Zhipu AI, a leading Chinese large model company (stock code: 02513.HK), officially listed on the Main Board of the Hong Kong Stock Exchange today, becoming the second domestic large model company to enter the capital market after MiniMax. The company publicly issued 37,419,500 shares at a price of 116.2 HKD per share. The stock opened at 120 HKD, up 3.27%, then slightly dropped to 117.2 HKD, with a market value exceeding 43 billion HKD. This IPO is expected to raise 4.3 billion HKD (approximately 550 million USD), mainly used for large model R&D, computing infrastructure construction, and commercialization ecosystem expansion.

 High Growth: Three-Year CAGR Over 130%, but Losses Continue to Widen

Zhipu AI, established in 2019, has rapidly joined China's top-tier AI companies with its GLM (General Language Model) series as the core. Financial data shows:

- Revenue grew at a compound annual growth rate of over 130% from 2022 to 2024, reflecting rapid growth in API calls, enterprise customization, and open-source ecosystems;

- Net losses continue to widen, mainly due to high-intensity R&D investment—R&D expenses accounted for over 80% of revenue in 2024;

- The company has established partnerships with leading customers in multiple industries such as finance, energy, government affairs, and the internet, and the GLM model performs well in tasks such as code generation, knowledge Q&A, and multilingual understanding.

 Technology-Driven: Open-Source and Closed-Source Dual Strategy

Zhipu AI adopts a strategy of "open-source to build ecosystem, closed-source to win business":

- In open-source: The GLM series models have been downloaded over ten million times on platforms like Hugging Face and ModelScope, becoming an important foundation for domestic large models;

- In closed-source: It launched enterprise-level models such as GLM-Edge and GLM-AllSpark, supporting private deployment and industry-specific fine-tuning;

- Toolchain completion: Provides an agent development framework, RAG engine, and low-code AI application platform, lowering the integration threshold for enterprises.

 Use of Funds: Betting on the Next Generation of Large Models and Computing Autonomy

The 4.3 billion HKD raised through this IPO will focus on:

1. R&D of the next-generation multimodal large model (GLM-V), enhancing visual-language-reasoning integration capabilities;

2. Domestic computing compatibility: Deepening cooperation with Huawei Ascend and Cambricon, building a full-stack domestic AI training and inference stack;

3. Industry solution implementation: Building benchmark cases in scenarios such as financial risk control, intelligent customer service, and scientific research assistance;

4. Global talent recruitment: Expanding the scale of Beijing, Shanghai, and Singapore R&D centers.

 AIbase Observation: Large Model Companies Enter the "Listing as a Sprint" Phase

Zhipu AI’s listing marks that Chinese large model startups are fully moving from the "technology validation period" into the "commercialization realization period." Although high R&D investment leads to short-term losses, the capital market still gives a high valuation, reflecting recognition of its technical barriers and long-term potential.

However, challenges remain severe:

- How to convert API usage into sustainable profits?

- How to maintain the domestic market and go global in the global competition with OpenAI, Anthropic, and Google?

- How to balance open-source ecosystem contributions and commercialization pace?

With MiniMax and Zhipu AI listing consecutively, 2026 will become a key turning point for the commercialization of Chinese large models. Zhipu AI's first-day performance is just the starting line of this long race.