Core Insights - The article discusses the significant advancements made by the Chinese AI computing company, Tensu Zhixin, which has recently launched a roadmap to surpass international giants like Nvidia's Hopper, Blackwell, and Rubin by 2025 to 2027 [2][4][5] - Tensu Zhixin's new architecture, the Tensu Tian Shu, has already demonstrated a performance improvement of approximately 20% over Nvidia's Hopper in key model scenarios, marking a substantial leap for domestic solutions [4][5][32] - The company aims to redefine the narrative of domestic GPU industry by moving away from a "benchmarking" approach to a self-defined leadership in AI computing [4][33] Group 1: Evolution of Computing in China - The roadmap released by Tensu Zhixin outlines a clear timeline and quantifiable breakthroughs, marking a departure from the traditional "follow and catch up" strategy adopted by many domestic companies [5][35] - The architecture is set to achieve a practical utilization rate of over 90% in executing attention mechanisms by 2025, showcasing the company's commitment to high efficiency [5][35] - The upcoming architectures, Tian Xuan and Tian Ji, are expected to further enhance performance and address industry-specific computational needs by 2026 [7][37] Group 2: Technological Innovations - Three core technological innovations underpin the aggressive roadmap: TPC BroadCast, Instruction Co-Exec, and Dynamic Warp Scheduling, which collectively enhance computational efficiency and resource utilization [10][39] - The company has adopted a problem-oriented research and development approach, addressing common industry pain points such as FP8 accumulation precision and matrix transposition overhead [11][40] - This focus on practical solutions has resulted in significant performance improvements, such as a 50% reduction in memory usage for model inference [14][44] Group 3: Redefining Value in Computing - Tensu Zhixin proposes a new value coordinate system for the computing industry, emphasizing high efficiency, predictability, and sustainability as key competitive factors [12][41] - The company aims to optimize total cost of ownership (TCO) by enhancing effective computing output per unit of power, thus reducing unnecessary costs for enterprises [14][43] - The design philosophy ensures that hardware remains adaptable to future algorithmic advancements, extending its lifecycle and value [16][44] Group 4: Market Positioning and Product Launch - The launch of the "Tongyang" series of edge computing products fills a gap in the domestic high-end edge computing market and establishes a comprehensive "cloud + edge + end" computing layout [18][46] - The product matrix covers a range of performance metrics from 100T to 300T, with specific models tailored for various applications [19][47] - Tensu Zhixin's products have already been deployed in over 300 customer environments, demonstrating their effectiveness in real-world applications [22][49] Group 5: Long-term Strategy in the GPU Industry - The competition in the GPU industry is fundamentally about building an open and collaborative ecosystem, which is essential for long-term success [52][52] - Tensu Zhixin is focused on creating a closed-loop ecosystem for domestic AI computing through hardware foundation, software adaptation, and partner collaboration [53][52] - The company aims to ensure that its products maintain value over a decade, positioning itself as a long-term player in the industry rather than merely a competitor to Nvidia [53][55]

Core Viewpoint - The GPGPU industry is transitioning from merely providing computational power to ensuring that the power is efficient, reliable, and cost-effective for real-world applications, especially in the context of AI and large models [1][3]. Group 1: Industry Trends - The demand for performance in AI has surged, with model training parameters growing from billions to trillions, necessitating a shift from simply increasing GPU numbers to addressing system engineering challenges [3]. - Data centers are evolving from hardware-centric operations to focusing on efficiency, reliability, and sustainability, with key metrics including PUE, TCO, and stability becoming critical [3][4]. - The average utilization rates for inference and training scenarios are low, highlighting inefficiencies in the current growth model of computational power [3][4]. Group 2: Company Developments - Tian Shu Zhi Xin has unveiled its fourth-generation architecture roadmap, aiming to surpass NVIDIA's Hopper architecture by 20% in performance by 2025 [6]. - The company is focusing on high-efficiency, predictable, and sustainable computing power, which is essential for long-term value [4][6]. - The introduction of innovative technologies such as TPC Broadcast, Instruction Co-Exec, and Dynamic Warp Scheduling aims to enhance performance and efficiency in their new architectures [8]. Group 3: Product Launches - The company plans to release multiple chip models, including the "Tian Gai" and "Zhi Kai" series, over the next three years, with a goal of doubling processing capabilities with each generation [9]. - The newly launched "Tong Yang" series includes various models designed for edge computing, emphasizing high performance and low latency for diverse applications [10][12]. - The Tong Yang series products have demonstrated superior performance compared to NVIDIA's AGX Orin in practical tests, showcasing their competitive edge in the market [12]. Group 4: Market Positioning - Tian Shu Zhi Xin aims to establish itself as a leader in the domestic edge computing market, focusing on high-performance, cost-effective solutions that connect AI with the physical world [12][20]. - The company has achieved significant performance improvements across various sectors, including internet AI, finance, and healthcare, with notable metrics such as a 70% increase in report generation efficiency [18]. - The firm emphasizes a comprehensive ecosystem approach, integrating software and hardware solutions to enhance user experience and performance [21][23].

智通财经记者 | 徐美慧 智通财经编辑 | 文姝琪 图片来源：天数智芯 天数智芯AI与加速计算技术负责人单天逸表示，未来3年，天数智芯将基于此次发布的四代架构，陆续发布多款产品，持续提升计算性能。 天数智芯登陆港交所上市后的第一个动作，便是官宣全栈对标英伟达。 1月26日，智通财经记者了解到，天数智芯正式对外发布四代架构路线图，其预期于2027年超越英伟达Rubin架构，随后将转向突破性计算芯片架构设计。 1月8日，天数智芯半导体股份有限公司正式在港交所主板挂牌上市，当日开盘涨31.54%。 作为通用GPU（GPGPU）产品及AI算力解决方案提供商，天数智芯成立于2015年，并于2018年正式启动通用GPU芯片设计。 此前，天数智芯战略与公共关系部副总裁余雪松在接受智通财经记者采访时表示，在AI大模型算法快速迭代的背景下，相比于专用芯片，GPGPU的灵活性 更能适应这种变化。 据弗若斯特沙利文，受AI大模型浪潮驱动，中国智能计算芯片市场正处于爆发期。预计到2029年，中国AI芯片出货量将从2024年的250万片增长至1140万 片，CAGR达32.1%。 这一增长中，GPGPU占据了重要地位。报告显示，202 ...

Core Viewpoint - TianShuZhiXin Semiconductor Co., Ltd. officially listed on the Hong Kong Stock Exchange, marking the gathering of the "Four Little Dragons" of domestic GPUs in the capital market [1][2] Group 1: Company Overview - TianShuZhiXin was founded in 2015 in Nanjing, initially focusing on computing acceleration hardware and software, later shifting to general-purpose GPU chips and AI computing solutions [2] - The company’s CEO, Gai Lujian, has 17 years of financial and investment experience and joined TianShuZhiXin in July 2020 [2] Group 2: Financial Performance - Revenue projections for TianShuZhiXin from 2022 to 2024 are 189 million, 289 million, and 540 million yuan, with a compound annual growth rate of 68.8% [2] - Net losses for the same period are projected at -553 million, -817 million, and -892 million yuan, with a net loss of -609 million yuan reported for the first half of 2025 [2][3] - R&D expenditures from 2022 to 2025 are approximately 456 million, 616 million, 773 million, and 451 million yuan for the first half of 2025, representing a significant percentage of total revenue [3] Group 3: Product Development - TianShuZhiXin has launched two product series: "TianZhai" for training and "ZhiKai" for AI inference optimization [3] - The average selling price of training GPUs decreased to 30,400 yuan, with a gross margin drop to 32%, both down over 20% year-on-year [3][4] - Upcoming products include TianZhai Gen 4 expected in Q2 2026 and TianZhai Gen 5 in Q1 2027, focusing on advanced training and high cost-performance deployment [5] Group 4: Market Position - The domestic general-purpose GPU market share increased from 8.3% in 2022 to 17.4% in 2024, with expectations to exceed 50% by 2029 [5]

Group 1 - The core viewpoint of the news highlights the rapid development of the AI computing power, large models, and GPU chip sectors within the computer and software development industry, with a focus on innovations in AI chip architecture and advancements in domestic GPUs [1] - Domestic GPU companies, such as Tianzu Zhixin and Birun Technology, are making progress with their products, which have begun small-scale shipments. The Tianpai series (training) is priced at approximately 30,000 to 40,000 yuan per unit, while the Zhi Kai series (inference) is around 10,000 yuan per unit [1] - Birun Technology is projected to have orders exceeding 1.2 billion yuan by 2025, with its next-generation BR20X chip expected to be commercialized in 2026, showcasing advanced GPGPU architecture, software platforms, and cluster technologies [1] Group 2 - In the large model sector, companies like Zhipu Huazhang (ToB) and MiniMax (ToC) are progressing through local deployment and overseas commercialization paths, having passed hearings at the Hong Kong Stock Exchange [1] - Zhipu Huazhang leverages Tsinghua University's research capabilities, with its GLM-4.7 model demonstrating strong performance in coding abilities, while MiniMax has achieved 73.1% of its revenue from overseas through efficient architecture and rapid iteration [1] - The industry is currently in a high R&D investment phase, with commercialization progress in sub-sectors such as AI servers and tax AI exceeding expectations [1]

Core Viewpoint - TianShuZhiXin Semiconductor Co., Ltd. officially listed on the Hong Kong Stock Exchange, marking the gathering of the "Four Little Dragons" of domestic GPUs in the capital market [1][2] Group 1: Company Overview - TianShuZhiXin was founded in 2015 in Nanjing, initially focusing on computing acceleration hardware and software, later shifting to general-purpose GPU chips and AI computing solutions [2] - The company’s CEO, Gai Lujian, has 17 years of financial and investment experience and joined TianShuZhiXin in July 2020 [2] Group 2: Financial Performance - Revenue projections for TianShuZhiXin from 2022 to 2024 are 189 million, 289 million, and 540 million yuan, respectively, with a compound annual growth rate of 68.8% [2] - The company reported net losses of 553 million, 817 million, and 892 million yuan for the same period, with a projected net loss of 609 million yuan for the first half of 2025 [2][3] - R&D expenditures from 2022 to 2025 are approximately 456 million, 616 million, 773 million, and 451 million yuan for the first half of 2025, representing over 200% of total revenue in the initial years [3] Group 3: Product Development - TianShuZhiXin has launched two product lines: "TianVai" for training and "ZhiKai" for AI inference optimization [3] - The average selling price of training GPUs decreased to 30,400 yuan, with a gross margin drop to 32%, both down over 20% year-on-year [3][4] - Upcoming products include TianVai Gen 4 expected in Q2 2026 and TianVai Gen 5 in Q1 2027, focusing on advanced training and high cost-performance deployment [4] Group 4: Market Position - The company’s shipment volume increased from 7,800 units in 2022 to 16,800 units in 2024, with a market share of 0.7% in training chips and 0.2% in inference chips as of the first half of 2025 [5] - The domestic general-purpose GPU market share surged from 8.3% in 2022 to 17.4% in 2024, with expectations to exceed 50% by 2029 [5]

Core Viewpoint - TianShu ZhiXin, a domestic general-purpose GPU manufacturer, officially listed on the Hong Kong Stock Exchange on January 8, 2025, raising approximately HKD 3.7 billion with an initial share price of HKD 144.6, which surged by 31.54% on the first trading day, reaching HKD 190.20 per share and a market capitalization exceeding HKD 47.55 billion [1][2]. Company Overview - TianShu ZhiXin is recognized as the first domestic general-purpose GPU company in China, achieving multiple industry milestones, including the first mass production of training and inference general-purpose GPU chips [8][12]. - The company has established a dual product line strategy, with "TianHua" targeting training and "ZhiKai" focusing on inference, aiming to connect the data center from training to deployment [12][15]. Financial Performance - Revenue for TianShu ZhiXin from 2022 to 2024 was approximately CNY 1.89 billion, CNY 2.89 billion, and CNY 5.39/5.40 billion, reflecting a compound annual growth rate of about 68.8% [17][19]. - The company reported a net loss of approximately CNY 5.54 billion, CNY 8.17 billion, CNY 8.92 billion, and CNY 6.09 billion in the first half of 2025, primarily due to high R&D expenditures [18][19]. Market Position and Competition - The general-purpose GPU market in China is characterized by an oligopolistic structure, with the top player holding a market share of approximately 91.9% [5][7]. - TianShu ZhiXin's market share is currently around 0.3%, indicating significant room for growth as the domestic GPU market is projected to grow at a compound annual growth rate of about 33% from 2025 to 2029 [5][6]. Investment and R&D Focus - Approximately 80% of the funds raised from the IPO will be allocated to R&D for general-purpose GPU chips and AI computing solutions, while 10% will be used for sales and marketing, and the remaining 10% for working capital [5][20]. - The company aims to enhance its software ecosystem and developer compatibility, particularly with CUDA, to lower migration costs for clients [21][22]. Future Outlook - The competitive landscape for general-purpose GPUs is expected to intensify, with domestic companies increasingly evaluated against specialized AI chips, emphasizing the need for efficient delivery and software toolchains [23][24]. - The ongoing "listing wave" in the domestic GPU sector is anticipated to create a more transparent competitive environment, where companies will be assessed based on their technological completeness and commercial viability [24][25].

Core Viewpoint - The strong market performance of domestic GPU companies, particularly Tensu Zhixin, reflects significant investor confidence in China's GPU industry and its potential for growth in AI computing solutions [2][4]. Group 1: Company Overview - Tensu Zhixin, known as one of the "Four Little Dragons" of domestic GPUs, officially listed on the Hong Kong Stock Exchange on January 8, with an opening price surge of over 30% [2]. - The company specializes in general-purpose GPU chips and AI computing solutions, offering products such as GPU chips, accelerator cards, and customized AI solutions for servers and clusters [4]. - Tensu Zhixin's products, including the Tianwei and Zhikai series, are designed for high performance, easy migration, and broad applicability, fully compatible with major AI ecosystems and deep learning frameworks [4]. Group 2: Financial Performance - The company has experienced rapid revenue growth, with revenues of 189 million RMB in 2022, projected to reach 540 million RMB in 2024 [5]. - Despite revenue growth, the company remains in a loss-making position, with losses of 524 million RMB in 2022 and projected losses of 892 million RMB in 2024 [5]. Group 3: IPO Details - Tensu Zhixin's IPO involved a global offering of 25.43 million shares, with a public offering in Hong Kong of 2.54 million shares at a price of 144.60 HKD per share, raising approximately 3.509 billion HKD after expenses [2]. - The IPO saw overwhelming demand, with a subscription rate of 414.24 times for the Hong Kong public offering and 10.68 times for the international offering, indicating strong investor interest [2][3].

Core Viewpoint - The company TianShu ZhiXin (09903) has garnered significant investor interest, achieving a subscription multiple of nearly 300 times and raising over 1 billion HKD during its IPO, indicating strong market confidence in its growth potential and product offerings [1][5]. Group 1: Company Overview - TianShu ZhiXin is a leader in the domestic general-purpose GPU sector, with a projected revenue compound annual growth rate (CAGR) of 68.8% from 2022 to 2024, and a continued high growth rate of 64.2% in the first half of 2025 [2]. - The company has a robust research and development (R&D) capability, with R&D expenditures from 2022 to the first half of 2025 amounting to 457 million, 616 million, 772.8 million, and 334 million CNY respectively, representing significant percentages of revenue [2][3]. Group 2: Product Development - The company has developed two main product lines: the TianYai series, which is the first domestically produced general-purpose GPU for AI model training, and the ZhiKai series, designed for inference applications, enhancing integer computation and optimizing data pathways [2][3]. - The product delivery has shown a doubling trend, with shipments increasing from 7,800 units in 2022 to 16,800 units in 2024, and 15,700 units in the first half of 2025 [3]. Group 3: Market Position and Growth Potential - The company is positioned in a high-growth industry, with the AI chip market in China expected to grow at a CAGR of 80.3% from 2022 to 2024, and the domestic GPU market share projected to rise from 8.3% in 2022 to 17.4% in 2024, potentially exceeding 50% by 2029 [3]. - The company is expected to solidify its market leadership, with a projected revenue exceeding 10 billion CNY in the next five years, and a price-to-sales (PS) ratio significantly lower than industry leaders like NVIDIA [4]. Group 4: Investment Appeal - The company’s IPO raised a net amount of 3.479 billion HKD, with 80% allocated for R&D and product commercialization, indicating a strong commitment to innovation and market expansion [4]. - The current market valuation is considered low, presenting a potential opportunity for long-term investors, especially in light of the ongoing AI development wave and domestic substitution trends [5].

该公司客户覆盖金融服务、医疗保健、交通运输、制造业及零售业等多个行业场景，客户数量在过去三 年翻超8倍，截止2025年6月，产品已在超过290名客户中完成超900次实际部署。而产品出货量从2022年 的0.78万片增至2024年的1.68万片，2025年上半年1.57万片，呈现翻倍增长趋势。 天数智芯具备高成长的行业土壤，根据弗若斯特沙利文，2022-2024年，中国AI芯片复合增长达80.3%， 预计到2029年将达到8981亿元，而国产通过GPU市场占有率持续提升，从2022年的8.3%提升至2024年 的17.4%，并预计到2029年提升至超过50%。该公司在行业中处于领导地位，按2024年按收入口径统 计，天数智芯训练型及推理型通用GPU产品均在国内同类型产品市场份额稳居前三。 智通财经APP了解到，天数智芯（09903）于1月5日完成招股，获得投资者蜂拥抢筹，根据富途数据， 认购倍数超300倍，申购额超过1000亿港元，发行价定为 144.6 港元，募资规模约37亿港元，IPO 市值 354.42亿港元。在暗盘期间，该公司股价大幅上涨，盘中涨幅一度超过49%，市场投资热度高涨。 实际上，该公司上市前 ...