Core Insights - A significant trend is emerging as leading domestic GPU companies are rapidly pursuing IPOs, with notable examples including Moer Thread and Muxi Co., which have set records for the speed of their listing processes [1][3][10] - This capital frenzy is driven by the expectation for self-controlled computing power and the exit demands of numerous VCPEs during their fund lifecycle [2] Group 1: Moer Thread and Muxi Co. - Moer Thread's IPO process is remarkable, with its application accepted on June 30, 2025, and approval granted on September 26, 2025, taking only 88 days. The IPO aims to raise 8 billion yuan [3][7] - Founded in 2020 by Zhang Jianzhong, a former NVIDIA executive, Moer Thread focuses on full-function GPUs and has developed the MUSA architecture to support AI computing and graphics rendering on a single chip [3][4] - Moer Thread's revenue figures show a growth trajectory, with revenues of 46.09 million yuan in 2022, projected to reach 438 million yuan in 2024, and 702 million yuan in the first half of 2025, marking a compound annual growth rate of 208% [7][9] - Muxi Co. followed closely, with its IPO approved on October 24, 2025, after 116 days of review, aiming to raise 3.904 billion yuan for the development of next-generation GPU chips [10][12] - Muxi, also founded in 2020, focuses on high-performance general-purpose GPUs and has established a complete domestic supply chain for design, manufacturing, and testing [10][12] Group 2: Other Companies and Market Dynamics - Suiruan Technology is taking a more cautious approach to its IPO, currently in the counseling phase, with a complex shareholder structure potentially affecting its progress [13][17] - Suiruan's products include the domestic cloud training chip "Suis 1.0" and the fourth-generation training and inference chip "Suiruan L600," which features significant advancements in computing power and storage [14][15] - Wallran Technology's IPO path remains uncertain, with reports suggesting it has submitted a confidential listing application to the Hong Kong Stock Exchange [19][24] - Tian Shu Zhixin is also rumored to be pursuing a listing, focusing on GPGPU products that have already achieved commercial success [25][27] Group 3: Industry Trends and Challenges - The market for AI computing power is expanding rapidly, with national strategies emphasizing self-sufficiency in computing capabilities, and the Sci-Tech Innovation Board providing a pathway for unprofitable but technologically critical companies to go public [28][30] - Companies are seizing the historic opportunity for domestic substitution, particularly in light of supply constraints from NVIDIA, creating a significant market gap [29][30] - The collective IPO rush signals a broader industry trend where national interests and capital markets are aligning to promote high-end computing autonomy [30][31] - The limited capacity of the capital market means that the first companies to successfully list will gain a competitive financing advantage, potentially leading to a "Matthew effect" in the industry [31][32]

Core Viewpoint - The domestic GPU manufacturer Tian Shu Zhi Xin Semiconductor Co., Ltd. is considering an initial public offering (IPO) in Hong Kong, potentially raising between $300 million to $400 million [1][2]. Company Overview - Tian Shu Zhi Xin, established in 2015, is a provider of general-purpose GPU chips and computing systems in China, focusing on developing autonomous and internationally leading general-purpose GPU products [5]. - The company's GPU products include the Tian Yai and Zhi Kai series, which are designed for high efficiency, easy migration, and broad compatibility with major AI ecosystems and deep learning frameworks [6]. Product Development - In December 2020, Tian Shu Zhi Xin successfully launched the "Tian Yai 100 chip," marking the first domestically developed cloud training chip based on a fully self-researched GPU architecture [6]. - In May 2022, the company introduced the Zhi Kai 100 inference chip, becoming the only domestic company to offer a complete solution for both cloud training and inference under GPU architecture [6]. - In November 2023, the Tian Yai 150 training chip supported the Zhiyuan Research Institute in completing heterogeneous mixed training of a large model, establishing the company as the first in China with heterogeneous computing power mixed training capabilities [6]. Market Position and Achievements - Tian Shu Zhi Xin's GPU products have assisted over 100 clients in training more than 500 models across over 20 industries [6]. - The company holds over 270 authorized intellectual property rights and has applied for more than 170 invention patents, receiving numerous accolades such as Shanghai's specialized and innovative enterprises and high-tech enterprises [6]. Financing History - The company has completed multiple rounds of financing, including a multi-hundred million yuan B round in 2019 led by Centurium Capital and Princeville Capital, and a 1.2 billion yuan C round in 2021 led by Yuanbo Capital and Centurium Capital [7]. - In the C+ and C++ rounds in 2022, investors included Financial Street Capital and HOPU Investment, raising over 1 billion yuan [9]. - Hainan Centurium Investment Partnership is the largest shareholder, holding 9.43% of the company, followed by Nanjing Youxun Equity Investment Partnership and Shanghai Xishi Enterprise Management Consulting Partnership with 8.52% and 7.13% respectively [9].