摘要 AI 应用投资机会梳理 20260112 AI 应用边际改善显著，大语言模型迭代加速，2025 年已达季度级别， 谷歌 Gemini、Anthropic 和 OpenAI 等头部实验室竞争激烈，模型性 能通过范式革新实现脉冲式提升，在线学习或终身学习成为新方向。 多模态模型发展潜力巨大，目前处于早期阶段，但未来有望实现跨越式 发展。OpenAI 的周活跃用户（WAU）已接近 10 亿，预计 2026 年底 可能达到 20 亿，AI 已成为全球流量格局中不可忽视的一部分。 国内外用户付费习惯差异影响国内 AI 应用市场，海外 C 端订阅模式在国 内推广受阻，B 端收费亦存在困难。教育等增值服务领域仍有机会实现 收入增长，AI 成果显著的公司将获得更多关注。 港股阿里巴巴、快手、美图和富博等公司在 AI 应用方面领先，值得关注。 阿里巴巴积极布局 AI 优化供应链和客户体验；快手利用 AI 改进内容推 荐；美图通过 AI 提升图像处理功能；富博在特定领域拥有先进 AI 技术。 OpenAI 大幅上修 2026-2029 年营收预期，探索电商和广告变现免费 用户，计划 2026 年实现 30 亿美元的免费用户 ...

Core Insights - DeepSeek has introduced a new research paper titled "Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models," in collaboration with Peking University, focusing on enhancing large language models (LLMs) through conditional memory and a new module called Engram [1][3][4]. Group 1: Research Background and Problem Statement - Current large language models primarily utilize Mixture of Experts (MoE) for sparsity, but existing Transformer architectures lack native knowledge retrieval mechanisms, leading to inefficient simulation of retrieval behavior [3][9]. - DeepSeek proposes conditional memory as a complementary approach to MoE, introducing the Engram module to address the limitations of current models [4][9]. Group 2: Engram Module and Its Functionality - The Engram module modernizes classic n-gram embeddings, enabling knowledge retrieval with O(1) time complexity [9]. - Engram separates static knowledge storage from dynamic computation processes, enhancing the model's ability to perform complex reasoning by offloading the reconstruction burden from the model's shallow layers [11][13]. Group 3: Performance Improvements - Engram has been scaled to 27 billion parameters, showing significant performance improvements over pure MoE baseline models under equivalent parameter and FLOPs conditions [11]. - Notably, Engram enhances knowledge retrieval capabilities, with improvements in metrics such as MMLU (+3.4), CMMLU (+4.0), and general reasoning tasks like BBH (+5.0) and ARC-Challenge (+3.7) [11][38]. Group 4: System Efficiency and Scalability - Engram's deterministic addressing supports prefetching from host memory at runtime with minimal performance overhead, allowing for efficient memory management [12][19]. - The architecture allows for the decoupling of parameter storage from computational resources, facilitating linear scalability with the number of accelerators [21][22]. Group 5: Experimental Results - Four models were trained: Dense-4B, MoE-27B, Engram-27B, and Engram-40B, all using the same training data and processes [35][36]. - Sparse architectures (MoE-27B, Engram-27B/40B) significantly outperformed the dense model (Dense-4B) across various benchmarks, demonstrating superior scaling properties [38]. Group 6: Long Context Training - Engram architecture has shown significant advantages in long-context tasks by preserving valuable attention capacity for global context processing [41]. - Controlled experiments indicate that Engram outperforms MoE models in complex retrieval tasks, confirming its architectural superiority [46].

Core Insights - DeepSeek has released a significant technological achievement by open-sourcing a new paper and module called Engram, which introduces a "lookup-computation separation" mechanism to enhance the performance of large language models in various tasks [2][5]. Summary by Sections Introduction of Engram - Engram is a scalable, lookup-based memory module designed to improve the efficiency of language models by separating memory retrieval from computational tasks [10][18]. Need for Engram - Traditional large language models rely on Transformer and Mixture-of-Experts (MoE) architectures, which combine memory and computation in a way that can lead to inefficiencies. Engram aims to address this by allowing models to handle factual memory and logical reasoning separately [8][9]. Core Technology of Engram - Engram utilizes modernized hashed N-gram embeddings, allowing for O(1) time complexity in memory retrieval, which significantly reduces computational costs while maintaining high retrieval speed [11][13]. Relationship with MoE - Engram provides a new axis of sparsity that complements MoE by offering static memory retrieval capabilities, thus optimizing parameter efficiency. In a 27 billion parameter model, Engram can utilize a large number of parameters for memory while consuming minimal computational resources during inference [15][16]. Performance Metrics - Engram has shown improved performance metrics across various benchmarks, such as achieving a loss of 1.950 on the Pile dataset and an accuracy of 60.4% on MMLU with 5-shot learning, outperforming both Dense and MoE models [17]. Community Reception - The Engram technology has received positive feedback from the community, with users highlighting its potential to separate memory pattern retrieval from neural computation, marking a new direction in model architecture design [18][19][21]. Future Implications - Observers speculate that Engram will be a core component of DeepSeek's upcoming V4 model, indicating a significant architectural advancement in memory and reasoning collaboration [22][23].

Group 1 - The core viewpoint of the news is that Zhiyun (02513.HK) has formed a strategic partnership with Didi to explore key technologies in General Artificial Intelligence (AGI) and their applications in the transportation sector [1] - Didi has been increasing its investment in large models and intelligent agents, leading to innovations such as AI travel assistants and business travel assistants [1] - Zhiyun has a strong foundation in large model architecture, training paradigms, and intelligent agent technology, and the partnership aims to enhance the deployment of agents in complex business scenarios [1] Group 2 - Zhiyun was established in 2019 and focuses on developing advanced general large models, launching China's first proprietary pre-trained large model framework, GLM, in 2021 [2] - The company has achieved significant growth in its cloud-based Model as a Service (MaaS) and subscription business, with over 2.9 million users on its API platform [2] - According to Frost & Sullivan, Zhiyun ranks first among independent general large model developers in China and second overall, with a market share of 6.6% as of 2024 [2] Group 3 - Zhiyun's R&D investments from 2022 to 2024 are projected to be 84 million yuan, 529 million yuan, and 2.195 billion yuan, with 1.595 billion yuan allocated for the first half of 2025 [3] - The company has a research team of 657 members, with R&D personnel making up 74% of its workforce [3] - On January 12, Zhiyun's stock surged by 31.40%, closing at 208.4 HKD per share, with a market capitalization of 91.744 billion HKD, reflecting a cumulative increase of 79.35% since its IPO [3]

Core Viewpoint - The article emphasizes that artificial intelligence (AI) is not merely a "replacement" for humans but rather a "creator" and "reshaper" in the financial industry, highlighting the irreplaceable role of human judgment, creativity, and responsibility in the face of advancing technology [3]. Group 1: Financial Innovation and Human Role - Liu Xiaochun, Vice President of Shanghai New Finance Research Institute, asserts that financial innovation should focus on the essence of finance rather than technology, maintaining that the core role of humans in technology application remains unchanged [4]. - He categorizes relevant technologies into three levels: financial technology (designing financial solutions), institutional technology (establishing reasonable distribution of benefits and risks), and scientific technology, emphasizing the need to balance these elements for effective financial technology implementation [5]. - Despite ongoing efforts to reduce workforce in the banking sector over the past two decades, the total number of employees has continued to grow, indicating that while technology replaces certain roles, it also creates new demands for technical services [5]. Group 2: Financial Intelligence and Technological Pathways - Yuan Yue, Chairman of Zero Point Data, outlines the evolution of financial technology into a new phase centered on "financial intelligence," transitioning from early automation to intelligent decision-making [6]. - He introduces a framework using the A-Z method to identify core technologies supporting risk control and service optimization, while critiquing the current hype around large language models, stating they are insufficient for high-sensitivity financial applications [6][7]. Group 3: AI's Impact on Content Creation and Financial Services - The conference also explored the intersection of financial technology and content creation, with discussions on how AI's rapid development fundamentally impacts various industries, particularly finance [7]. - Zhang Wenyu from Zhejiang University highlights that the emergence of ChatGPT marks a shift from "weak AI" to a new era of AI that mimics human-like responses, emphasizing the importance of human creativity and insight in the face of AI advancements [8]. - Financial entrepreneur Zhu Guangye acknowledges the reality that many repetitive tasks in finance will be replaced by AI, but stresses that not all foundational work can be easily substituted, particularly in areas requiring human judgment and experience [9]. Group 4: Tools and Applications in Financial Technology - Tian Li, COO of Yingfan Technology, discusses the launch of the first intelligent agent swarm, aimed at enhancing user understanding and transforming passive data into personalized insights for financial institutions [9]. - The consensus among various industry experts is that financial technology provides precise tools for content creation, while content creation supports user education and brand communication for financial products, indicating a deep integration that will drive high-quality industry development [9].

Investment Rating - The industry investment rating is "Increase" [5] Core Insights - The era of AI for Science (AI4S) is transforming scientific research, particularly in materials development, which has become increasingly complex due to multi-objective optimization requirements. AI4S utilizes AI algorithms to enhance molecular structure insights through quantum physics calculations and integrates real-world data from high-throughput robotic laboratories, significantly shortening research cycles [1] - The potential market size for AI4S in the pharmaceutical sector is estimated at approximately $108.2 billion, based on a 33% value share of the preclinical research market within the global pharmaceutical market of $1.64 trillion. Additionally, assuming a 25% penetration rate in sectors such as chemicals, pharmaceuticals, new energy, alloys, displays, and semiconductors, the total AI4S market demand could reach around $148.6 billion [2] - Key application areas for AI4S include innovative drug development, where the complexity of drug research aligns well with AI capabilities, and space photovoltaics, particularly with perovskite materials that can significantly enhance satellite energy efficiency [3] Summary by Sections AI4S Empowerment in Scientific Research - AI4S capabilities encompass "reading, computing, and doing." For instance, the company Tai Holdings has developed a patent data mining platform that can extract literature and patent data in one hour with a 95% accuracy rate, and over 200 AI models that enhance research speed and precision [1] Market Size and Potential - The pharmaceutical sector's AI4S market potential is approximately $108.2 billion, while the overall market demand across six sectors could reach about $148.6 billion under a 25% penetration assumption [2] Notable Application Areas - Innovative drug development is a primary focus for AI4S due to the high investment and complexity involved. Additionally, perovskite materials in space photovoltaics present a promising area for AI optimization, addressing technical challenges related to stability and efficiency [3][4]

Group 1 - The core viewpoint of the article highlights that 2023 may be a breakout year for AI for Science (AI4S), driven by significant advancements in capabilities that allow for more extensive scientific research [1] - AI4S is now recognized as a strategic battleground among global tech giants, moving beyond the "proof of concept" stage in laboratories [1] - AI4S is aiding scientists in efficiently identifying new research opportunities, such as predicting protein functions, designing new materials, and discovering new targets [1] Group 2 - Related A-share concept stocks mentioned include Health元 and Northeast Pharmaceutical [2]

Group 1: IPO Dynamics - Beijing Zhipu Huazhang Technology Co., Ltd. (Zhipu) officially listed on the Hong Kong Stock Exchange on January 8, 2026, with the stock code "2513," becoming the "first global large model stock" [3] - Zhipu plans to issue 37,419,500 H-shares, with a subscription rate of 1,159.46 times for public offerings in Hong Kong and 15.28 times for international offerings, raising over HKD 4.3 billion at an issue price of HKD 116.2 per share [3] - Zhipu's revenue from 2022 to 2024 is projected to grow from CNY 0.57 billion to CNY 3.12 billion, with a compound annual growth rate (CAGR) of 130% [4] Group 2: Market Position and Growth - The domestic large language model market is expected to grow 20 times in the next six years, with enterprise demand leading the way, providing Zhipu with a competitive advantage [5] - Zhipu's revenue is expected to exceed USD 100 million in 2025, with projections of approximately CNY 1.6 billion and CNY 2.7 billion for 2026 and 2027, respectively [5] Group 3: Other Companies' IPOs - Shanghai Tianshu Zhixin Semiconductor Co., Ltd. (Tianshu) also listed on January 8, 2026, with a total issuance of 25,431,800 shares and a subscription rate of 414.24 times for public offerings in Hong Kong [7] - Tianshu's revenue grew from CNY 1.89 billion in 2022 to CNY 5.40 billion in 2024, with a CAGR of 68.8% [9] - MiniMax Group Inc. (MiniMax) listed on January 9, 2026, with a total issuance of 29,197,600 shares, achieving a subscription rate of 1,837.17 times for public offerings in Hong Kong [10] - MiniMax's revenue is projected to grow from USD 3.5 million in 2023 to USD 30.5 million in 2024, with a year-on-year increase of 782.2% [11] Group 4: Financial Performance - Shenzhen Jingfeng Medical Technology Co., Ltd. (Jingfeng) listed on January 8, 2026, with a revenue of approximately CNY 1.49 billion in the first half of 2025, representing a nearly 400% year-on-year growth [14] - Hunan Mingming Hen Mang Commercial Chain Co., Ltd. (Mingming) achieved a retail sales volume of CNY 661 billion in the first three quarters of 2025, a year-on-year increase of 74.5% [17] - Hunan Sangnisendi Group Co., Ltd. (Sangnisendi) reported revenues of CNY 1.07 billion and CNY 2.45 billion for 2023 and 2024, respectively, with a year-on-year growth of 129.5% [24] Group 5: Industry Insights - The AI and semiconductor sectors are experiencing rapid growth, with companies like Zhipu and Tianshu leading innovations in large models and computing power [4][9] - The food and beverage retail sector is also expanding, with companies like Mingming and COMMUNE establishing significant market positions [17][21] - The medical technology field is advancing with companies like Jingfeng and DeShi Biotech focusing on robotic surgery and AI in medical imaging [14][35]

Core Insights - The report presents a systematic framework for automated factor research based on Large Language Models (LLM) and Prompt Engineering, aiming to explore the potential applications of AI in the entire quantitative investment chain [1] - The framework was first applied to low-frequency price-volume factors, optimizing the classic Alpha158 factor library and transitioning from an "optimization" paradigm to a "generation" paradigm [1] - AI demonstrated strong factor discovery capabilities in both fundamental and high-frequency data domains, successfully generating new factors and enhancing traditional factor libraries [1] - The report also explores AI's application in unstructured text analysis, utilizing the Gemini model to interpret sentiment from extensive research memos, creating unique sentiment indicators that effectively integrate into stock selection strategies [1] Group 1: Low-Frequency Price-Volume Factor Optimization - The framework was initially applied to the optimization of low-frequency price-volume factors, using the Alpha158 factor library as a foundation for optimization experiments [1] - AI identified logical flaws in original factors and proposed effective improvements, with optimization effects being consistent across multiple time windows from 5 to 60 days [1] - New factors generated by AI, with low correlation to sample factors, showed robust out-of-sample performance, with some factors achieving an Information Coefficient Information Ratio (ICIR) above 1.0 [1] Group 2: Fundamental and High-Frequency Factor Discovery - In the fundamental dimension, AI not only generated enhanced versions of classic factors but also innovatively expanded value, quality, and growth factors from novel perspectives [1] - In the high-frequency dimension, AI was empowered to directly generate Python code, uncovering a set of novel and high-performing high-frequency factors, with some strong signal factors achieving annualized returns exceeding 60% [1] - Integrating the AI-generated high-frequency factor library into the AGRU neural network model significantly improved annualized excess returns from 18.24% to 25.28% [1] Group 3: Alternative Data Processing and Sentiment Analysis - The report investigates AI's potential in processing alternative data, analyzing nearly one million words of research memos using the Gemini 2.5 Pro model [1] - A weekly sentiment factor was constructed, revealing unique asymmetric predictive capabilities, where negative sentiment strongly predicted future price declines, achieving annualized excess returns of 8.26% [1] - This sentiment factor exhibited low correlation with traditional price-volume and fundamental factors, serving as an independent and effective supplementary information source [1] Group 4: Comprehensive Strategy Development - A multi-dimensional information fusion strategy was developed, integrating AI-discovered high-frequency factors with low-frequency market data into the AGRU neural network to form a core Alpha [1] - The final strategy, enhanced by AI sentiment factors for risk adjustment, improved annualized excess returns from 11.15% to 11.81% while maintaining turnover rates [1] - The strategy demonstrated a significant increase in the information ratio from 2.18 to 2.31, validating AI's potential to empower quantitative research across multiple stages and achieve a "1+1>2" effect [1]

Group 1 - MiniMax, an AI large model company, has officially listed on the Hong Kong Stock Exchange, closing at 345 HKD with a rise of over 109% [3] - Another domestic AI large model company, Zhipu, also debuted on the Hong Kong market the previous trading day [3] - The recent listings of domestic AI large model companies are seen as a sign that the industry is transitioning from the technology development phase to a stage where technology and commercialization are synchronously implemented, with business models becoming clearer [3] Group 2 - According to a report by Sullivan, the market size of China's large language model is projected to reach 5.3 billion CNY in 2024, and is estimated to grow to 101.1 billion CNY by 2030, with a compound annual growth rate of 63.5% from 2024 to 2030 [3]