每经AI快讯，1月29日，港股大模型概念股MINIMAX-WP涨幅扩大至20%，股价一度涨至590港元创上 市来新高。 ...

Core Viewpoint - Alibaba's chip design unit, Pingtouge, has launched a high-end AI chip named "Zhenwu 810E" following reports of its potential IPO, indicating significant progress in external market penetration and product development [1][2]. Group 1: Product Launch and Features - The "Zhenwu" PPU has been deployed in multiple clusters on Alibaba Cloud, serving over 400 clients including State Grid, Chinese Academy of Sciences, Xiaopeng Motors, and Sina Weibo [1]. - The PPU features a self-developed parallel computing architecture and inter-chip communication technology, with a memory of 96G HBM2e and an inter-chip bandwidth of 700 GB/s, suitable for AI training, inference, and autonomous driving [1]. - The overall performance of the "Zhenwu" PPU surpasses Nvidia's A800 and mainstream domestic GPUs, and is comparable to Nvidia's H20, with an upgraded version reportedly outperforming Nvidia's A100 [2]. Group 2: Strategic Integration and Market Position - Alibaba is integrating its AI, cloud computing, and chip technologies through a collaboration known as "Tongyun Ge," which includes Tongyi Lab, Alibaba Cloud, and Pingtouge, allowing for focused hardware development and optimized AI frameworks [2]. - The company aims to establish "Tongyun Ge" as an AI supercomputer, leveraging its self-developed chips, leading cloud services, and open-source models to achieve maximum efficiency in training and deploying large models [2]. - Alibaba and Google are currently the only two global tech companies with competitive capabilities across large models, cloud, and chip sectors [4].

每经AI快讯，1月29日，港股大模型概念股盘初拉升，MINIMAX-WP涨超10%，智谱涨超6%。 （文章来源：每日经济新闻） ...

"岭南文化大模型通过备案，是我们打造地域文化AI新范式的关键里程碑""备案是合规的基石，更是信 任的起点"……以"备启新程 粤见未来"为主题的广东省备案大模型发布会（第二期），1月28日下午在 粤港澳大湾区生成式人工智能安全发展联合实验室（以下简称"联合实验室"）举行，最新一批备案大模 型企业现场领取"生成式人工智能服务备案证明"，并分享备案给企业带来的实实在在好处。截至目前， 广东备案大模型达到124款，AI产业从"单点突破"迈向"生态共振"。 广东备案大模型数量达124款 大模型是当前全球科技竞速与产业跃迁的"核心引擎"。据2025年12月国家网信办相关公告显示，全国已 完成备案的生成式人工智能服务累计达748款。 广东坚决贯彻落实中央决策部署，牢牢把握发展机遇，推动大模型备案工作走在全国前列。截至目前， 广东通过备案的大模型数量达到124款，其中最新一批通过备案的大模型有19款，位居全国前列。大模 型备案数量激增，是广东AI产业从"单点突破"迈向"生态共振"的生动写照。 当天参与发布会的既有刚通过备案的岭南文化等不同领域行业大模型企业代表，涉及文化、教育、电 竞、传媒、工业制造、市场营销等，又有正在申请 ...

在发布会上，8家新完成备案的大模型企业相关负责人进行主题分享，结合自身备案大模型讲解"大模型 特色在哪里""备案给企业带来哪些特效""充分发挥备案大模型应用创新力有什么特长"。 斑马儿童科教大模型试验室负责人柳景明表示，目前已拥有斑马AI学、斑马百科、斑马图书、斑马玩 教具等全方位业务矩阵。基于斑马科教大模型，今年推出"斑马口语"。 21世纪经济报道记者 林典驰 深圳报道 1月28日下午，在粤港澳大湾区生成式人工智能安全发展联合实验室（简称：联合实验室）举行，最新 一批备案大模型企业现场领取"生成式人工智能服务备案证明"。 截至目前，广东备案大模型达到124款，AI产业从"单点突破"迈向"生态共振"。 大模型是当前全球科技竞速与产业跃迁的"核心引擎"。据2025年12月国家网信办相关公告显示，全国已 完成备案的生成式人工智能服务累计达748款。 截至目前，广东通过备案的大模型数量达到124款，其中最新一批通过备案的大模型有19款，位居全国 前列。大模型备案数量激增，是广东AI产业从"单点突破"迈向"生态共振"的生动写照。 当天参与发布会的既有刚通过备案的岭南文化等不同领域行业大模型企业代表，涉及文化、教育、电 ...

本报记者 昌校宇 2026年以来，港股市场再度成为公募机构的重点布局方向。中国证监会官网公布的信息显示，截至1月28日，年内已有27 只港股主题基金密集上报，科技、医药、周期三大赛道成为布局焦点。与此同时，资金正借道ETF持续南下，其中，富国中证 港股通互联网ETF份额突破1000亿份，成为首只跻身"千亿俱乐部"的港股主题ETF。 在资源品价格上涨的周期背景下，港股周期板块也进入公募机构的"掘金"视野。永赢基金申报的港股通周期慧选混合基 金、汇添富基金上报的港股通周期精选混合基金等，将目光投向有色金属、能源等传统周期行业。 新基金密集申报的背后，是资金持续流入港股市场的趋势性力量。今年以来，港股主题ETF持续获得资金净流入，并催生 出规模增长迅速的产品。Wind资讯数据显示，截至1月28日，富国中证港股通互联网ETF最新规模达到910.07亿元，最新份额为 1001.51亿份，成为首只份额超千亿的港股主题ETF。 兴银基金指数与量化投资部负责人、基金经理林学晨对《证券日报》记者表示："展望2026年，港股市场在估值低位、资 金持续流入以及科技股与全球产业链深度绑定的综合优势下，有望迎来估值修复与盈利增长的双轮 ...

Core Insights - Fourier Exploration aims to integrate embodied intelligence technology with brain-machine interface (BMI) technology to enhance rehabilitation efficiency and improve patient quality of life [1][4] - The CEO, Gu Jie, announced plans to scale the introduction of BMI technology in robotic rehabilitation training within the next 1-2 years, emphasizing the importance of recognizing patient intent during training [1][3] Group 1: Technology Integration - The combination of BMI and robotic systems allows for real-time recognition of patient intentions, enabling robots to provide appropriate assistance during rehabilitation [3][4] - Traditional rehabilitation methods often lead to passive training, while the new approach encourages active participation from patients, resulting in better rehabilitation outcomes [3][4] Group 2: Development Challenges - There is a lack of large-scale datasets and necessary hardware and software infrastructure for BMI and embodied intelligence technologies globally [1][6] - The company has initiated a collaborative project with several academic institutions to enhance interdisciplinary cooperation and support the exploration of BMI and embodied intelligence integration [6] Group 3: Technological Advancements - Recent advancements in BMI hardware have led to lighter, modular devices, and the transition from single EEG systems to multi-modal signal acquisition methods has improved accuracy and reliability [5][6] - The development of large models has transformed the processing of EEG signals, allowing for more effective classification and interpretation of complex brain signals [5]

Core Viewpoint - The article emphasizes that AI glasses without display functions are better suited to meet the underlying needs of traditional eyewear, suggesting they are likely to see earlier market adoption compared to AR glasses [1] Industry Growth Potential - The smart glasses industry is expected to enter a rapid growth phase, with companies in the supply chain poised to benefit from an upturn in market conditions [2] - The launch of the Ray-Ban Meta smart glasses marks a new product cycle, featuring improved audio-visual quality and integration with Meta's social platforms, enhancing user experience and application scenarios [2] Market Penetration Insights - AI glasses are currently at a nascent stage, with a projected penetration rate of only 0.1% in 2024, similar to the early adoption phase of smartwatches [3] - Forecasts indicate that global AI glasses shipments could reach 291 million units by 2034, with a compound annual growth rate (CAGR) of 55.16% from 2025 to 2034 [3] Product Development Challenges - Key factors hindering the rapid penetration of smart glasses include reliance on traditional offline channels for customer engagement, necessitating partnerships between brands and retailers [4] - The comfort of AI glasses is influenced by weight and design, with current models averaging around 50g, while traditional glasses weigh about 20g; achieving a balance between functionality and comfort is crucial [4] - Battery capacity and performance are critical for convenience, requiring advancements in battery materials and technology to enhance longevity [4] - The system-on-chip (SoC) is vital for the performance of AI glasses, impacting user experience through its ability to handle various tasks efficiently [4]

Core Viewpoint - The article discusses the strategic positioning of the company "月之暗面" (Moon's Dark Side) in the competitive landscape of AI models, particularly in light of the upcoming release of DeepSeek's new model. The release of K2.5 is seen as a defensive move rather than an offensive one, aimed at stabilizing the company's position before potential disruptions from competitors [5][9]. Group 1: K2.5 Model Release - The K2.5 model was released on January 27, 2026, with founder 杨植麟 (Yang Zhilin) personally promoting it, which is unusual in the industry and signals a strategic shift [5][7]. - K2.5 is viewed as an upgrade rather than a groundbreaking model, focusing on enhancing coding capabilities and the Agent system, especially for overseas markets [11][12]. - The release is interpreted as a necessary step to maintain relevance and presence in the market, especially before DeepSeek's anticipated new model launch [8][14]. Group 2: Industry Context and Competition - The AI model industry is undergoing a significant reshuffle, with several companies transitioning away from being solely "base model" providers, leaving 月之暗面 as one of the few remaining in this category [16][17]. - The narrative in the industry is shifting from "who can create the strongest model" to "who can effectively utilize the model," indicating a higher risk for companies that continue to focus solely on base models [18][22]. - 月之暗面 is attempting to pivot towards coding and Agent capabilities, but it still heavily relies on model performance for revenue, particularly from overseas markets [21][22]. Group 3: Strategic Concerns and Anxiety - The company faces anxiety regarding its dependency on model capabilities, especially with competitors like DeepSeek potentially outpacing it in model advancements [23][24]. - The transition to focusing on coding and Agent systems presents long validation cycles, making it challenging for 月之暗面 to demonstrate its strategic shift compared to competitors who can quickly showcase model improvements [25]. - The memory of the impact from DeepSeek's previous model release in 2025 creates pressure for 月之暗面 to avoid being caught off guard again, prompting the early release of K2.5 as a preemptive measure [26][27].

Investment Rating - The report assigns an "Accumulate" rating for the smart glasses industry, indicating a positive outlook for investment opportunities in this sector [6][75]. Core Insights - The smart glasses industry is expected to enter a rapid growth phase, with related companies poised to benefit from an upturn in market conditions [6][75]. - The report highlights that AI glasses without display functions are likely to meet the underlying demand logic of traditional eyewear, potentially leading to quicker market penetration [2][6]. Summary by Sections Product Development and Market Penetration - Continuous improvement in product capabilities is anticipated, with sales expected to increase. Key factors affecting rapid market penetration include reliance on traditional offline channels for final delivery, the need for lightweight materials for comfort, battery capacity constraints, and the performance of SoC chips [4][24][32][33][38]. Market Trends and Projections - The report draws parallels with the smart watch market, predicting that AI glasses will have a penetration rate of only 0.1% in 2024, similar to the early stages of smart watches. It forecasts that global AI glasses shipments will reach 291 million units by 2034, with a CAGR of 55.16% from 2025 to 2034 [21][23][75]. Key Product Launches - The report discusses the launch of Ray-Ban Meta, which features significant improvements in audio and video quality, and integrates with Meta's social platforms, enhancing user experience and application scenarios [10][11][15][71]. Investment Recommendations - The report recommends investing in leading domestic lens brand Mingyue Lens and company Inpai, which are expected to benefit from the industry's growth. Mingyue Lens is actively collaborating with top brands in the industry, while Inpai is expanding into the smart sports equipment sector [6][75][76].