Investment Rating - The industry investment rating is "Outperform the Market - A" [9] Core Insights - The overseas AI computing sector has experienced a significant rebound, driven by five key factors reflecting optimism in the industry, increasing downstream application demand, supply from upstream chip manufacturers, confidence from major players, and infrastructure investment [7][22] - Major cloud providers in North America reported a total capital expenditure (Capex) of $71.9 billion in Q1 2025, a 62% year-on-year increase, indicating a strong consensus on AI as a key industry trend [3][18] - The rapid growth in token usage, with Google's monthly average token calls projected to increase from 9.7 trillion in April 2024 to 480 trillion in April 2025, signifies a surge in demand for computing power, particularly in AI applications [4][19] - Broadcom, a leader in ASIC chips, reported quarterly revenue exceeding $15 billion, with AI business revenue surpassing $4.4 billion, reflecting a 46% year-on-year growth, as demand for low-cost ASIC chips rises [5][20] - NVIDIA's CEO emphasized the strong and sustained demand for AI, marking the beginning of a decade-long infrastructure buildout, with the company's stock reaching a historic high of $160.98 and a market cap exceeding $3.92 trillion [6][21] Summary by Sections Capex - North American cloud giants reported a total Capex of $71.9 billion in Q1 2025, up 62% year-on-year, with optimistic forecasts for continued investment in AI infrastructure [3][18] Token - Microsoft's Azure AI infrastructure processed over 100 trillion tokens in Q1 2025, a fivefold increase year-on-year, while Google's token usage is expected to grow 50 times within a year, indicating a rapid increase in demand for AI computing power [4][19] ASIC - Broadcom's ASIC chip revenue reached over $4.4 billion, a 46% increase year-on-year, as the demand for AI applications shifts towards more cost-effective ASIC solutions [5][20] NVIDIA - NVIDIA's stock hit a record high, with the CEO highlighting the ongoing strong demand for AI and the company's strategic roadmap for product updates, reinforcing its leadership in the AI computing sector [6][21] Stargate - OpenAI and Oracle announced a significant expansion of their Stargate AI computing infrastructure project, with an initial investment of $100 billion, aiming to enhance AI infrastructure in the U.S. [7][22]

出品丨虎嗅汽车组 作者丨李赓 头图丨视觉中国 从历史经验看，高通在特定赛道的表现，最终"决定权"往往不在自己手里：在智能手机的黄金十年，是三星、小米们的极致探索，成就了骁龙的霸主地 位；而在PC市场，微软与传统软件生态的步调迟疑，也让高通的Windows on ARM战略至今未能真正突破。 当战场转移至智能汽车，"决定者"的角色再次出现，顺势落在了中国汽车行业众多的智驾参与者头上。 中国智驾伙伴，推着高通往前走 6月底，高通再次在苏州举办了"汽车技术与合作峰会"。 在这场活动上，一幅完整的中国智能汽车产业图谱被精准地呈现出来。与会者名单本身就是一份宣言：从零跑、上汽通用、北汽等整车厂，到卓驭、 Momenta、元戎启行与博世这样的核心一级供应商 （Tier1） ，再到几乎覆盖所有细分领域的软件合作伙伴，高通的座上宾囊括了产业链的每一个关键 角色。 然而，与智能手机时代高通定义行业范式、引领技术演进的"固定模式"不同，进入智能驾驶时代：中国合作伙伴的力量变得更加强大，反倒成为了整个 生态的主要推动者。 一个最明显的迹象，在于一众中国合作伙伴在实际解决方案打造过程中的引领作用。 举一个核心例子。 此前，高通汽车业 ...

Core Insights - Qualcomm's historical performance in specific sectors often depends on external players, as seen in the smartphone era where companies like Samsung and Xiaomi propelled Snapdragon's dominance [1] - In the smart automotive sector, the decision-making power has shifted to numerous Chinese automotive participants, indicating a change in the ecosystem dynamics [2][8] Group 1: Qualcomm's Position in the Automotive Industry - Qualcomm hosted an "Automotive Technology and Cooperation Summit" in Suzhou, showcasing a comprehensive map of the Chinese smart automotive industry with key players from various segments [6][7] - Unlike the smartphone era where Qualcomm defined industry standards, in the smart driving era, Chinese partners have become the main drivers of the ecosystem, leading in the development of practical solutions [8] - The recent Snapdragon 8797 chip represents a shift from Qualcomm's previous product positioning, serving as a flexible "integrated central computing platform" rather than a purely smart driving chip [12] Group 2: Competitive Landscape and Market Dynamics - The Snapdragon 8797 chip boasts an estimated sparse equivalent computing power exceeding 700 TOPS, meeting the performance needs for both cockpit and smart driving applications [13] - Chinese partners are pushing for stronger AI performance, with some opting to use the more powerful 8797 chip for cockpit solutions instead of the newly released 8397 [14][15] - Xiaomi's approach is notably radical, integrating the consumer-grade Snapdragon 8 Gen 3 chip into the cockpit and deeply combining it with NVIDIA's Thor chip, highlighting a trend towards cross-brand chip integration [17][22] Group 3: Market Challenges and Opportunities - Qualcomm's product rollout speed is not industry-leading, with the Snapdragon 8797 expected to enter mass production in early 2026, lagging behind competitors [23] - Despite challenges in software support compared to NVIDIA, Qualcomm's partners can leverage their capabilities to fill gaps, allowing Qualcomm to focus on chip design and production [25] - The current market landscape shows that Qualcomm's mid-tier products, such as the 8620 and 8650, are gaining traction as mainstream options, driven by the cost-effectiveness of combining Qualcomm hardware with Chinese partners' software [26][28] Group 4: Innovation and Strategic Shifts - The intense competitive environment in China is fostering innovation, compelling all participants, including chip giants, to adopt more agile and market-responsive development models [34] - Major global players are increasingly focusing on the Chinese market, with strategies evolving to operate independently within China while developing competitive products for the global market [36] - The collaboration between companies like GAC Toyota and local partners such as Huawei and Xiaomi signifies a broader trend of integrating local resources to enhance competitiveness in the automotive sector [35]

Core Insights - The traditional consumer electronics component suppliers are increasingly entering the automotive industry due to the ongoing wave of new energy [2] - Qualcomm's Snapdragon digital chassis solution is driving the transformation of vehicles from mechanical engineering products to intelligent mobile terminals, focusing on connectivity, computing, and AI [2] - The Snapdragon Ride platform supports various driving assistance and cockpit integration features, which are essential for automotive companies to compete in the future [2] Group 1: Qualcomm's Automotive Solutions - Qualcomm has supported over 210 vehicle models from various Chinese automotive brands in the past two years using the Snapdragon digital chassis [2] - The Snapdragon Ride platform includes specific chips for urban and highway navigation assistance, as well as a unique platform that supports both digital cockpit and ADAS functionalities [2][4] - The Snapdragon Ride Elite version supports over 40 cameras and multimodal sensors for AI-based sensor fusion, enabling precise 360-degree external coverage [4] Group 2: Industry Adoption and Collaboration - More than 20 automotive companies have announced or are developing models with ADAS features based on the Snapdragon Ride platform, including major global and Chinese automakers [5] - The platform's flexible architecture allows automotive manufacturers to run both digital cockpit and driving assistance functions on the same SoC, facilitating integrated electronic and electrical architecture [4] - Generative AI is utilized in the Snapdragon Ride platform to personalize settings based on driver behavior and preferences, enhancing the driving experience [4]

随着先进驾驶辅助系统(ADAS)技术不断成熟，消费者对高速和城市通勤场景下驾驶辅助体验的期待提升，L2级高速NOA与城市NOA加 速渗透，带动域控方案对前视一体机方案的替代，并推高中高算力SoC需求。Canalys(现并入Omdia)预计，2025年中国市场L2级及以上功 能渗透率将达62%，较2024年显著提升，高速NOA与城市NOA分别达到10.8%和9.9%。 Canalys预计2027–2028年，城市NOA增速将反超高速NOA，市场结构快速演变，将重塑SoC竞争格局。基于当前表现与节奏判断，高通 凭借Snapdragon Ride平台(包括骁龙8650、8620和8775等SoC)在产品性能与市场落地的表现，已连续两年位列Canalys中国ADAS SoC领导 力矩阵的"冠军阵营"。 驾驶辅助普及加速：市场倾向于平台化、性能和成本最优平衡SoC厂商 这一策略对SoC产品的平台化能力、算法复用效率以及端到端的交付能力提出了更高要求。同时，主机厂产品迭代节奏明显加快，小改 款车型频繁推出以应对市场竞争，这使得主机厂对SoC厂商的工具链完整度、FAE(现场应用工程师)服务响应能力和算法支持强度提出了 更加 ...

随着先进驾驶辅助系统（ADAS）技术不断成熟，消费者对高速和城市通勤场景下驾驶辅助体验的期待提升， L2级高速NOA与城市NOA加速渗透，带动域控⽅案对前视一体机⽅案的替代，并推高中高算力SoC需求。 Canalys（现并入Omdia）预计，2025年中国市场L2级及以上功能渗透率将达62%，较2024年显著提升，高速 NOA 与城市NOA 分别达到10.8%和9.9%。预计2027–2028年，城市NOA增速将反超高速NOA ，市场结构快 速演变，将重塑SoC竞争格局。基于当前表现与节奏判断，高通凭借Snapdragon Ride平台（包括骁龙8650、 8620和8775等SoC）在产品性能与市场落地的表现，已连续两年位列Canalys中国ADAS SoC领导力矩阵的 "冠 军阵营"。 当前， 驾驶 辅助 普及加速 ： 市场倾向于平台化、性能和成本最优平衡 SoC厂商 越来越多主机厂 在推进 驾驶辅助普及 进程中，选择 。 将中高阶ADAS功能下放⾄10-20万元主力⻋型 为抢占先机 ，以 ⽐亚迪、 吉利 奇瑞 和 红旗 为代表的主流主机厂 、 采取 跨平台 的 驾驶 辅助 ⽅案 排布策略 ，系统性地推出 ...

Core Insights - The collaboration between CheLianTianXia, Qualcomm, BAIC Group, and ZhuoYu Technology aims to develop integrated cockpit and driving assistance solutions for vehicles priced below 200,000 yuan, indicating a significant market opportunity [1][2] Group 1: Product Development - CheLianTianXia showcased the AL-A1 cockpit and driving assistance integration product at the Shanghai Auto Show, which is based on the Qualcomm QAM8775P chip and is expected to enter mass production by Q4 2025 [1] - The integrated cockpit project aligns with future trends in cross-domain integration and central computing architectures, offering a cost-effective solution for vehicles priced under 200,000 yuan [2] Group 2: Market Strategy - BAIC Group plans to implement the integrated cockpit and driving assistance in its vehicles, aiming to lead the industry with its platform-based approach that covers L2 ADAS to urban NOA functionalities [2] - The first batch of Robotaxi vehicles, the Extreme Fox Alpha T5, is set to launch this year, with 120 units to accelerate the mass production of L4 autonomous driving technology [2] Group 3: Technology and Innovation - ZhuoYu Technology's integration solution utilizes a single SoC to support various advanced driving assistance features, achieving approximately 30% cost reduction compared to separate systems [3] - The core technology includes an inertial dual-eye perception system that offers a low-cost alternative to traditional LiDAR, with a point cloud density ten times that of standard LiDAR [3] Group 4: Chip Technology - Qualcomm's Snapdragon 8775 chip, launched in January 2023, provides a dense computing capability of 96 TOPS, supporting both digital cockpit and ADAS functionalities [3][4] - The chip is optimized for scalable performance, allowing automakers to choose suitable performance levels for different vehicle tiers, thus enhancing user experience with immersive graphics and audio [4] Group 5: Industry Perspective - The automotive industry is currently facing a disconnect between smart cockpits and driving assistance systems, with many companies using multiple chips for integration, which is seen as insufficient [4] - The ultimate goal is to achieve complete integration of hardware and software, allowing for a centralized computing architecture that enhances user feedback and perception during vehicle operation [4]

Core Insights - The collaboration among CheLianTianXia, ZhiYu Technology, Qualcomm, and BAIC Group aims to launch the cockpit and driving integration domain control project, expected to achieve global mass production by Q4 2025 [1] - The cockpit and driving integration project utilizes Qualcomm's QAM8775P chip and is designed to meet the cost requirements for vehicles priced below 200,000 yuan, indicating a significant market potential [1][2] - The integration solution offers advanced features such as cross-layer memory parking, high-speed navigation, and urban navigation, representing a comprehensive approach to cockpit and driving integration [1][2] Technical Aspects - The core technology of the integration solution includes inertial navigation dual vision, end-to-end large models, and Hypervisor virtualization for resource management and functional isolation, enhancing system safety and usability [2] - Qualcomm's Snapdragon 8775 chip provides a computing power of 96 TOPS, supporting mixed critical workloads and ensuring isolation for specific ADAS functions, which is crucial for automotive safety [2][3] - The integration platform aims to deliver a user-friendly experience by learning from vast driving data, adapting to various scenarios, and reducing reliance on manual rules [3] Market Positioning - BAIC Group emphasizes that the cockpit and driving integration aligns with the automotive industry's shift towards centralized electronic and electrical architecture, aiming to provide a cost-effective urban NOA (Navigation on Autopilot) solution for consumers [3] - The integration AI platform will be applied across all major models, including the Alpha T5 and S5, with a target to make urban NOA accessible at a price point of 100,000 yuan [3] - The initiative highlights the importance of collaboration in the automotive sector, focusing on technological innovation to contribute to the development of smart vehicles [3]