记者丨彭新 编辑丨倪雨晴 10月24日， 沐曦集成电路（上海）股份有限公司（简称"沐曦"）的首次公开发行（IPO）申请将上会审议。中国GPU初创公司冲刺 资本市场再迎"关键一战"。 沐曦成立于2020年，五年来，先后推出用于智算推理的"曦思N"系列GPU、用于训推一体和通用计算的"曦云C"系列GPU。传统数据 中心GPU既要兼顾科学计算，又要满足AI训练需求，沐曦的GPU芯片针对云端AI训练和推理场景深度优化，旨在解决大规模人工智 能算力瓶颈。 进入2025年以来， 主要GPU初创企业沐曦、摩尔线程、壁仞等公司上市进程大幅加速，即将迎接更严格的考验。 创业以来，这些 公司曾吸引了前所未有的大额投资、高级人才和资源，剑指回报丰厚的GPU芯片市场，并誓言挑战英伟达在中国市场的垄断地位。 然而，在人工智能热潮带来的市场机会持续走高的情况下，市场环境却陡然发生变化。英伟达高端GPU在中国市场供应几乎"归 零"，这个被动形成的真空地带，成为国产GPU企业从未有过的历史性机遇。 IPO加速 9月26日，在各方期待下，中国GPU公司摩尔线程科创板IPO申请成功通过上交所上市委审议。从6月30日申请IPO到成功过会，仅用 了 ...

Core Viewpoint - The IPO applications of Chinese GPU companies, including MuXi Integrated Circuit and MoEr Thread, are progressing rapidly, marking a significant moment for the domestic GPU industry as they aim to challenge NVIDIA's dominance in the market [1][4][11]. Group 1: Company Developments - MuXi Integrated Circuit, established in 2020, has developed the "XiSi N" series GPUs for intelligent computing inference and the "XiYun C" series for training and general computing, targeting AI training and inference scenarios [1][5]. - MoEr Thread, also founded in 2020, has successfully passed the IPO review process in a record 88 days, potentially becoming the largest IPO on the STAR Market this year [3][4]. - Both companies have seen accelerated IPO processes amid a changing market environment, with significant investments and talent acquisition aimed at capturing the lucrative GPU market [1][5]. Group 2: Market Environment - The market for AI GPUs in China is experiencing rapid growth, with the market size projected to increase from 142.86 billion yuan in 2020 to 996.72 billion yuan by 2024, reflecting a compound annual growth rate of 62.5% [12]. - The exit of NVIDIA from the Chinese market due to U.S. export restrictions has created a unique opportunity for domestic GPU companies to fill the void [11]. - The competitive landscape is shifting, with domestic companies like MoEr Thread and MuXi aiming to establish themselves as viable alternatives to NVIDIA, which has seen its market share in China plummet from 95% to nearly 0% [11]. Group 3: Technological Challenges - Despite advancements, domestic GPU companies still face significant technological gaps compared to NVIDIA, particularly in software ecosystems and manufacturing processes [8][9]. - The transition costs for developers to switch from NVIDIA's CUDA to domestic alternatives remain a major hurdle for the adoption of domestic GPUs [8]. - The manufacturing capabilities of domestic companies are limited to 7nm or 14nm processes, while NVIDIA has moved to 4nm, posing challenges for future competitiveness [9]. Group 4: Future Outlook - The success of domestic GPU companies will depend not only on technological advancements but also on the maturation of the entire industry ecosystem [9]. - The push for domestic AI chip independence is gaining momentum, with increasing support from policies and market dynamics favoring local manufacturers [7][11]. - The rapid adaptation of domestic GPUs to emerging AI applications, such as DeepSeek, indicates a growing recognition of their capabilities in the market [12].

Core Viewpoint - The successful IPO of Moore Threads on the Sci-Tech Innovation Board highlights the strong support from the Shanghai Stock Exchange for hard technology enterprises and the emphasis on artificial intelligence and new productivity by policy makers [1][2]. Company Overview - Moore Threads, established in 2020, focuses on the research, design, and sales of full-function GPUs and related products, aiming to provide computing acceleration platforms for high-performance computing fields such as AI, digital twins, and scientific computing [1][2]. - The company is one of the few domestic GPU manufacturers that integrates both graphics rendering and AI computing, offering a comprehensive product layout from cloud to edge [1]. Financial Performance - In the first half of 2025, Moore Threads is projected to achieve a revenue of 702 million yuan, surpassing the total revenue of 608 million yuan from 2022 to 2024, which reflects a compound annual growth rate of over 200% [2]. - The revenue growth is attributed to the continuous expansion of the product matrix and sustained market demand, with the company launching a new generation of GPU architecture chips annually from 2021 to 2024 [2]. Product Development - Moore Threads has introduced four generations of chips, including "Sudi," "Chunxiao," "Quyuan," and "Pinghu," with the upcoming "Changjiang" smart SoC chip set to launch in 2024 [2]. - The performance of Moore Threads' products is approaching or has reached international advanced levels, with specific products like the MTT S80 graphics card showing single-precision floating-point performance comparable to NVIDIA's RTX 3060 [2]. Market Position and Trends - Moore Threads is currently the only domestic manufacturer capable of mass-producing full-function GPUs, with products adaptable to various application scenarios, including large model training, digital twins, consumer electronics, and cloud computing [3]. - The demand for computing power driven by AI and large model training is surging, positioning GPUs as a core computing engine and emphasizing the importance of achieving self-sufficiency in GPU technology [3]. Industry Outlook - According to Frost & Sullivan, the global GPU market is expected to reach 3.6 trillion yuan by 2029, with China's GPU market projected to grow to 1.36 trillion yuan, increasing its market share from 30.8% in 2024 to 37.8% [3]. - The AI chip market in China is forecasted to grow from 142.54 billion yuan in 2024 to 1.34 trillion yuan by 2029, with a compound annual growth rate exceeding 50%, making GPUs one of the fastest-growing segments [3][4]. Technological Advancements - As the digital economy advances, the application scenarios for GPUs are expanding beyond traditional graphics processing to include AI computing, digital twins, intelligent driving, and the metaverse, necessitating higher requirements for versatility, computing performance, and ecosystem compatibility [4]. - Companies adhering to a full-function GPU technology route are likely to achieve broader industrial application, with Moore Threads positioned as a leading player in this domain [4].

Core Viewpoint - The article highlights the rapid progress of Moer Technology's IPO process, reflecting strong policy support for AI and advanced productivity in China [2][3]. Company Overview - Moer Technology, established in 2020, focuses on the research, design, and sales of full-function GPUs and related products, aiming to provide computing acceleration platforms for high-performance computing fields such as AI and digital twins [2][3]. - The company has a diverse product layout covering cloud to edge, supporting digital transformation across various industries [3]. Financial Performance - Moer Technology is projected to achieve a revenue of 702 million yuan in the first half of 2025, surpassing the total revenue from 2022 to 2024, which was approximately 608 million yuan [3]. - The company has experienced a compound annual growth rate (CAGR) of over 200% from 2022 to 2024, with revenues of 46 million yuan, 124 million yuan, and 438 million yuan in those respective years [3]. Product Development - Moer Technology has launched four generations of GPU architecture chips from 2021 to 2024, with the latest being the intelligent SoC chip "Changjiang" [3][4]. - The products have achieved performance metrics close to or at international advanced levels, with the MTT S80 graphics card's single-precision floating-point performance nearing that of NVIDIA's RTX 3060 [4]. Market Position and Trends - Moer Technology is currently the only domestic company capable of mass-producing full-function GPUs, with products applicable in various fields such as large model training, digital twins, and cloud computing [4]. - The global GPU market is expected to reach 3.6 trillion yuan by 2029, with China's GPU market projected to grow to 1.36 trillion yuan, increasing its market share from 30.8% in 2024 to 37.8% [4][5]. - The AI chip market in China is anticipated to grow from 142.54 billion yuan in 2024 to 1.34 trillion yuan by 2029, with GPUs being one of the fastest-growing segments [5]. Strategic Importance - The demand for computing power driven by AI and large model training is surging, positioning GPUs as critical components in this landscape [4][5]. - Companies adhering to a full-function GPU technology route are expected to achieve broader industrial applications, with Moer Technology being a leading player in this domain [5].