深度解读：Deepseek 如何影响智驾产业 摘要 Q&A 知行汽车科技的基本情况和近期发展如何？ 知行汽车科技于 2023 年 12 月在港交所上市，成为自动驾驶领域的首家上市公 司，目前备受关注。公司每日流通量约为 2 亿，并在德国设有子公司。此外， 公司在苏州总部设有工厂，同时在马来西亚也有一个海外工厂，主要服务于海 外客户。知行汽车科技专注于自动驾驶技术的量产落地，核心竞争力包括算法 软硬一体、行波一体、仓储一体以及深厚的工程化能力。目前，公司已与吉利、 • 知行汽车科技专注于自动驾驶技术量产落地，已与吉利、长城等车企实现 多款车型量产，交付 50 万套自动驾驶系统，核心竞争力在于算法软硬一体 化及深厚的工程化能力。 • 知行汽车科技与地平线合作推出 IDC510 产品，基于 G6 系列芯片，支持高 速 NVP、自动泊车等功能，定位中端市场，已应用于奇瑞 iCar 05 项目， 预计 2025 年下半年量产。与地平线征程三芯片合作的新产品已在新纪元 ETES 及 iCar V23 车型上量产，并获得欧盟 GSR 认证。 • Deepseek 技术通过模型蒸馏降低算力需求，利好中小算力平台玩家和尚未 达 ...

Summary of Key Points from the Conference Call Industry Overview - The conference call discusses the evolution of China's renewable energy grid pricing policies and the implications of recent policy adjustments on the renewable energy sector [2][3][4]. Core Insights and Arguments - **Evolution of Pricing Policies**: China's renewable energy pricing policy has evolved since the introduction of the Renewable Energy Law in 2006, with significant changes in 2009, 2016, 2019, and 2022, culminating in a goal for full market participation by 2030 [2][3]. - **Market Transition**: Starting from June 1, 2024, new projects will default to full market pricing, eliminating government-set prices and relying on market mechanisms for revenue confirmation [2][3]. - **Sustainable Pricing Mechanism**: A new "sustainable development pricing settlement mechanism" will be implemented, allowing for a "multi-return, less-supplement" approach to ensure project revenues align with market prices [5][6]. - **Current Participation Rate**: As of 2023, nearly 50% of China's renewable energy generation is market-based, with expectations for this to increase as policies are adjusted [6][7]. - **Impact on Existing Projects**: The policy adjustments provide unexpected guarantees for existing projects, allowing them to continue operating under current pricing mechanisms as long as they do not exceed local coal power benchmark prices [7][8]. Additional Important Content - **Incremental Project Pricing Mechanism**: New incremental projects will be priced through competitive bidding, categorized by technology type to avoid price undercutting, with the highest bid determining the price [10][11]. - **Market Price Settlement**: The new market price settlement mechanism will not be based on actual transaction prices but on average prices of similar projects, encouraging operational efficiency among renewable energy companies [11][12]. - **Future Trends**: By 2025, provinces will begin long-cycle continuous settlement trials, with an increase in trading frequency for cross-province renewable energy transactions [13][14]. - **Investment Opportunities**: The renewable energy sector remains attractive for investment, particularly in offshore wind and companies with strong competitive advantages in green energy [18]. - **Challenges and Opportunities**: The market faces challenges such as price pressures in certain regions, but recent policies are expected to drive high-quality development and improve the economic viability of renewable energy projects [16][17]. - **Software Development**: Policies favoring price signals will benefit software vendors related to energy trading and power forecasting, enhancing the operational capabilities of renewable energy companies [21]. - **Storage Policy Changes**: The removal of mandatory storage requirements eases the burden on renewable energy companies, although storage remains crucial for optimizing generation and improving economic returns [22][23]. - **Future Pricing Mechanisms**: Anticipated changes in pricing mechanisms for regulatory power sources may include new policies for pumped storage and flexible coal power, contributing to a more robust renewable energy consumption framework [24].

Investment Rating - The report does not explicitly state an investment rating for the industry or specific companies within it. Core Insights - The DeepSeek AI app has gained significant traction, topping app-store downloads in the US and being available in over 170 countries, showcasing its low-cost and high-performance capabilities [3][26] - The DeepSeek model's training cost is reported at US$5.6 million, with a charge of US$0.14 per million input tokens, making it approximately 94% cheaper than OpenAI's pricing [3][26] - Major Chinese internet companies like Alibaba, Tencent, and Baidu are also advancing in AI and LLMs, with their efforts being underappreciated by investors [5][39] - The report anticipates that the DeepSeek model could lead to a global shift towards lower-cost AI models, accelerating adoption rates and application development [6][28] Summary by Sections DeepSeek Technology - DeepSeek utilizes a Mixture of Experts (MoE) architecture with 671 billion parameters, optimizing for efficiency and cost [4][31] - The model bypasses expensive CUDA infrastructure, directly optimizing for PTX, which contributes to its lower operational costs [4][27] Competitive Landscape - The report highlights that DeepSeek's success may overshadow other AI products like Bytedance's Doubao, which was previously the top productivity app in China [56][59] - Alibaba's Qwen model and Tencent's Hunyuan are also noted for their advancements in AI, with Alibaba's Qwen2.5-Max being released with over 20 trillion tokens of pre-training data [39][40] Global Implications - The report suggests that the success of DeepSeek may inspire US models to adopt similar low-cost approaches, potentially leading to a significant inflection point in AI adoption by 2025 [6][28] - Geopolitical scrutiny may arise due to the rapid adoption of DeepSeek, with some countries already restricting its use [34][36] Market Performance - DeepSeek's app climbed to 1 in both productivity and overall app rankings in the US and China shortly after its release, indicating strong market interest [55][56] - The report notes that the rapid rise of DeepSeek's app has led to a significant shift in the competitive landscape of AI applications in China [55][56]

一直致力在这种AI上面AI应用和Agent的一个创新的应用公司在2023年年初就投入了这一块的研发和一些商品化的一些落地和订单的一些逐步的转化时至今日其实已经过去了两年整的时间了我认为在这个过程中我们确实也走了一些弯路因为毕竟AI是一个新鲜的事物但是在这个过程中我们也取得了一个莫大的成绩 然后完成了一个是智能培训考试系统的一个情况当时接的是DeepSeq的V3未来上线可能会更新到R1这样的一个程度主要是在课程制作上和考试环节两大层面上去实现了DeepSeq的一个融合另外与DeepSeq上我们还做了数据分析 也是和用它的RE模型去做的大概是这两类的一个深度的应用和整合当然其实对于我们这种厂商来讲的这种因为我们自己不做这种通用大模型的训练那对于模型来讲的更多的其实是一个使用和复能的一个过程那这个过程中我们也去适配了多样化的一个厂商就是主要还是与客户的需求以及它这个成本 需求功能应用等等有比较大的一个关联性这就是目前我在弹幕型适配的一个情况我就在与大家之前公布的基础上更新这么多内容然后接下来我就介绍一下我们的今天的一个专家的嘉宾他是田总是我们的首席创新官是协同智能研究院的院长田总在咱们的B端的领域其实是有着非常深厚 ...

【九点特供】合肥助力! 安徽百信完成DeepSeek-V2版本在 产服务器上的模型适配,这家公司刚刚亲承和百信合作共 同推广销售异腾服务器产品;东风、长安两大汽车央企均筹 划童年 九点特供 2025.02.10 08:04 星期一 【市场主线】 上周五市场全天冲高回落,创业板指大涨超2.5%。盘面上,DeepSeek和比亚迪智驾概念股延续强势,光伏板块同样展开反弹, 仅银行等少数板块下跌。指数层面,两市成交额近2万亿目仅1只个股跌停,连续第二天实现了"指数放量大涨+跌停消灭"的目 标，而之所以能够大涨，就好比一个人在黑夜里摸着石头过河，等发现自己已经上岸了，便开始加速奔跑。如图所示，周五上证 冲了稻体上沿后回落，创业板指更是接近收复了年初的跌幅，但现在毕竟不像去年春节后（当时A股是从前一年国庆假期后一路 跌到2月），去年9月底之前也是连续跌了近4个月，因此经过连续两日的放量后，指数大概率会盘整一番，乐观预期下，等下方 的均线跟上后,有望突破箱子C并回到上方的箱体B开启新的区间震荡,关键还是看量能,量能看什么,看消息驱动,比如2月的 海外宏观(本周即将宣布"对等关税"措施）,3月的全国两会,4月的财报季。 题材 ...

09 Feb 2025, 08 Feb 2025, 08 Feb 2025、 test-article2 测试文章2 Commentary 评论 APAC Flows - DeepSeek propelled HK/China rally and drove East Asian tech outflows 亚太地区资金流向 - DeepSeek 推动香港/中国股市上涨，并导致东亚科技资金流出 09 Feb 2025, 09 Feb 2025、 Commentary 评论 U.S. Merger Arbitrage Monitor (2/7/25) 美国兼并套利监测（2/7/25） APAC Flows - DeepSeek propelled HK/China rally and drove East Asian tech outflows 亚太地区资金流向 - DeepSeek 推动 香港/中国股市上涨，并导致东亚科 技资金流出 更多一手调研纪要和研报数据加V：shuinu9870 更多一手调研纪要和研报数据加V：shuinu9870 更多一手调研纪要和研报数据加V：shuinu9870 更多一手调研纪要 ...

Market Insights | Markets | Equities 市场洞察 | 市场 | 股票 GS US TMT .. what a week .. GS US TMT ......多么美好的一周. GS US TMT: All references to "we/us/our" refer to the views and observations of the desk. … looking ahead, while many TMT companies have gotten "over" the 1Q / CY25 guidance wall-of-worry that weighed on sentiment during preview season (think NET, PINS, SPOT, TTWO 'clearing event' reactions) – it does feel like investors are starting to underwrite a tinge of optimism about 2Q/2H+ inflections across TMT - ...

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 Markets Strategy 03 February 2025 This material is neither intended to be distributed to Mainland China investors nor to provide securities investment consultancy services within the territory of Mainland China. This material or any portion hereof may not be reprinted, sold or redistributed without the written consent of J.P. Morgan. China Equity Strategy (852) 2800-1087 wendy.m.liu@jpmorgan.com J.P. Morgan Securities (Asia Pacific) Limited/ J.P. Morgan Broking (Hong Kong ...

这个Cybernet 包括你家光棚号公司的财报当中我们也看到关于DCI级别的东西越来越多那么其实作为基础设施中非常重要的一部分那么在北美市场上不管从光的角度来看还是从能源的角度来看整个DCI的这个大趋势已经逐步成形了已经逐步成形了所以呢这个领域里面还是会有更强的这个Beta还是会有更强的这个Beta 那么所以这两个是就是我们看到的一些一些变化当然今年呃年中的动头戏还是要开1.6T那么这个我觉得十点应该在一季度末二季度的时候会看到这些新品不开始有有有放量开始有放量那么除此以外呢呃还是会聚焦在我们的这个能源的这个赛道上 客户或受邀客户会议中嘉宾的发言内容仅代表其个人观点国胜证券研究人员的发言内容仅代表当地的判断具体请参见完整版的研究报告未经国胜证券事先书面许可任何机构或个人严禁录音转发及相关解读事前违反上述情形的我们将保留追究相关方法律责任的权利 好的各位投资朋友大家好欢迎参加国事通讯的周例会那么本周的我们的主题叫算力依旧开发开发起点要创新之旅因为在过去的几周一日其实整个板块已经有一个大的波动从开业第一天大家觉得有了这个deepfake以后整个算力受到了很大的冲击加上节假日期间英伟达波动也比较大 那么这个带来了一 ...

Summary of Conference Call Records Industry or Company Involved - The discussion primarily revolves around the **technology industry**, specifically focusing on **Apple Inc.** and its innovations in **AI**, **AR**, and **hardware components**. Core Points and Arguments 1. **Automation and Human-Computer Interaction**: The enhancement of automation capabilities significantly improves human-computer interaction experiences, with a focus on the differentiation between various downstream applications like mobile and PC, which are considered mature categories. The emphasis is on the industry's progress and valuation over the next two to three years, particularly regarding profit release elasticity and certainty [1][2][3]. 2. **AI and Hardware Innovations**: Apple is seen as a cautious leader in AI, with a focus on its proprietary architecture for AI applications. The company is expected to leverage its M-series chips for both device and cloud-based AI models, enhancing user privacy and security [5][6][8]. 3. **Market Trends and Product Launches**: The upcoming launches of new iPhone models, including the SE with a full-screen design and self-developed baseband chips, are anticipated to drive sales. The iPhone 17 is expected to start production in June, with significant updates planned for iOS 19 [6][7][9]. 4. **Innovation Cycle**: Apple is entering a three-year innovation cycle, which is expected to be the most intense in its history. This cycle is driven by the integration of AI into existing technologies, leading to increased hardware prices and a potential boost in sales [8][12]. 5. **AI Glasses and Market Competition**: The market for AI glasses is projected to grow significantly, with major players like Samsung and Xiaomi expected to release competitive products. The overall shipment of AI glasses is forecasted to reach close to 5 million units by 2025, with a tenfold increase anticipated by 2028 [12][14]. 6. **Component Demand and Supply Chain**: The demand for components such as DRAM is expected to rise due to the high resource consumption of large AI models. The need for increased battery capacity and the introduction of new suppliers are also highlighted as critical factors for growth in the consumer electronics sector [19][20]. Other Important but Possibly Overlooked Content 1. **Consumer Feedback and Sales Projections**: The actual impact of AI on sales growth remains uncertain, with a strong emphasis on consumer feedback being crucial for accurate sales forecasts [9]. 2. **Regulatory and Tariff Concerns**: Previous concerns regarding tariffs and their impact on product pricing have been addressed, with Apple successfully applying for exemptions in the past. This suggests a potential for stability in pricing despite regulatory challenges [17][18]. 3. **Emerging Technologies and Innovations**: The discussion includes the potential for new technologies such as foldable screens and AR glasses, indicating ongoing innovation beyond current product lines [10][11]. 4. **Brand Competition**: The competitive landscape among brands like Xiaomi, Samsung, and Lenovo is noted, with a focus on how these companies can leverage AI to enhance user engagement and market share [20].