After Zhipu AI's listing, another Chinese major model star company has entered the Hong Kong stock market. MiniMax (Shanghai Xiyu Technology) saw its stock price surge over 15% in the dark pool trading today, reaching as high as HKD 199.8, far exceeding the issue price of HKD 165, and its market value is approaching HKD 60 billion. The company's global offering of 25.3892 million shares successfully raised HKD 4.189 billion (approximately USD 536 million), with the funds to be mainly used for large model R&D, computing infrastructure construction, and global business expansion.

 Capital Enthusiasm: Domestic Large Models Enter the "Value Realization" Phase

MiniMax's strong dark pool performance reflects the capital market's high recognition of the technical strength and commercial prospects of domestic large models. As a representative of the first-tier domestic large models, MiniMax is known for its MoE (Mixture of Experts). Its open-source models are widely popular among developers on platforms such as Hugging Face and ModelScope, and its closed-source API services have covered multiple industries including gaming, social media, finance, and smart hardware.

The proceeds from this IPO will focus on:

- Next-generation multimodal large models (supporting unified understanding of text, voice, images, and videos);

- Self-developed AI chips and compatibility with domestic computing power, reducing reliance on overseas GPUs;

- Overseas developer ecosystem building, promoting model exports.

 Competing with Zhipu AI, the "Dual Power" Pattern of Domestic Large Models Begins to Emerge

At the same time, Zhipu AI (02513.HK) also completed its listing on the Hong Kong stock market this week, with an initial opening increase of 3%. Although both companies belong to the large model sector, their approaches differ slightly:

- MiniMax: More focused on high-concurrency, low-latency general AI services, strong in speech synthesis, dialogue understanding, and edge deployment;

- Zhipu AI: Focuses on knowledge-intensive scenarios (such as research, finance, and government affairs), with the GLM series showing outstanding performance in logical reasoning and code generation.

The consecutive successful IPOs of the two companies mark that Chinese large model startups are moving from the "technology validation phase" to the "commercialization acceleration phase," with significant improvement in capital market confidence.

 AIbase Observation: Challenges Behind High Valuation

Despite the bright dark pool performance, MiniMax still faces three major challenges:

1. Profit pressure: Large model R&D investment is huge, how to turn API usage into sustainable profit?

2. International competition: OpenAI, Anthropic, and Google continue to iterate, and the performance gap of closed-source models still exists;

3. Ecological construction: Although open-source models are popular, how to avoid being "exploited for free" and achieve commercial feedback?