, Zhipu's price-to-sales ratio is 1,750 times, which is much higher than competitors like Anthropic (20-30 times) or domestic companies like Tencent and NetEase (under 8 times). Even MiniMax in the same industry has a ratio of 150 times. This suggests high market expectations, but whether these will be met remains uncertain:
- Technical Challenges: Zhipu’s GLM-5.2 model is not as powerful as Anthropics Claude Fable5.
- Competitive Landscape: The domestic AI market is competitive, with companies like Baidu and Alibaba also developing large-scale models.
- Opportunities: The global AI market is growing rapidly, with enterprises spending $2.1 trillion on AI by 2029, giving Zhipu an advantage through its B-side services and open-source ecosystem.
In conclusion, Zhipu’s current market value reflects high expectations, but whether these will be justified depends on its ability to convert its technological vision into actual revenue. Time will tell, and there is considerable uncertainty involved.
(End of Analysis)