流动性方面，科创信息技术ETF盘中换手24.21%，成交8538.25万元，市场交投活跃。拉长时间看，截 至9月29日，科创信息技术ETF近1周日均成交9852.21万元，居可比基金第一。 截至9月29日，科创信息技术ETF近3年净值上涨127.24%，指数股票型基金排名44/1879，居于前 2.34%。从收益能力看，截至2025年9月29日，科创信息技术ETF自成立以来，最高单月回报为32.25%， 上涨月份平均收益率为9.53%。 消息面上，在2025世界人工智能大会(WAIC)上,国产GPU企业摩尔线程创始人兼CEO张建中首次提出 的"AI工厂"理念。以"AI工厂"理念为核心蓝图,将芯片研发、集群搭建与软件生态构建的全栈能力深度 整合,持续推动国产算力基础设施向AGI时代所需的规模化、高效率、高可靠模型训练"超级工厂"升级。 有券商表示，算力产业链的高景气度已经确定，生成式AI的竞争已转向算力基础设施的军备竞赛，巨 头们正通过前所未有的资本投入，争夺有限的电力、土地和芯片资源，以奠定未来AI时代的竞争优 势。AI巨头的天价投资和长期规划，为整个生态提供了清晰的需求预期，这意味着算力供应链的全链 条受益 ...

Core Viewpoint - Nvidia has established itself as a cornerstone of the AI era, with its investments in startups indicating its ambition to build a vast ecosystem over the next decade [2][22]. Investment Strategy - Since 2023, Nvidia has significantly increased its investment frequency, rising from approximately 20 investments in 2022 to around 50 by the end of 2023, maintaining a pace of 50-60 investments annually thereafter [3][4]. - Nvidia's investments span various stages of company development, from seed rounds to D, E, and F rounds, as well as acquisitions [3]. Focus Areas - The majority of Nvidia's investments are concentrated on the AI industry chain, covering AI computing power, large models, and AI applications, primarily within the United States, with occasional investments in Europe and Israel [4][16]. - Nvidia's investment strategy is not solely focused on financial returns but aims to strengthen its ecosystem, with a clear preference for companies that utilize its technology and products [9][12]. Investment Entities - Nvidia's primary investment activities are conducted through its Corporate Development Department, led by Vishal Bhagwati, and NVenture, led by Sid Siddeek, each with distinct investment philosophies [8][10]. - The Corporate Development Department has significantly increased its investment frequency, averaging around 40 investments annually from 2023 to 2025, nearly tripling its previous rate [9]. - NVenture, established in 2021, has also accelerated its investment pace, from approximately 14 investments in 2023 to 20 in 2024 [12]. Unicorns and Performance - Nvidia has successfully invested in 20 unicorns, with its Corporate Development Department outperforming NVenture in terms of post-investment valuations [16][19]. - Notable investments include You.com, Reka AI, and Weka.io, which have all achieved unicorn status and rely on Nvidia's GPU technology [17][18][21]. Future Ecosystem Development - Nvidia's investments are evolving to encompass not only AI models and infrastructure but also energy and embodied intelligence sectors, aiming to create a unified AI infrastructure for the next 5-10 years [26][28]. - The concept of the AI Factory, introduced by Nvidia, aims to integrate AI development with industrial processes, covering the entire AI workflow from data collection to large-scale inference [30]. Financial Growth - Nvidia's long-term equity investments have seen substantial growth, with the value increasing from $1.3 billion in fiscal year 2024 to $3.4 billion in fiscal year 2025, reflecting a nearly threefold increase in just one year [31].

Core Insights - The future of computing power competition is shifting from individual data centers to cross-regional computing networks [2][5] - NVIDIA's new Spectrum-XGS Ethernet technology integrates geographically dispersed data centers into a unified AI super factory to address the physical limits of single data centers [2][3] - The Scale-Across model emerges as a third pillar of AI computing, allowing integration of data centers across different cities, countries, and even continents [3][5] Industry Trends - The demand for AI computing power is pushing traditional data centers to evolve into AI data centers (AIDC) due to the exponential growth of AI technology and AIGC [5] - The integration of global resources through Spectrum-XGS can significantly reduce training time and enhance research efficiency for large models [4][5] - The need for high-speed, low-latency optical communication infrastructure is expected to rise, particularly for 1.6T/3.2T optical modules and hollow-core fibers [6][7] Market Impact - Companies involved in optical fiber and optical module production are experiencing increased attention and stock performance, with notable gains in companies like Yangtze Optical Fibre and Cable [2] - The demand for hollow-core fibers is anticipated to grow rapidly due to their advantages in low latency and high capacity, making them suitable for long-distance data transmission [7] - The transition to GW-level AI supercomputing centers is likely to drive growth in high-end PCB manufacturing and liquid cooling technologies [6][7]

Group 1 - NVIDIA's CEO Jensen Huang praised TSMC as a great company that will continue to grow at an astonishing speed in the AI era, indicating a new industry called "AI factories" will emerge in Taiwan, presenting significant opportunities for the region [1][2] - Huang announced that the Blackwell Ultra GB300 has entered full production with successful output increases, and TSMC along with NVIDIA's ecosystem partners, including Foxconn, Quanta, Wistron, and ASUS, are performing exceptionally well in this regard [2] - NVIDIA is the global leader in AI chips, and Huang mentioned the upcoming advanced Rubin platform with six product designs already ordered from TSMC, including CPU, GPU, NVLINK switch chips, and optical switch chips [2] Group 2 - Huang expressed excitement for more factories in Taiwan, noting that NVIDIA has already begun its first factory with Foxconn and hopes to establish more [3] - Huang highlighted the potential for U.S. government initiatives to support chip manufacturing, suggesting that TSMC could also benefit from such measures, and he regards TSMC as one of the greatest companies in human history and a smart investment target [3]

"这次WAIC大会，基本上国内所有AI芯片、服务器厂商都开始拥抱超节点。我们在去年提倡超节点的 时候，大家还觉得很陌生，今年对这个趋势已没有任何疑问。"上海曦智科技有限公司（以下简称"曦智 科技"）创始人、首席执行官沈亦晨博士表示。 记者还注意到，算力基础设施每年都在竞速创新，摩尔线程、沐曦科技、燧原科技等厂商都展出了最新 的产品和解决方案。而且国产厂商也不再"比参数"，还出现了罕见的合作。比如四位国产AI芯片（沐曦 科技、天数智芯、燧原科技和壁仞科技）高管首度同台，出现在大模型独角兽——阶跃星辰的圆桌论坛 上。 "不做超节点就掉队！"回顾2025年世界人工智能大会（WAIC）的精彩时刻，"超节点"绝对有一席之 地。 其中最受瞩目的，当属华为首次展出的昇腾384超节点真机，即Atlas 900 A3 SuperPoD，其集成了384颗 昇腾NPU和192颗鲲鹏CPU，通过全新高速网络MatrixLink全对等互联，这超节点就像一台超级"AI服务 器"，算力规模300 PFLOPs，接近英伟达GB200 NVL72系统的2倍。 《中国经营报》记者在现场看到，在昇腾384超节点面前，不缺少前来打卡的人流以及向工作 ...

Investment Rating - The report maintains an investment rating of "Outperform" for the electronics sector [5]. Core Insights - Major companies like Meta, Microsoft, and Apple are increasing capital expenditures to enhance their AI capabilities and overall business performance [1][2][3]. - The semiconductor industry is expected to rebound significantly, with a projected compound annual growth rate of 7.54% from 2025 to 2034 [15]. - The electronics sector has shown strong performance, with a 1.22% increase in the index over the past week, ranking 4th among 31 sectors [29][32]. Summary by Sections Company Performance - Meta reported Q2 revenue of $47.5 billion and plans to increase its annual capital expenditure to between $66 billion and $72 billion, focusing on AI and smart glasses [1]. - Microsoft achieved Q4 revenue of $76.44 billion, a year-on-year increase of 18%, with a notable 39% growth in Azure revenue [2]. - Apple’s Q3 revenue reached $94.04 billion, with a 13% increase in iPhone sales and a 10% overall growth [3]. Industry Trends - The semiconductor market is projected to grow from $627.76 billion in 2025 to approximately $1,207.51 billion by 2034, indicating strong growth potential [15]. - The collaboration between Innosilicon and NVIDIA aims to promote the 800 VDC power architecture in AI data centers, enhancing efficiency and reliability [4][9]. Market Performance - The electronics sector's PE ratio stands at 56.63, with a 10-year percentile of 67.15%, indicating a relatively high valuation compared to historical averages [37][40]. - The semiconductor sub-sector has a PE ratio of 87.13, while consumer electronics stands at 30.44, reflecting varying growth expectations across segments [40]. Investment Recommendations - Key companies to watch in the computing power supply chain include Shenghong Technology, Huadian Technology, and Industrial Fulian, among others [11]. - For the storage industry, focus on companies like Zhaoyi Innovation and Baiwei Storage, while in the consumer electronics sector, companies like Luxshare Precision and Xiaomi Group are recommended [11].

Core Viewpoint - The rapid development of artificial intelligence (AI) has made AI chips a global discussion hotspot, with NVIDIA dominating the market due to its GPU advantage, leading to record-high performance and market capitalization. AMD's CEO predicts that the market for AI and large computing system accelerators will exceed $500 billion in a few years [1] Group 1: Full-Function GPU Development - The evolution of computing power is closely tied to the development of full-function GPUs, which have transitioned from single-task graphics cards to versatile processors that support various applications, including AI [2] - Full-function GPUs have four core engines: AI computing acceleration, modern 3D graphics rendering, physical simulation and scientific computing, and ultra-high-definition video encoding and decoding [3] Group 2: Moore Threads' Innovations - Moore Threads, established in 2020, has developed a complete computing acceleration system, launching four generations of GPU architectures and intelligent SoC products, covering AI intelligence, professional graphics acceleration, and desktop graphics acceleration [5] - The company aims to build an "AI factory" to enhance the efficiency of advanced model production, addressing the bottlenecks in large model training for the AGI era [6] Group 3: AI Factory Efficiency - The efficiency of the "AI factory" is determined by five core elements: generality of accelerated computing, effective chip computing power, single-node efficiency, cluster efficiency, and cluster stability [7] - Moore Threads emphasizes the importance of full-function GPUs and full precision in achieving high efficiency in AI model training [9] Group 4: Technical Breakthroughs - The self-developed MUSA architecture allows for significant improvements in resource utilization and reduces the development cost of new chips, achieving a 30% performance increase in Transformer computing [11] - Innovations in memory systems and communication have led to a 50% bandwidth saving and a 60% reduction in latency, enhancing the effective computing power of single chips [12] Group 5: Cluster Solutions - The "KUA" cluster, based on full-function GPUs, aims to provide a comprehensive system-level solution for large-scale GPU computing, supporting over 1,000 computing nodes with ultra-low communication latency [17] - The KUA cluster incorporates advanced technologies to enhance training efficiency and stability, achieving over 99% effective training time [19] Group 6: Industry Applications - Moore Threads' full-function GPUs are driving innovations across various sectors, including physical simulation, AIGC, scientific computing, and intelligent manufacturing, with a vision to empower developers and serve multiple industries [21][25]

Core Insights - Huang Renxun emphasizes the importance of AI as the greatest "technological equalizer," suggesting that in the future, everyone will be a programmer, artist, or writer [21][22][23] - The allocation of the scarce H100 chips is based on a simple principle: first come, first served, with a smooth process for partners to plan ahead [28][25] - Huang Renxun takes pride in personally reviewing employee compensation and claims to have created more billionaires among executives than any other CEO [6][8][45] Group 1 - Huang Renxun revealed that NVIDIA has scrapped $50 billion worth of graphics cards, indicating the high demand for chips from tech giants like Zuckerberg and Musk [4][26] - The company is fully embracing AI across all levels, with employees being liberated from mundane tasks to pursue greater creativity, ultimately leading to growth and job creation [20][18] - Huang Renxun believes that the future will require AI as a co-pilot for programmers, making traditional coding methods obsolete [24][21] Group 2 - The H100 chip's value remains high, with a residual value of 75-80% after one year, thanks to the open CUDA platform that enhances performance [33][34] - Huang Renxun agrees with Musk's insight that the future will require 50 million H100-level computing chips, marking the beginning of a multi-trillion-dollar infrastructure wave [35][37] - The emergence of efficient open-source models like DeepSeek from China is seen as a victory for the U.S. tech stack, reinforcing its global standard [40][41] Group 3 - Huang Renxun acknowledges the significant compensation for top AI researchers, asserting that it is reasonable given the value they create [8][44] - He confirms his deep involvement in employee compensation, using machine learning to assist in the process, and states that he always increases salary expenditures [5][47] - The trend of small, elite teams driving innovation is highlighted, with companies like OpenAI and DeepSeek operating with around 150 top talents [9][46]

Core Insights - A significant market battle worth billions is unfolding, highlighted by a recently surfaced photo of Nvidia's CEO Jensen Huang and Xiaomi's CEO Lei Jun, marking their first public appearance together in 12 years [1][3] Group 1: Nvidia's Strategic Moves in China - Jensen Huang's frequent visits to China in 2025, including three trips to major cities, indicate Nvidia's focus on penetrating the Chinese market, especially after facing a $13.5 billion revenue loss due to U.S. export restrictions [4][5] - Nvidia's approval to export the H20 chip to China is a crucial development, allowing the company to resume sales in a market that contributes $17.1 billion annually to its revenue [4] - The introduction of the RTX Pro GPU, designed for AI applications, aligns with U.S. export regulations, showcasing Nvidia's adaptability in the face of geopolitical challenges [5] Group 2: Transition to AI Infrastructure - Nvidia is transitioning from a hardware supplier to an AI infrastructure provider, as evidenced by the announcement of the NVLinkFusion architecture, which supports third-party CPU and AI accelerator integration [7] - This technology offers a bandwidth of 900GB/s, significantly surpassing traditional protocols, positioning Nvidia as a key player in the evolving AI landscape [7] - Huang's statement that "China has sufficient computing power" reflects a strategic shift towards collaboration and ecosystem building rather than maintaining a monopoly [7] Group 3: AI Factories and Robotics - Nvidia's strategy in China includes establishing "AI factories," which represent a shift from traditional data centers to AI-driven operations, aiming to create value through continuous data generation [9][10] - The potential of humanoid robots as a trillion-dollar industry is highlighted, with China serving as a critical commercialization hub due to its lower manufacturing costs and technological advantages [11] - Nvidia's collaboration with local companies like Xiaomi is essential for integrating AI capabilities into various sectors, leveraging China's manufacturing strengths [13] Group 4: Strategic Partnership Dynamics - The renewed partnership between Huang and Lei signifies a deeper strategic alignment, as both companies have evolved from hardware manufacturers to ecosystem builders [17] - The mutual need for collaboration arises from Nvidia's requirement for local partners to maintain influence amid U.S. restrictions and Xiaomi's need for advanced computing power to enhance its automotive technology [18] - The partnership is seen as a pragmatic approach to balancing political risks and commercial interests, with both companies benefiting from shared technological advancements [18]

Core Viewpoint - A significant market battle worth billions is unfolding, highlighted by a recent meeting between Nvidia's CEO Jensen Huang and Xiaomi's CEO Lei Jun, marking a notable shift in the tech landscape over the past 12 years [1][4]. Group 1: Nvidia's Strategy in China - Nvidia's frequent visits to China in 2025 indicate a strategic focus on the Chinese market, especially after facing a $13.5 billion revenue loss due to U.S. export restrictions [5][4]. - The approval of H20 chip exports to China is a crucial development for Nvidia, allowing the company to resume sales and mitigate losses [5][4]. - Nvidia's new RTX Pro GPU is designed for AI applications and complies with U.S. export regulations, showcasing the company's adaptability [5][4]. Group 2: Transition from Hardware to AI Infrastructure - Nvidia is evolving from a hardware supplier to a provider of AI infrastructure, as evidenced by the introduction of the NVLink Fusion architecture, which enhances system design flexibility for cloud service providers [7][4]. - Huang's statement that "China has sufficient computing power" reflects Nvidia's shift towards becoming an ecosystem builder rather than a technology monopolist [7][4]. Group 3: AI Factories and Robotics - Nvidia's strategy includes establishing AI factories in China, which are expected to redefine data centers by focusing on AI computation rather than traditional data storage [11][4]. - The Chinese manufacturing sector, which accounts for about 30% of global manufacturing value, presents a significant opportunity for Nvidia's AI factory strategy [12][4]. Group 4: Humanoid Robots as a Future Industry - Huang identifies humanoid robots as a potential trillion-dollar industry, with China playing a critical role in commercialization due to lower manufacturing costs and strong supply chain capabilities [14][4]. - The Chinese government's support for humanoid robots as a disruptive technology further enhances the business environment for Nvidia [16][4]. Group 5: Strategic Partnership Between Nvidia and Xiaomi - The historical relationship between Nvidia and Xiaomi, marked by mutual respect and understanding, lays a foundation for future collaboration, especially in light of current geopolitical challenges [22][4]. - Both companies have transformed from hardware manufacturers to ecosystem builders, creating a complementary relationship that benefits both parties [22][4]. - Nvidia's collaboration with Xiaomi is seen as a pragmatic approach to balance political risks and commercial interests in the evolving tech landscape [22][4].