Core Insights - The rapid popularity of the DeepSeek large model has led many companies and individuals to privatize its deployment, but most are operating in an unsecured manner [1][2] - A significant percentage of active Ollama model servers are exposed to the internet without authentication, increasing the risk of data breaches and service interruptions [1] - Security experts emphasize the need for immediate protective measures for those deploying DeepSeek services to mitigate potential threats [2] Group 1 - The monitoring platform by Qihoo 360 found that out of 8,971 Ollama model servers, 6,449 are active, with 88.9% operating without security measures [1] - The lack of security authentication has led to unauthorized access, which can result in data leaks and service disruptions [1][2] - Attackers can potentially issue commands to delete deployed DeepSeek and Qwen model files, posing a significant threat to system security [1] Group 2 - Automated scripts have been reported to scan for unsecured DeepSeek servers, leading to resource theft and server crashes for some users [2] - Six security vulnerabilities in the Ollama AI framework were disclosed, which could be exploited for various malicious activities, including DDoS attacks and model theft [2] - Experts recommend that all users of DeepSeek services implement effective security measures, such as modifying configurations and ensuring data transmission encryption [2]

更多资料加入知识星球:水木调研纪要 关注公众号:水木纪要 Global Research 12 February 2025 China Equity Strategy A multi-faceted look at DeepSeek's implications More to go on the DeepSeek rally? The launch of DeepSeek's R1 model has put AI development in China back on investors', radar. Our tracking of AI related China-listed stocks (summarised in Figure 1) shows that since the start of the year the stocks are up 15% and have outperformed MŞCbChina by 9%. Fundamentally Al contributes a limited proportion of most companies" revenue and ...

ab 12 February 2025 Global Research China Equity Strategy A multi-faceted look at DeepSeek's implications More to go on the DeepSeek rally? The launch of DeepSeek's R1 model has put AI development in China back on investors' radar. Our tracking of AI related China-listed stocks (summarised in Figure 1) shows that since the start of the year the stocks are up 15% and have outperformed MSCI China by 9%. Fundamentally AI contributes a limited proportion of most companies' revenue and we think the fierce compet ...

Investment Rating - The report assigns a "Buy" rating to companies such as JCET, Will Semi, and NAURA, while it recommends a "Sell" rating for companies like Sanan and Silan [4]. Core Insights - The report highlights three main implications of DeepSeek on China's tech supply chain, including the potential for better utilization of locally produced high-performance computing (HPC) chips, driving innovation in edge generative AI, and improving market sentiment leading to a re-rating of Chinese semiconductor stocks [1][10]. - The report emphasizes that the rapid development of generative AI will accelerate innovation in edge applications, stimulating demand for smarter devices and faster replacement cycles [20]. - The report notes that the Chinese semiconductor index (CITIC) rose by 17% in 2024, partly due to improved liquidity and government support, and suggests that DeepSeek may reshape investor perceptions of the Chinese tech supply chain [28]. Summary by Sections Section 1: DeepSeek's Impact on AI Supply Chain - DeepSeek is expected to enhance the utilization of local HPC chips and create a more sustainable business model for the local HPC supply chain [10]. - The strong demand for DeepSeek has led to its deployment on various local GPUs, which may benefit the domestic semiconductor supply chain [2][12]. Section 2: Edge AI Innovation - The report anticipates that edge AI will gain mainstream penetration from 2025 to 2027, driven by the popularity of DeepSeek [3][20]. - Major Chinese OEMs are rapidly integrating DeepSeek into their products, which is expected to have a significant positive impact on the entire tech supply chain [25]. Section 3: Stock Selection and Ratings - The report identifies preferred stocks such as JCET, Will Semi, and NAURA, while recommending a sell for Sanan and Silan [4]. - The report provides a detailed overview of various semiconductor companies, including their market caps, current prices, and target prices [7]. Section 4: Market Performance and Investor Sentiment - The report indicates that the Chinese semiconductor index has underperformed compared to the SOX index since 2023, suggesting a potential for re-rating as investor sentiment improves [28][29]. - The report includes performance comparisons of the Chinese semiconductor index against other indices, highlighting a significant recovery in 2024 [29].

DeepSeek以黑马之姿引爆全球人工智能界。热潮之下，网上迅速冒出一系列操作培训课程。 铺天盖地的付费教程靠谱吗？从相关报道看，目前市面上的DeepSeek课程大致可分为"小白"操作教学、 进阶实战课程，以及AI赚钱模式培训这几种。前两类售价较低、内容基础，后者则主打"快速搞钱"，鼓 吹DeepSeek如何成为"AI印钞机"，课程价格自然也不菲。虽然说得天花乱坠，但很多买家吐槽，本以为 能学到独家的赚钱技巧，结果讲授的全是拼拼凑凑的网络信息。更有甚者把卖课当噱头，将用户引入私 域，再通过其他方式宰客赚钱，实属"挂羊头卖狗肉"。 可见，蹭DeepSeek热度的培训课虽多，但本质上都是在兜售"信息差"。各路"导师"将网课吹得神乎其 神，只是为"淘金者"制造付费挖宝的幻觉。事实上，相较于传统模型，DeepSeek的推理能力已大大增 强，用户不需要再费尽心思打磨提示词，更没必要额外学习那些似是而非的结构化语言，只需要"讲人 话"，明确告知其具体需求，得到高效回答并非难事。 可见，蹭DeepSeek热度的培训课虽多，但本质上都是在兜售"信息差"。各路"导师"将网课吹得神乎其 神，只是为"淘金者"制造付费挖宝的幻觉。事 ...

Q&A 阿里 AI+DeepSeek：阿里云 摘要 阿里云与苹果的合作有哪些潜在机会和影响？ 如果未来苹果手机在中国接入阿里的大模型，考虑到国内接近 3 亿的 iOS 用户， 这将带来庞大的文本和图片处理需求，以及频繁的调用次数。这不仅有助于满 足强大的算力需求，还能推动 AI 终端应用场景的落地。手机作为便携且联网设 备，在智能语音助手、信息推送、图像和文本识别等方面具有丰富的 AI 应用场 景。此外，阿里可以借鉴苹果现有的三层模型架构，包括端侧小模型、自研语 言模型以及与 OpenAI 合作的大模型，从而进一步提升其 AI 技术能力。 目前阿里的商业模式主要通过卖算力、提供训练平台派和调用平台百炼进行变 现。如果未来与苹果加强合作，有可能探索更多商业模式，例如参考 ChatGPT 的一些定位模式或引流广告等方式。同时，在 23 年集团新 CEO 接管后，公司进 行了组织战略调整，将战略改为"AI+公有云"，并成立了三个事业部：公有云、 混合云及海外业务事业部。这些调整推动了收入增长，并改善了利润率。 过去两年间阿里云采取了哪些降价措施？效果如何？ 阿里云如何应对国产大模型 DeepSeek 扩容需求？ 随 ...

ab 12 February 2025 Global Research China Semiconductors Initial implications of DeepSeek and China's AI progress Three major implications of DeepSeek for China's tech supply chain Mainly stimulated by DeepSeek and improved investor sentiment, the China semi stock index has risen 7.5% after Chinese New Year, outperforming the HS300/SOX indexes by 5.3ppt/2.8ppts. We think DeepSeek has shown real AI computing innovation cost efficiently, and we believe more AI propagation is positive for AI computing spending ...

e公司讯，腾讯元宝今日在应用商店发布更新，并宣布DeepSeek R1模型联网、满血上线。 其支持对用户的提问进行深度思考、逻辑推理和详细回答，同时也支持联网搜索信息，以提高推理和回 答的时新性和权威性。 官方表示，腾讯元宝是依托于腾讯混元、DeepSeek大模型，基于跨知识领域和自然语言理解能力的大 模型AI产品。 ...

证券时报网讯，2月13日，四川长虹(600839)宣布，旗下AI TV正式接入DeepSeek，并支持在"深度思考 （满血R1）"和"快速响应"两个版本间自由切换，这意味着行业首个能深度思考的AI TV智能体已全面 上线。 目前，长虹电视用户终端的升级工作已全部结束，所有搭载了沧海智能体的长虹AI TV均已正式接入 DeepSeek，用户无需复杂的操作步骤即可使用DeepSeek。 此外，长虹控股集团旗下工业互联网"双跨"平台也已同时接入DeepSeek。 ...

多位分析师向记者指出，苹果和阿里巴巴合作的可能性很高，Counterpoint高级分析师Ivan Lam向21世 纪经济报道记者表示："双方大概率是进行合作。苹果需要的是更加四平八稳的合作伙伴，目前几个互 联网大厂更有合作基础，特别是团队和持续投入方面，短期也不需要大模型能赚钱。" 他进一步表示："苹果在中国更倾向去采取多供模式，所以未来会有其他大模型厂商进来。预计今年苹 果就会在中国落地Apple Intelligence。" 苹果牵手阿里巴巴？ 苹果和阿里巴巴的消息一出，就引发业界的广泛关注。对于中国地区的iPhone用户而言，已经等待苹果 AI颇久，其他手机厂商早已上阵。 虽然苹果向来在新技术上"不赶早、不怕晚"，但是随着大模型的日新月异，端侧AI的应用在快速迭代， 手机企业之间的角逐愈演愈烈。尤其是在中国这样白热化的大市场上，苹果和安卓、鸿蒙阵营之间的 AI竞争正在加剧。 此番若苹果和阿里巴巴能合作落地，将加快苹果在中国市场的AI进展，为苹果的生成式AI策略拼上重 要的一块版图。 苹果AI正在加速进入中国市场。 2月11日，有消息称，苹果正和阿里巴巴合作，共同为中国iPhone用户开发AI功能，双方已 ...