摘要 2025 年中国汽车销量预计同比增长 8.4%，达 3,408.6 万辆，但 2026 年乘用车销量预计下滑 5%，受购置税政策变化影响，市场将向头部企 业集中，如鸿蒙智行、零跑和吉利江淮等有望提升市场份额。 智能化方面，L3 级智能驾驶预计 2026 年落地，高阶辅助驾驶成卖点。 比亚迪 E3 平台 3.0 和吉利 SEA 浩瀚平台规模化落地。比亚迪计划在 10 万元以上车型搭载激光雷达，加速智驾平权。 比亚迪海外工厂投产增加产能，规避关税风险，自主品牌通过技术换市 场，主导新能源和智能化。RoboTaxi 在政策、技术和成本推动下，预 计 2026 年进入商业化拐点，无人矿卡渗透率将显著提升。 智能汽车产业链是机器人产业链的原始积累，核心技术可迁移，供应链 相通性及规模化量产能力将推动机器人发展。中国新能源汽车产业链经 验或可复制到机器人领域。 2026 年汽车行业投资集中在购置税减免带来的乘用车销量增长、智能 驾驶平权和全球化产能落地、Robot Taxi 商业化及无人物流车进展、具 身智能发展。推荐吉利汽车、长城汽车 H 股，以及速腾聚创、德赛西威 等智能化标的。 Q&A 2025 年汽车行业的 ...

Core Viewpoint - The company is issuing convertible bonds to unspecified investors, with specific risks and conditions outlined for potential investors, including credit ratings and shareholder participation commitments. Group 1: Convertible Bonds Issuance - The company is issuing convertible bonds that require investors to meet specific suitability criteria for the Sci-Tech Innovation Board [3] - The bonds have a redemption clause, with a redemption price of 112% of the face value, including the last interest payment, if not converted into shares [3] - The company has engaged a credit rating agency, with the bonds rated AA- and a stable outlook [4] Group 2: Risks Associated with Bonds - The bonds are not secured, which may pose repayment risks for investors [5] - Investors who do not meet the suitability criteria may face losses if the redemption price is lower than their acquisition cost [3] Group 3: Shareholder Participation - Major shareholders, including Qingyi Investment and Ningbo Zhunzhi, may participate in the bond subscription and have committed to adhere to relevant regulations [6][7] - Shareholders with over 5% ownership, including Xu Yihua and Xu Wei, have also expressed potential interest in subscribing to the bonds under similar commitments [8][9] Group 4: Financial Performance and Risks - The company has experienced fluctuations in net profit, with figures of 15,210.36 million, 21,517.24 million, 12,469.06 million, and -1,516.90 million over recent periods, indicating potential volatility in earnings [13] - The company's gross profit margin has declined from 40.09% to 35.47%, raising concerns about future profitability [14] - Accounts receivable have increased significantly, with balances of 49,124.86 million, 51,281.85 million, 62,359.73 million, and 46,219.54 million, indicating cash flow risks [15][16] Group 5: Profit Distribution Policy - The company aims for a stable profit distribution policy, prioritizing cash dividends when conditions allow, with a minimum cash distribution of 10% of distributable profits [18][19] - The company has specific conditions for cash dividends, including positive distributable profits and sufficient cash flow [20][21] - The company plans to maintain its current profit distribution policy post-bond issuance, subject to regulatory changes [33]

Investment Rating - The report rates the optical components industry with a positive outlook, indicating a robust growth trajectory and significant investment opportunities. Core Insights - The optical components industry is projected to grow from $8.5 billion in 2019 to $15 billion in 2024, with a compound annual growth rate (CAGR) of 12.03%. It is expected to further increase from $17.2 billion in 2025 to $26.366 billion by 2029, with a CAGR of 11.27% [4][42][49]. Industry Definition - Optical components are core devices used to alter light paths, control spectra, or complete optoelectronic energy conversion. They include lenses, prisms, gratings, mirrors, filters, optical fibers, lasers, LEDs, photodiodes, and micro-nano diffraction elements. These components are essential for applications in barcode scanning, motion optics, photography, autonomous driving LiDAR, high-end cameras, AR/VR, and more [5][4]. Industry Characteristics - The industry is characterized by high-precision manufacturing, multidisciplinary integration, and a coexistence of customization and small-batch production. The manufacturing process requires extreme precision, often within nanometers, and involves advanced techniques such as ultra-precision diamond turning and ion beam polishing [9][10][11][12]. Development History - The optical components industry has evolved from manual grinding of spherical lenses in the early 1900s to a highly automated and precise manufacturing process today. The industry has transitioned through phases of industrialization, precision, and now to intelligent manufacturing, with significant contributions from countries like Germany, Japan, and China [14][15][16]. Industry Chain Analysis - The optical components industry chain consists of upstream raw material and equipment suppliers, midstream optical component manufacturers, and downstream end-users and system integrators. The industry exhibits a collaborative culture that binds materials, design, manufacturing, and application [17][18]. Market Dynamics - The market is experiencing a healthy differentiation in profits, with upstream materials accounting for over 30% of BOM costs. Leading companies like Schott maintain high gross margins, while some midstream manufacturers face negative margins due to intense competition [19][24][25]. Future Trends - The industry is expected to see a shift towards high-value segments driven by AI and high-speed optical communication. Companies that can integrate technology and respond quickly to market demands will likely gain a competitive edge [61][62][63].

Core Viewpoint - The company Desay SV aims to develop intelligent cockpit solutions that meet future smart travel scenarios, focusing on enhancing human-machine interaction [1] Group 1: Business Growth and Strategy - The intelligent cockpit business is expected to continue its growth into the first half of 2025, driven by a diversified product portfolio, market extension strategies, new product development, and strong new customer orders [1] - The rapid advancement of AI and smart driving technologies will lead to richer application scenarios for intelligent cockpits, resulting in continuous upgrades and increased penetration of higher-level intelligent cockpit products [1] Group 2: Market Opportunities and Challenges - The market space for intelligent cockpits is expected to expand continuously due to the integration and ongoing upgrades of smart driving and intelligent cockpit technologies, which will enhance vehicle efficiency, optimize costs, and improve user experience [1] - The cross-domain integration poses higher challenges for the industry, necessitating high-level R&D and engineering capabilities in both smart driving and intelligent cockpit fields [1] Group 3: Competitive Position - The company has established a leading position in both the intelligent cockpit and smart driving sectors, and the cross-domain integration is anticipated to further enhance its competitive edge in the industry [1]

| 项目类别 | 数据指标 最新 环比 | | 数据指标 | 最新 | 环比 | | --- | --- | --- | --- | --- | --- | | | 主力合约收盘价（日，元/吨） | 92,800.00 | -2040.00↓ 前20名净持仓（日，手） | -182,801.00 | -14994.00↓ | | 期货市场 | 主力合约持仓量（日，手） | 575,421.00 | -17708.00↓ 近远月合约价差（日，元/吨） | 1,340.00 | +1880.00↑ | | | 广期所仓单（日，手/吨） | 12,920.00 | -200.00↓ | | | | 现货市场 | 电池级碳酸锂平均价（日，元/吨） | 92,750.00 | 0.00 工业级碳酸锂平均价（日，万元/吨） | 90,350.00 | 0.00 | | | Li₂CO₃主力合约基差（日，元/吨） | -50.00 | +2040.00↑ | | | | 上游情况 | 锂辉石精矿(6%CIF中国)平均价（日，美元/吨） | 1,200.00 | 0.00 磷锂铝石平均价（日，元/吨） | 9,925.00 ...

Policy and Regulations - The National Energy Administration has launched a pilot program for "Artificial Intelligence + Energy" to promote the integration of AI technology with the energy industry, aiming to address industry pain points and cultivate new growth areas [8][13]. - The National Development and Reform Commission has released a list of eligible projects for Real Estate Investment Trusts (REITs) in the infrastructure sector, including clean energy projects and energy storage facilities [15]. - Shanghai has strengthened safety management for public electric vehicle charging facilities, requiring compliance with national safety standards and encouraging the use of high-quality equipment [16][17]. Market Insights - In the first ten months of 2025, Shenzhen sold 22,358 new energy light trucks, leading the market with a penetration rate of 78.8%, although growth has slowed due to market saturation [19][20]. - The sales of pure electric light trucks in Shenzhen accounted for 97.1%, with box-type trucks making up 75.7% of sales, and 62.1% of the trucks having a battery capacity of 96-110 kWh [23][24]. - The top three companies in the Shenzhen new energy light truck market hold a combined market share of 54.5%, with the leading company, Farizon Auto, capturing 21.3% of the market [25]. Company Monitoring - Farizon Auto plans to launch its first Robovan, the "Shentong T6," in January 2026, featuring an autonomous driving system designed for logistics applications [39][41]. - ZF Friedrichshafen and Cal Power have entered into a comprehensive strategic partnership in the field of commercial vehicle intelligent driving, focusing on L4 autonomous driving solutions [44][45]. - Guoxuan High-Tech has advanced the design of a 2 GWh production line for solid-state batteries, indicating progress in battery technology for electric vehicles [39].

Core Insights - The first Horizon Technology Ecosystem Conference was held on December 8, 2025, focusing on the development of smart vehicles and robots, aiming to promote the integration of smart technology into everyday life [1] Group 1: Technological Advancements - Horizon's founder and CEO, Yu Kai, emphasized the need for a solid technological foundation for the widespread adoption of smart technology, highlighting three core pillars: BPU (intelligent computing architecture), compilers, and algorithms [2] - The newly released fourth-generation BPU architecture "Riemann" achieves a tenfold increase in key computing power and enhances energy efficiency for large language models by five times, marking over a 1000-fold performance improvement in BPU over the past decade [4] - The fourth-generation compiler, OpenExplorer 4.0, reduces compilation time from hours to minutes and improves model performance by 20%, indicating a shift towards machine learning-driven self-evolution in the industry [4] Group 2: Business Model Innovation - Horizon introduced a new collaboration model "HSD Together," which allows partners to focus on their strengths while Horizon provides comprehensive algorithm services, reducing R&D costs, computing power consumption, and time to market by 90% [5] - The HSD driver assistance system has already been applied in models like the Xingtu ET5 and Deep Blue L06, achieving over 12,000 activations within two weeks of launch, making advanced driving assistance accessible in the 150,000 RMB market [5] Group 3: Market Expansion - Horizon aims to democratize advanced urban driving assistance for vehicles priced at 100,000 RMB, with a new solution based on a single Journey 6M chip set to go into mass production [6] - The Journey series chips have achieved significant market penetration, with over 10 million units of the Journey 6B/L models and over 1 million units of the Journey 6E/M models shipped in the first year [6] Group 4: Robotics and Open Source Models - Horizon has expanded its technology ecosystem beyond the automotive sector, becoming China's largest consumer robotics computing platform, collaborating with over 100 partners and connecting more than 100,000 developers [7] - Two open-source models, HoloMotion and HoloBrain, were launched, focusing on motion intelligence and spatial awareness for robots, respectively, while Horizon collaborates with over 500 educational institutions to cultivate talent in the smart industry [8] - Horizon's mission is to empower smart vehicles and robots to enhance safety and quality of life, with a goal of achieving "HSD million-unit production" in the next 3 to 5 years [8]

创新为要 在金龙客车厦门制造基地的焊装车间，数台工业机器人挥舞机械臂，在空中划出精准的弧线。伴随着激 光传感器的细微闪光，一块块钢板被精准焊接成车身骨架，整个过程在现代化生产线上高效完成。这条 国内客车行业首条柔性化智能焊装线，实现了6种车型的混线生产，生产效率提升35%，产品一次交验 合格率达99.2%。 "智能制造不是选择题，而是必答题。"公司工程研究院副院长宋光吉站在总装车间的数据大屏前介绍。 屏幕上实时跳动着生产进度、质量数据和设备状态，每条产线都配备了5G+视觉检测系统，能自动识别 装配问题并实时报警。这套智能系统使公司在2024年实现了关键工序100%数字化管控。 智能生产线驶出的不仅是传统客车，更是面向未来的智慧产品。在金龙智慧园区，记者试乘了正在运营 的L4级别量产自动驾驶巴士，无方向盘、无油门、无刹车踏板，极具科技感和现代感。公司智能车营 销中心部长柯永瑞告诉记者，"这款巴士前后安装有激光雷达、毫米波雷达等传感器，不会像人一样'开 小差'，能持续监测路面情况、周围物体，实现路牌识别、自主避障、变道绕行"。通过激光雷达、毫米 波雷达和视觉感知系统的融合，车辆能在100毫秒内完成对周围环境的感知 ...

智驾国产芯片格局变化 20251208 摘要 蔚来汽车智驾方案采用全栈自研，主推世界模型，但效果相对落后，明 年主要任务是提升车位到车位功能的接网率及处理复杂案例，乐道和萤 火虫系列预计仍将使用英伟达方案。 小鹏汽车中高端车型将搭载自研图灵芯片，算力达 1,000+TOPS，算法 重点在于 VLA 和世界模型的迭代，深度融合 BL 模块，并计划在 Robotic 业务线中使用图灵芯片优化 Robot Taxi 的通行效率和安全性。 理想汽车自研 M100 苏马赫芯片预计 2026 年 Q2 量产，首发于高端改 款车型，AD Max 系统将存在 M100 与地平线混合方案并存的情况， AD Pro 系统继续采用地平线方案，但可能升级至 G6H 版本，算法方面 坚定走 VOL 路线。 小米汽车计划在高端车型上采用英伟达 42 系列芯片，自研玄戒 O2 芯片 暂缓使用，明年将采用类似特斯拉的架构，以世界模型为主，加语言模 型辅助，解决停车场出入、道路标识牌识别及城市通勤问题。 比亚迪高端方案将升级至英伟达索尔方案，首发于仰望 U8 改款车型， 由 Momenta 提供 R6 大模型 plus 版本算法。终端方案天翼 ...

有分析师对《证券日报》记者表示，地平线通过HSD方案有效压缩了高阶智驾的硬件与算法成本，为车 企提供了兼具性能与性价比的技术路径，进而推动智能驾驶从"可选"向"标配"演进，加速智能驾驶普 惠。 AI能力向物理世界延伸 在推动智能驾驶普及的同时，地平线也在夯实其AI计算的技术底座。会上，地平线正式发布第四代BPU 架构"黎曼"，并同步推出编译器"天工开物4.0"及智能驾驶大模型，构建起以BPU、编译器和基础模型为 核心的三大技术支柱。 12月8日，在2025地平线技术生态大会上，地平线创始人兼CEO余凯回顾了过去两年地平线在高阶辅助 驾驶领域持续发力的成果。 "从2021年起，地平线持续推动高阶驾驶向高而行，2024年4月24日征程6家族重磅发布、2025年4月18日 征程6P与HSD（地平线城区辅助驾驶系统）发布。2025年11月18日起，地平线智驾HSD正式首发量 产，降低了高阶智驾门槛。首搭HSD量产落地的星途ET5和深蓝L06上市仅两周，HSD激活量超12000 辆，市场反响热烈，展现出地平线在高阶辅助驾驶领域的强大实力与广阔前景。"余凯表示。 技术底座持续迭代 "传统的两段式端到端、一段式端到端+规则 ...