冲刺全球国产大模型第一股，两家国内龙头港股上市在即

Group 1 - The core viewpoint of the articles highlights that MiniMax and Zhipu AI, part of the "AI Six Dragons," have received approval from the China Securities Regulatory Commission and are participating in the Hong Kong Stock Exchange listing hearing, aiming to become the "first stock of global large models" [1] - The listing process for both companies is significantly faster than market expectations, with MiniMax targeting a listing window in January 2026, marking it as one of the fastest cases since the new filing system for mainland companies going public in Hong Kong [1] - The commercialization of domestic large models has made substantial progress, with MiniMax's overseas product, talkie, achieving revenue in the tens of millions of dollars, leading the domestic market in AI companionship applications [1] Group 2 - The event signifies a transition for the domestic large model industry from the technology development phase to a capital realization phase, which is expected to reshape the competitive landscape and valuation system of the global AI sector [1] - According to CITIC Securities, the accelerated listing of domestic large model companies will create a positive cycle of "financing-research-commercialization," promoting the deep integration of multimodal technology and vertical application scenarios [1][2] - The growth logic of the large model industry has shifted from technology validation to commercial scalability, with significant profitability breakthroughs in content generation and intelligent companionship applications since 2025 [2] Group 3 - The demand for computing power in large model training and inference is exponentially increasing, benefiting infrastructure suppliers, including upstream hardware and cloud computing services, as well as downstream vertical application enterprises [2] - Electric Broad Media holds a 55% stake in Dacheng Caizhi, indirectly owning shares in Zhipu AI [3] - Capital Online is a core computing power partner of Zhipu AI, providing underlying GPU computing power cluster support [4]