Core Insights - Google has significantly upgraded its Gemini 3 AI model's Deep Think mode, focusing on addressing complex challenges in modern scientific research and engineering, sparking a new wave of "Gemini AI frenzy" globally [1][7] - The new Deep Think mode is now available to Google AI Ultra subscribers and is the first time Google has provided this functionality through the Gemini API to select researchers, engineers, and large enterprises [1][8] Performance Metrics - The updated Deep Think model achieved a score of 48.4% on Humanity's Last Exam (HLE), setting a new industry standard for contemporary AI models [5][7] - It scored 84.6% on the ARC-AGI-2 reasoning task benchmark, verified by the ARC Prize Foundation, and obtained a 3455 Elo rating on the Codeforces competitive programming platform [5][6] - The model also demonstrated gold medal-level performance in the written sections of the 2025 International Physics and Chemistry Olympiads, achieving 50.5% on the CMT-Benchmark [4][6] Application and Functionality - Deep Think is designed to facilitate practical applications, enabling researchers to interpret complex data and engineers to model intricate physical systems through coding [2][7] - The model's capabilities extend beyond mathematics and programming, now encompassing interdisciplinary research problems requiring a combination of physical intuition, structured chemical inference, mathematical formalization, and coding solutions [4][6] - The upgrade emphasizes structured reasoning and scalable inference capabilities, allowing for iterative exploration of multiple hypothesis spaces and continuous refinement through a "generate-validate-revise" loop [8] Market Positioning - The updated Deep Think model positions Google in direct competition with other AI products like OpenAI's ChatGPT and Anthropic's Claude, marking a shift from abstract reasoning to practical applications in large-scale research and engineering workflows [7][8] - By defining Deep Think as a specialized reasoning mode for scientific, research, and engineering challenges, Google aims to attract attention from developers and institutions through its impressive performance metrics and clear application scenarios [7][8]

Core Viewpoint - The article discusses the rapid development and funding of a 3D AI chip company, 算苗科技 (Suanmiao Technology), which has completed two rounds of financing totaling nearly 1 billion RMB, aimed at developing domestically produced 3D computing chips for AI applications [3][10]. Group 1: Company Overview - 算苗科技 focuses on the research and development of 3D computing chips, with its core product being a customized chip for AI model inference [4]. - The company aims to address the "memory wall" issue that limits AI model computation, as current AI chips face significant inefficiencies due to memory bandwidth constraints [4][5]. - 算苗科技's A4 chip has demonstrated a throughput of 1.26 to 2.19 times that of NVIDIA's H200 in inference tasks on major open-source models [5]. Group 2: Funding and Market Position - The recent funding rounds were led by prominent investors, including Source Code Capital and Shixi Capital, indicating strong market interest and support for the company's vision [3][10]. - The company is positioned to leverage its expertise in 3D IC technology to create a competitive edge in the AI chip market, which is expected to grow significantly [10][19]. Group 3: Technological Innovation - 算苗科技 utilizes a 3D stacked architecture that allows for significantly higher memory bandwidth (up to 32 TB/s), which is crucial for AI model inference [4][13]. - The company’s approach contrasts with traditional GPU architectures, focusing on specialized ASIC designs that optimize performance for specific tasks rather than general-purpose computing [14][15]. Group 4: Strategic Focus - The company has chosen to concentrate on AI model inference rather than training, as it anticipates that 90% of future AI computing demand will be for inference tasks [15][18]. - 算苗科技 believes that the future of AI computing lies in architectural innovation, particularly through 3D stacking and ASIC optimization, which aligns with the growing demand for efficient computing solutions [28][29].

Group 1 - The core point of the article highlights the transition from generative AI to agent-based AI, with companies like Tencent and Alibaba engaging in a "red envelope war" to attract users, while computing power rental firms emerge as significant beneficiaries [1][2] - The demand for AI infrastructure has surged due to the ongoing competition among major tech firms, leading to price increases from cloud service providers like UCloud, which announced a price hike for all products starting March 1, 2026, citing rising hardware costs and increased demand [2][3] - Amazon Web Services (AWS) also raised prices for its machine learning capacity blocks by approximately 15%, marking a significant shift in its pricing strategy after 20 years of price reductions, driven by high demand for GPU resources [2][3] Group 2 - The explosion in demand for inference capabilities, particularly driven by the rise of agents, has led to a significant increase in computing power requirements, with agents causing a tenfold to fiftyfold increase in token consumption compared to standard chat applications [4][5] - The shift in computing demand from a training-only model to a dual-driven model of training and inference is noted, with inference increasingly dominating the demand landscape [5] - The article discusses the potential for increased deployment of edge and endpoint AI inference, suggesting a hybrid architecture that balances real-time requirements and privacy concerns with the need for powerful cloud-based models [5]

Core Viewpoint - Aixin Yuan Zhi Semiconductor Co., Ltd. has officially listed on the Hong Kong Stock Exchange, raising a total of HKD 2.961 billion with an issue price of HKD 28.20 per share, focusing on edge computing AI chips and expanding into growth markets like smart vehicles [1] Group 1: Company Overview - Aixin Yuan Zhi was established in May 2019, primarily engaged in the sale of semiconductor products and related technical services [2] - The company focuses on three main areas: terminal computing, smart vehicles, and edge AI inference, with applications in smart cities, intelligent transportation, smart homes, smart manufacturing, and smart consumer products [2] Group 2: Market Position and Product Development - Aixin Yuan Zhi ranks among the top five in the global visual edge AI inference chip market by shipment volume, with an expected shipment of 9.3 million AI inference SoCs in 2024 [2] - The company operates a fabless model, concentrating on chip design and sales, and aims to expand production scale in visual edge AI inference SoCs [2] Group 3: Financial Performance - The revenue from terminal computing products is a significant part of Aixin Yuan Zhi's total revenue, contributing HKD 235 million (87.2% of total revenue) in the first three quarters of 2025 [3] - The company reported revenues of HKD 50.23 million, HKD 230 million, HKD 473 million, and HKD 269 million for the years 2022, 2023, 2024, and the first three quarters of 2025, respectively [5] - The adjusted net losses for the same periods were HKD 444 million, HKD 542 million, HKD 628 million, and HKD 462 million [5] Group 4: Strategic Acquisitions - Aixin Yuan Zhi's revenue growth in the first three quarters of 2025 was driven by the successful acquisition of Zhejiang Huatu Microchip Technology Co., Ltd., expansion of terminal computing products, and rapid sales growth of smart vehicle SoCs and edge AI inference products [5] - The company acquired control of Zhejiang Huatu in October 2023, holding 99.04% of its shares, which has allowed for an expanded product portfolio and enhanced operational efficiency [5] Group 5: Research and Development Investment - As a technology-driven company, Aixin Yuan Zhi is in a capital investment phase, with significant R&D expenditures of HKD 446 million, HKD 515 million, HKD 589 million, and HKD 414 million for the years 2022, 2023, 2024, and the first three quarters of 2025, respectively [6] - The funds raised from the IPO will be used to optimize existing technology platforms, invest in R&D projects, expand sales, and for general corporate purposes [6]

Group 1 - In 2026, South Korean investors significantly increased their investments in Hong Kong stocks, shifting their focus from traditional sectors to hard tech fields such as AI models and semiconductors [2] - In 2025, Xiaomi Group was the most favored stock among South Korean investors with a net purchase amount of $87.75 million, while hard tech stocks were less represented in the top ten [2] - By February 10, 2026, AI model leader MiniMax-WP topped the buying list, indicating a clear trend towards hard tech investments [2] Group 2 - The overall enthusiasm of South Korean retail investors for overseas stock investments has reached historical highs, with trading commissions from the top 12 securities firms amounting to 1.95 trillion KRW (approximately $1.3 billion) in 2025, more than doubling from the previous year [3] - Foreign exchange fee income related to retail overseas investment also saw a significant increase, rising from 147.7 billion KRW in 2024 to 452.6 billion KRW in 2025, reflecting a growing demand for overseas quality assets [3] - Hard tech stocks in Chinese Hong Kong markets are becoming a key choice for South Korean investors in their overseas allocations [3]

Core Viewpoint - The company Tongda Hai (301378) anticipates a net loss of between 90 million to 135 million yuan for the year 2025, primarily due to reduced investment in information technology by court clients, project delays, and increased R&D costs [1][2]. Financial Performance - The company has issued a performance forecast indicating a negative net profit for the year 2025, with losses expected to range from 90 million to 135 million yuan. The main factors contributing to this forecast include a decrease in IT investment from court clients, project delays, and rising R&D expenses. The company commits to adhering to regulatory requirements and will promptly disclose the formal annual report [2]. Business Development - In the announcement regarding stock price fluctuations, the company noted that its AI large model applications and services are still in the early stages, with significant uncertainty surrounding large-scale commercialization. The revenue driven by AI remains relatively low, and future attention should be directed towards the progress of commercialization in this area [3]. Stock Performance - From January 9 to January 13, 2026, the company's stock was identified as experiencing abnormal fluctuations due to a cumulative deviation of over 30% in three consecutive trading days. On February 2, 2026, there was a net inflow of 55.1284 million yuan in principal funds, with the stock price increasing by 12.63% in a single day, indicating active short-term capital movement [4].

Market Overview - The Hong Kong stock market experienced a decline, with the Hang Seng Index halting a three-day rally, closing at 27,032.54, down 0.86% [1][2] - The Hang Seng Tech Index fell by 1.65%, while the Hang Seng China Enterprises Index dropped by 1% [1][2] Sector Performance Technology Sector - Major tech stocks saw significant declines, with Meituan and NetEase both dropping over 4%, and Baidu and Tencent falling more than 2% [2][4] - Other tech companies like Kuaishou, Xiaomi, JD.com, and Alibaba also reported losses [2][4] Financial Sector - Financial stocks, including banks, Chinese brokerage firms, and insurance companies, mostly declined, adding pressure to the market [2] Consumer Sector - The restaurant sector faced downturns, with Jiumaojiu falling over 5% and Haidilao dropping more than 3% [6] - The overall consumer demand in the restaurant industry is stabilizing, with expectations for improvement in 2026 [6] Entertainment Sector - The film and entertainment sector was weak, with Emperor Culture Industries dropping nearly 6% [7] Energy Sector - The power equipment sector showed strength, with Harbin Electric rising over 13% [9] - The oil sector was active, with China Petroleum and Sinopec both gaining over 2% [14] Heavy Machinery Sector - Heavy machinery stocks continued to rise, with China National Heavy Duty Truck Group and Sany International both increasing over 5% [12] Semiconductor Sector - Semiconductor stocks saw gains, with Tensun Zhixin and Zhaoyi Innovation both rising over 20% [10] Notable Stocks - Dipo Technology surged by 8.34%, reaching a market value of 29.707 billion HKD [16] - The stock has increased 2.46 times since its IPO price of 26.66 HKD [21]

Group 1 - The collaboration between OpenDataLab, DeepLink, and domestic chip manufacturers has led to the adaptation of over 10 mainstream domestic computing power solutions, enhancing the ecological compatibility and adaptability of the MinerU project [1] - MinerU's self-developed VLM model achieves an accuracy rate of 99% in capturing elements from PDFs and complex web pages, enabling precise restoration and structured extraction of intricate mathematical formulas and nested structured tables [1] - The core value of MinerU lies in its cross-industry applicability and high parsing precision, serving as an efficient data production engine for large model development and a precise document parsing tool for government and enterprise sectors [1] Group 2 - TaiChuang YuanQi has completed the adaptation of over 30 AI large models to domestic computing power, including models like DeepSeek, Qianwen, and MinerU, covering various series such as Qwen3 Dense/MoE and multi-modal understanding models [2] - The ongoing updates and adaptations aim to accelerate the integration of intelligent computing with industry, enhancing the capabilities of both AI models and domestic chip manufacturers [2]

Group 1 - Dipo Technology (1384.HK), known as the "first stock of enterprise-level large model AI applications," saw its stock price surge by 9.89% to HKD 92.25, reaching a market capitalization of over HKD 30 billion, marking a new high since November 10 of the previous year [1] - Since its IPO on October 28, the stock has increased by 246% from its initial price of HKD 26.66 over approximately three and a half months [1] - The AI large model sector is experiencing significant interest, with multiple companies like Deepseek, Zhiyu, and MiniMax announcing new updates and releases, including the launch of the new generation base model GLM-5 by Zhiyu [1] Group 2 - The Hang Seng Index Company is set to announce the results of its review for the Hang Seng Index series for the fourth quarter of 2025 on February 13, with changes to the constituent stocks effective from March 9 [1] - A total of 43 new economy and technology companies, including Dipo Technology, meet the criteria for inclusion in the Hong Kong Stock Connect [1]

Core Viewpoint - Zhongke Shuguang's refinancing policy marks a significant step in accelerating its computing power layout, with plans to issue up to 8 billion yuan in convertible bonds to support its main business and enhance domestic computing power supply capabilities [1] Group 1: Financing and Investment - The company plans to invest the raised funds entirely into projects that support large model training and meet the AI model's demand for large-scale computing power and high-concurrency storage [1] - This convertible bond issuance is the first in the Shanghai market following the new refinancing policy, setting a benchmark for technology companies seeking financing [1] Group 2: Industry Impact - The projects funded by the convertible bonds are expected to lower the deployment threshold for large model applications and support the intelligent transformation of the industry [1] - The launch of three sets of WanKa super clusters provided by Zhongke Shuguang at the core node of the National Supercomputing Internet in Zhengzhou represents the largest domestic AI computing power pool currently in operation [1] Group 3: Financial Metrics - The current stock price corresponds to a P/E ratio of 46x for 2026 and 38x for 2027, with a target price raised by 35% to 98.98 yuan, maintaining an "outperform industry" rating [1]