Summary of Key Points
As the “first stock in the global large-model sector,” Zhipu AI saw its value soar 24 times within half a year after its listing in Hong Kong in January this year, reaching a peak market capitalization of HK$1.33 trillion. However, recent events such as the release of restricted shares and rumors about the withdrawal of its A-share IPO caused the stock price to plummet by 15% in one day, losing HK$110 billion. After clarification, the stock price rebounded by 10%, recovering HK$70 billion, resulting in a “bungee-style” fluctuation. The high valuation of HK$700 billion, supported by revenue of HK$724 million, has sparked controversy. This reflects both the potential of the technology and business growth as well as the capital expectations driving the market. However, it also exposes the bubble in the AI industry: while the technical barriers are not as insurmountable as imagined, the excessive valuation created by capital speculation may not be sustainable. Nevertheless, there is indeed a real demand for AI services, which could accelerate the maturation of the industry. For the education sector, the benefits of AI belong to those who can integrate it into practical applications to solve real problems, rather than just chasing trends.
I. Zhipu’s “Bungee” Fluctuations: Not Caused by Rumors, but by Excessive Valuation
The sharp fluctuations on July 6-7 were ostensibly triggered by the combination of restricted share releases and rumors, but in reality, they reflected the inherent fragility of a highly valued stock:
- Pressure from Restricted Shares: On July 8, 25.68 million restricted shares were released, with a market value of over HK$460 billion, creating significant selling pressure that caused the stock price to drop by 14.67%.
- Rumors about IPO Withdrawal: The market speculated that Zhipu was withdrawing its A-share listing application, shattering the expectation that Hong Kong stock prices would converge with those of the A-share Science and Technology Innovation Board (which favors unprofitable hard-tech companies with higher valuations). Without this premium, the stock price dropped another 9%.
- Clarification and Rebound: The company issued a statement in the afternoon confirming that its listing application was ongoing and that it had reported the rumors to authorities. This led to a 10% increase in the stock price. However, the root cause was the high valuation; any negative news could trigger such significant fluctuations.
II. A 24-Fold Increase is Not Just Speculation: Three Pillars of Support
Zhipu’s price rise from HK$116 to HK$2980 was not based on mere speculation. There are three solid reasons:
- Technical Strength: Developed by the Tang Jie and Zhang Peng team at Tsinghua University’s KEG Laboratory, Zhipu’s GLM model is among the best in China.
- Business Success: Revenue for 2025 is projected to be HK$724 million, more than doubling. More importantly, the annual recurring revenue (ARR) from its MaaS platform has increased from less than HK$30 million to HK$1.7 billion, indicating ongoing demand for its API services—a rare achievement among large-model companies.
- Capital Expectations: Investors bet that Zhipu would be listed on the A-share Science and Technology Innovation Board, expecting Hong Kong stock prices to align with the higher valuations there. They bought Hong Kong shares in anticipation of a price difference.
III. Is HK$700 Billion in Market Capitalization Excessive?
Several comparisons highlight the exaggerated nature of Zhipu’s valuation:
- Price-to-Revenue Ratio: Zhipu’s ratio is 900 times, meaning it would take nearly a thousand years to earn back its current market value at current revenue levels. In contrast, OpenAI’s ratio is only 17 times (revenue HK$88.3 billion, valuation HK$1.5 trillion), and Nvidia’s is 30 times (revenue HK$800 billion, valuation HK$25 trillion).
- Industry Comparison: Other leading large-model companies in China have much lower valuations: Yuezhi Dianmian at HK$22 billion, Baichuan at HK$20 billion, and MiniMax at HK$934 billion. Industry experts consider Zhipu’s valuation to be excessively high.
- Overestimating the Future: A market capitalization of HK$700 billion may include projected growth for the next 5-10 years. Any slowdown in MaaS revenue growth, competition from rivals, or delays in the A-share listing could lead to a price crash.
IV. Three Layers of Bubbles in the AI Industry: Opportunities Within
Zhipu’s story reflects three layers of bubbles within the broader AI industry:
- Technical Bubble: While many believe large-models have high barriers, the technology is relatively transparent (based on Transformer frameworks, data, and computing power), and several companies claim to be close to GPT-4. The real challenges lie in data availability, use cases, and ecosystem integration (e.g., applying AI in education or healthcare).
- Capital Bubble: Early-stage AI companies can raise significant funds with support from large corporations or impressive PPTs. However, this is not irrational; AI represents a once-in-a-decade technological revolution, and investors are eager to participate.
- Application Bubble: Profitable applications are still limited, and even OpenAI incurs substantial losses. Nevertheless, its annual revenue of over HK$80 billion shows that commercialization is feasible, though it’s still in its early stages. The demand for AI services is real, although not yet widespread.
V. Implications for the Education Sector
The news highlights the importance of practical applications for AI in education:
- Don’t Chase Trends: The education sector needs to focus on integrating AI into real solutions, such as grading assignments, personalized tutoring, and content generation, rather than just chasing short-term gains.
- AI is a Long-Term Revolution: Like the internet and mobile technology, AI will transform education. Educators should focus on implementing practical solutions rather than reacting to market fluctuations.
Conclusion
Zhipu’s extreme price movements are not a sign of a myth or a bubble bursting, but normal fluctuations in a technological revolution. For individuals, there’s no need to rush into buying stocks or worry about the future of AI. The true value of AI lies in its practical applications in specific industries, especially education. By integrating AI into classrooms, homework, and learning processes, we can harness its potential to transform education.
---
This translation maintains the structure and tone of financial journalism, adapting the content to fit the target audience while ensuring accuracy and clarity.