Investment Rating - The report maintains a "Recommended" rating for the industry [5] Core Insights - The computer (Shenwan) index fell by 3.44% during the week of March 23-27, 2026, underperforming major indices due to market conditions. The Shanghai Composite Index decreased by 1.09%, the Shenzhen Component Index by 0.76%, and the ChiNext Index by 1.68% [1][10][19] - Google Research introduced TurboQuant, a vector quantization compression algorithm that reduces the key memory usage of large models to 3 bits without losing accuracy. This innovation addresses the significant memory consumption of KV cache during model inference, which expands linearly with context window growth [3][20] - The report suggests that the continuous rapid iteration of large model-related software and hardware technologies is expected to lower application costs, promoting the widespread adoption of AI applications and driving sustained growth in AI hardware demand [3][20] Summary by Sections Market Review - The Shenwan computer index experienced a decline of 3.44%, with sub-sectors such as computer equipment, IT services II, and software development showing respective declines of -3.12%, -3.14%, and -3.90% [1][10][12] Major Events - Significant announcements include: 1. Guanglian Da reported a revenue of 6.068 billion yuan for 2025, a decrease of 2.17% year-on-year, while net profit increased by 61.77% to 405 million yuan [2][17] 2. Zhiming Da's revenue grew by 61.87% to 709 million yuan, with net profit soaring by 425.27% to 102 million yuan [2][17] 3. Kingsoft Office achieved a revenue of 5.929 billion yuan, up 15.78%, and a net profit of 1.836 billion yuan, an increase of 11.63% [2][17] 4. Hengsheng Electronics reported a revenue of 5.783 billion yuan, down 12.13%, but net profit rose by 18.01% to 1.231 billion yuan [2][17] Investment Perspective - The report highlights that the computer index's significant drop is influenced by international factors, with the introduction of TurboQuant potentially impacting the storage sector's stock prices due to its focus on reducing the key cost pain point of large models [3][19][20]

Group 1 - The second Intelligent Robotics Key Technology Conference (IRCTC2026) will be held from May 15 to 17, 2026, in Hangzhou, Zhejiang Province, aiming to promote breakthroughs in key technologies and deepen the integration of industry, academia, and research [3][5]. - The conference will cover various topics including autonomous navigation, human-robot interaction, intelligent perception, machine learning, and collaborative robotics [6][5]. - The deadline for paper submissions is April 10, 2026, with a requirement for original work that has not been published elsewhere [4][7]. Group 2 - Accepted papers will be compiled into the CNKI electronic version of the conference proceedings, and selected papers may be recommended for publication in various journals, including those indexed by EI and core Chinese journals [12]. - The conference registration fees vary by participant category, with discounts available for early registration before April 30, 2026 [18]. - The conference will feature a series of events including keynote speeches, specialized forums, and poster presentations, with a detailed schedule provided [15][14].

Investment Rating - The report maintains a "Buy" recommendation for the industry, highlighting the acceleration of technological iterations and the emergence of liquid cooling as a key solution in data centers [2]. Core Insights - The rapid development of large models and AI applications has led to an explosive demand for computing power, making liquid cooling a necessary solution for data centers [5][7]. - Traditional air cooling has reached its technical limits, and liquid cooling is expected to enter a phase of rapid growth as it becomes a mandatory requirement for high-power density servers [5][7]. - The report emphasizes the importance of liquid cooling technology in meeting the increasing power and heat dissipation requirements of modern data centers, driven by advancements in chip performance [5][22]. Summary by Sections 1. Liquid Cooling as an Efficient Cooling Solution - The data center industry is experiencing robust growth, driven by AI and intelligent computing center demands [11]. - The performance of chips is rapidly improving, with liquid cooling gradually becoming the mainstream cooling solution due to the increasing power density of chips [18][22]. - Liquid cooling is positioned as a key pathway to achieving green and low-carbon goals, with stringent PUE (Power Usage Effectiveness) requirements being enforced [32][35]. 2. Power Upgrade Breaks Air Cooling Limits - Traditional air cooling faces significant thermal bottlenecks, while liquid cooling solutions are becoming more refined and effective [22]. - The cooling system accounts for approximately 15%-20% of the initial investment in data centers, making the choice of an efficient cooling system critical [18]. - The report outlines various liquid cooling technologies, including cold plate and immersion cooling, which are emerging as effective solutions [22][46]. 3. Investment Recommendations - The report identifies three companies with comprehensive solution capabilities: - **Invid**: A leading provider of precision temperature control and energy-saving solutions, with significant deliveries in the liquid cooling sector [6]. - **Shenling Environment**: A specialized supplier of integrated temperature control solutions, with a strong customer base including major internet companies [6]. - **Tongfei Co., Ltd.**: Focused on industrial temperature control technology innovation, expanding into core components for data center liquid cooling [6]. 4. AI-Driven Demand Growth - The report notes that AI applications are driving significant growth in downstream demand for data centers, with major cloud service providers increasing their capital expenditures to support this trend [48]. - The capital expenditures of leading companies like Amazon, Alphabet, and Microsoft are shifting towards AI-specific infrastructure, indicating a substantial increase in demand for data center construction [48].

Core Viewpoint - The article highlights the successful IPO of Shandong Jishijiao Technology Co., Ltd. (referred to as "Jishijiao"), an AI computer vision algorithm platform, which has seen significant growth in revenue and market presence since its inception in 2015 [4][8][24]. Company Overview - Jishijiao was founded by three university classmates, including Chen Zhenjie, who has a background in business management from Peking University and a keen interest in AI technology [7][10]. - The company focuses on providing end-to-end AI computer vision solutions across various industries, with a community of over 100,000 AI algorithm developers and a marketplace showcasing 1,517 algorithms [7][15]. Financial Performance - Jishijiao's revenue has grown from 102 million RMB in 2022 to 128 million RMB in 2023, and is projected to reach 257 million RMB in 2024, with a gross margin increase from 30.6% in 2022 to 44.9% in the first three quarters of 2025 [8][22]. - The company reported a profit of 871,000 RMB in 2024 after previous losses, indicating a turnaround in financial performance [22]. Market Position - As of 2024, Jishijiao ranks eighth in the emerging enterprise-level computer vision solutions market in China, holding a market share of 1.6% [29]. - The market for enterprise-level computer vision solutions is expected to grow from 2.2 billion RMB in 2020 to 11.1 billion RMB in 2024, with a compound annual growth rate of 49.9% [28]. Business Model - Jishijiao operates as a platform company, connecting algorithm developers with enterprise clients, which allows for efficient service delivery and cost reduction [30][31]. - The company has transitioned from a simple intermediary to a provider of standardized AI solutions, focusing on both off-the-shelf and customized offerings [18][31]. Future Directions - The company plans to enhance its AI visual language model to address more general recognition needs and assist enterprises in their AI transformation processes [33].

Core Insights - The core focus of the articles is on China Merchants Bank's digital transformation and AI strategy, emphasizing the goal of becoming the first intelligent bank in the industry by 2025 [2][3]. Group 1: Digital Transformation Strategy - China Merchants Bank has adhered to a "technology-driven" strategy for many years, with the "14th Five-Year" digital financial development plan outlining the creation of an intelligent bank as a core task for the next five years [2]. - The bank's AI first strategy was initiated following Chairman Miao Jianmin's proposal to establish the first intelligent bank in the industry, leading to active deployment of large model applications [2]. Group 2: Large Model Applications - The bank has identified 1,588 independent work items for large model applications, categorizing them into high, medium, and low-value based on quantifiable standards, prioritizing high-value scenarios for implementation [2]. - By the end of 2025, the bank aims to have implemented a total of 856 large model application scenarios [2]. Group 3: Efficiency and Performance Improvements - The average iteration cycle for large model applications has been reduced from 32 days in 2024 to 8 days in 2025, significantly enhancing efficiency [3]. - In 2025, the daily input-output token scale for large models increased by 10.1 times compared to 2024, reaching a total of 26 billion tokens, positioning the bank at a leading level in the banking industry [3]. - The cumulative replacement of manual labor hours by large model applications exceeded 15.56 million hours in 2025, equivalent to the workload of over 8,000 full-time employees [3].

Core Insights - The article highlights that domestic large models have surpassed overseas models in usage for a month, with a total call volume of 9.82 trillion tokens, a 33.4% increase from the previous week [3] - In contrast, overseas models, represented by Anthropic and Google, saw a total call volume of 2.99 trillion tokens, a decrease of 15.4% from the prior week [3] Group 1: Token Consumption and Market Dynamics - OpenRouter's data indicates that the average weekly token consumption has more than doubled compared to the fourth quarter of 2025, driven primarily by OpenClaw [5] - OpenClaw contributed to 20% of the token consumption on the OpenRouter platform during the week of March 9-15, 2026, leading to a price increase across the industry [5] - The rise in token consumption is attributed to the complexity of tasks that models are now capable of handling, which can increase token usage by tenfold or even a hundredfold compared to simpler queries [6] Group 2: Industry Trends and Future Outlook - The AI industry is entering a practical application phase, with the emergence of OpenClaw marking a shift from discussions about token value to recognizing its actual consumption [7] - The demand for tokens is expected to grow non-linearly, impacting the supply-demand balance for computing power and leading to increased costs for essential hardware like GPUs and CPUs [6] - The pricing logic in the cloud computing industry is shifting towards premium monetization, as evidenced by price hikes from companies like Alibaba and Baidu, indicating a demand-driven pricing cycle [6]

Core Insights - The core viewpoint of the article emphasizes the significant breakthrough of large models in technology, which can simulate and partially replace human capabilities [1] Group 1: Company Developments - China Merchants Bank (CMB) held its 2025 annual performance meeting in Hong Kong [1] - CMB's Chief Information Officer, Zhou Tianhong, highlighted the identification of 1,580 relatively independent work items across the bank's workforce [1] - The implementation rate of high and medium-value work items reached 69% by the end of last year, while low-value work items had a 39% implementation rate, resulting in 856 specific work items being executed [1]

Core Viewpoint - Shandong Jishijiao Technology Co., Ltd. (referred to as "Jishijiao") officially listed on the Hong Kong Stock Exchange on March 30, 2026, under the stock code "6636" [3]. Company Overview - Founded in 2015, Jishijiao focuses on providing AI computer vision solutions and large model solutions, successfully expanding into commercially viable AI large model solutions [6]. - According to Frost & Sullivan, Jishijiao ranks eighth in the emerging enterprise-level computer vision solutions market in China, with a market share of 1.6% based on 2024 revenue [6]. Financial Performance - Jishijiao's revenue for 2022, 2023, and 2024 is projected to be 102 million, 128 million, and 257 million yuan, respectively, with a compound annual growth rate (CAGR) of 59.2% [6]. - In the first three quarters of 2025, Jishijiao's revenue increased by 71.6% to 136 million yuan compared to the same period in 2024 [6]. - The gross profit margins for 2022, 2023, 2024, and the first three quarters of 2025 are 30.6%, 25.9%, 40.2%, and 44.9%, respectively [6]. - Jishijiao achieved profitability for the full year in 2024 [6]. Industry Outlook - The computer vision solutions and large model solutions industries are experiencing significant growth. The market size for emerging enterprise-level computer vision solutions in China is expected to grow from 2.2 billion yuan in 2020 to 11.1 billion yuan in 2024, with a CAGR of 49.9% [7]. - The penetration rate is projected to increase from 20.5% to 30.2% during the same period, with an expected market size of 97 billion yuan by 2029 and a CAGR of 54.3% from 2024 to 2029 [7]. - In the enterprise large model application solutions sector, the market size is expected to grow from 5.8 billion yuan in 2024 to 52.7 billion yuan by 2029, with a CAGR of 55.5% [7]. Investment Journey - Jishijiao has completed multiple rounds of financing since its establishment, with Chuxin Capital being an early investor, completing a Pre-A round investment in May 2016 [7]. - The founder of Jishijiao, Chen Zhenjie, expressed gratitude for the support from Chuxin Capital throughout the journey from startup to IPO, highlighting the importance of trust and patience from investors [9].

Summary of Conference Call Records Industry Overview - The conference call discusses the rapid growth of AI technologies and the gaming industry in China, highlighting the significant increase in token usage and the development of AI models and chips [1][2]. Key Points on AI and Tokens - China's token usage has surged over 1000 times in two years, with an average daily token call volume exceeding 140 trillion as of March [2]. - Major AI models are being released domestically, focusing on cost-effectiveness and infrastructure advantages [2]. - Anthropic's AI Agent capabilities are intensifying, with new features allowing direct interaction with user devices [1]. Company Highlights - Kunlun Wanwei launched several products including Matrix-Game 3.0 and SkyReels V4, targeting creators with the Skywork Super Agent [2]. - Kuaishou and Xiaomi are making significant investments in AI, with Xiaomi planning to invest over 60 billion yuan in the AI sector over the next three years [2]. - Kingsoft Cloud reported a 95% year-on-year increase in AI cloud revenue, contributing to a 34.9% growth in public cloud revenue [2]. Chip Development - Alibaba Cloud's XuanTie ecosystem introduced the flagship CPU C950, supporting large AI models and forming a comprehensive AI chip system [3]. - The integration of GPU, CPU, and AI inference chips is expected to enhance cloud product capabilities and profitability [3]. Market Sentiment and Recommendations - The market sentiment towards AI's impact on software, media, and the internet is overly pessimistic, with low PE ratios in gaming and internet sectors indicating potential investment opportunities [4]. - Recommendations include companies like Kingsoft Cloud, Tencent, Alibaba, and Baidu, with a focus on gaming companies such as Giant Network and 37 Interactive Entertainment [4]. - The potential for AI to enhance rather than disrupt the gaming industry is emphasized, with AI expected to create demand through improved workflows and digital asset management [4]. Additional Insights - The gaming sector is seeing new game tests and releases, with expectations for new titles to exceed low growth forecasts for mobile games [4]. - AI's role in enhancing platform value and commercial monetization is highlighted, particularly for high-engagement platforms [4].

Investment Rating - The industry investment rating is "Positive" (maintained) [1] Core Insights - The report highlights the continuous investment in AI by major companies, with ByteDance's Doubao model achieving a daily token call volume exceeding 100 trillion, marking a growth of over 60% in less than two months [3][28] - The gaming sector is expected to see a resurgence with the upcoming launch of highly anticipated titles like "Yihuan," which has already surpassed 30 million global pre-registrations [4][10] - The report suggests a strong potential for domestic models in video generation, music, and coding, which will drive demand for token and cloud computing services [3] Industry Data Overview - "Luokewang Kingdom: World" ranked first in the iOS free game chart in mainland China, while "Peace Elite" topped the iOS revenue chart [10][14] - The film "Rescue Plan" achieved the highest box office for the week, grossing 0.81 billion [22] - The daily consumption of AIGC (AI-Generated Content) has seen significant growth, with ByteDance's Doubao model leading the way [27] Industry News Summary - The report notes that the AIGC sector is experiencing rapid advancements, with the daily token call volume for ByteDance's Doubao model surpassing 100 trillion [27][28] - OpenAI has officially shut down its AI video generation tool Sora, indicating a strategic shift towards productivity and programming tools [27] - The upcoming launch of OpenAI's next flagship model, Spud, is anticipated within weeks, reflecting ongoing innovation in the AI space [27]