Summary of Key Points
Zhipu and MiniMax are among the first Chinese large-scale models to list on the Hong Kong stock market. Initially, the market favored MiniMax's "global C-side story" (300 million users, 70% of revenue from overseas). On its first day of trading, MiniMax's stock price doubled, while Zhipu's stock price fell below its issue price. Six months later, the situation reversed: Zhipu's market value exceeded one trillion Hong Kong dollars due to its B-side (government and enterprise) business, making it the first Chinese large-scale model company to reach this milestone. However, its stock price plummeted when its technological leadership was questioned. MiniMax, on the other hand, faced issues such as a loss of C-side users, poor model performance, and copyright lawsuits, resulting in its market value shrinking from HK$41 billion to less than HK$7 billion. Both companies had to cut salaries, issue shares, and seek additional funding to survive. Although they took different approaches—Zhipu focusing on the B-side and MiniMax on the C-side—they both encountered challenges related to rapid technological advancement and unstable business models. Ultimately, both will need to maintain a leading edge in technology to succeed.
A Divided Market at Launch
When the two companies went public in January this year, the market's reaction was starkly different:
- MiniMax was favored: Its stock price rose 109% on the first day, and its market value was twice that of Zhipu. The reason was simple: MiniMax emphasized a global C-side user base of 300 million users across more than 200 countries, with 70% of its revenue coming from overseas, which seemed more appealing than Zhipu's focus on government and enterprise privatization deployments.
- Zhipu faced skepticism: Its stock price fell below its issue price during trading. This was because Zhipu aimed to serve domestic governments and enterprises, providing them with customized models (installed on their servers to ensure data security), which was not as attractive to investors as the C-side approach.
The market believed that the C-side represented the future and offered greater potential for profit. However, six months later, reality proved otherwise.
Zhipu's B-side Success
Zhipu's turnaround was due to its focus on the B-side market:
- A natural fit for government and enterprises: Its team comes from Tsinghua University's KEG Laboratory, with expertise in both technology and compliance requirements. Government and enterprise clients value stability, security, and practical problem-solving capabilities, which Zhipu provided.
- Fast revenue growth: Revenue increased from HK$57.4 million in 2022 to HK$312 million by 2024, with an annual growth rate of 130%. In the first half of 2025, revenue reached HK$191 million, a year-on-year increase of 325%.
- Strong pricing power: After the release of GLM-5 in 2026, some services saw price increases of 30%-100%, but demand still grew by 400%, indicating that customers were willing to pay for advanced technology.
- Policy support: Domestic central and state-owned enterprises have been purchasing Chinese-made large-scale models, and Zhipu has secured contracts in valuable sectors like government and finance.
Zhipu's Challenges
Despite its success, Zhipu faces several challenges:
- Uncertain market position: Although it led in revenue in 2024, its market share was only 6.6%, indicating a highly fragmented industry where competitors could easily steal business.
- Scalability issues: Most of its revenue comes from customized deployments, which require significant manpower and are not as scalable as SaaS software.
- Valuation dependence on AGI prospects: Its market value of one trillion Hong Kong dollars is based on expectations of its ability to achieve AGI (general artificial intelligence). Any doubts about its technological leadership can lead to a sharp drop in stock price, as seen when Kimi K3 was released.
MiniMax's C-side Struggles
MiniMax's C-side strategy once seemed promising:
- Global reach: Its products, such as Hailuo AI and Talkie, targeted overseas users, with 70% of its revenue coming from overseas subscriptions, making it the only Chinese model company to profit from C-side subscriptions.
- Multimodal capabilities: It started with multimodal technology, which is more suitable for C-side interactions.
However, it now faces:
- Falling model performance: The newly released M3 model ranked low in evaluations, and JPMorgan downgraded its rating, stating that it was still in the development stage.
- User loss: Changing the subscription model to a token-based system increased user dissatisfaction. Monthly active users of its C-side products declined significantly, and its overseas business was impacted by copyright lawsuits from companies like Alibaba and ByteDance.
- Profitability issues: Despite rapid revenue growth (from $3.46 million to $79 million), increasing costs and low subscription prices limited profitability.
A Path Forward for Both Companies
Both Zhipu and MiniMax need to address the issue of rapid technological advancement and unstable business models. They must continuously develop better models to retain customers and maintain their market values. Although their approaches differ, both recognize the importance of a technology-driven strategy.
The Road Ahead
The ultimate goal is AGI (general artificial intelligence), but no one knows when it will be achieved. All large-scale model companies are investing heavily in research. While Zhipu has established a strong B-side foundation, MiniMax needs to stabilize its C-side business. The industry's common challenge is the rapid pace of technological innovation, where even leaders can quickly fall behind.
In conclusion, there is no one-size-fits-all business model for large-scale models. The path to success requires continuous technological advancement and a balance between stability and growth. The fluctuations experienced by Zhipu and MiniMax reflect the uncertainties of this rapidly evolving market.