克雷西 发自 深圳 量子位 | 公众号 QbitAI 一个问题，在当前的智能手机中，如果AI需要成为具有自主意识、会主动实现功能的"常驻能力"，而不只是一个需要频繁被动焕新的"功能模 块"，什么样的芯片架构才能真正跟得上这样的改变？ 联发科给出的答案是：以更犀利的算力和更友好的能效表现，创造 超性能+超能效双NPU架构 ，始终让"AI Always on"。 这是一次从技术形态到使用方式的转变：目的是 让AI不再依赖被动唤醒，而是作为系统能力始终在线 、随时响应，融入用户的每一次操作。 这一趋势正在形成共识。 随着大模型下沉，端侧AI的使用频率越来越高，从输入法里的预测补全，到拍照时的构图建议，从锁屏摘要到图像生成，AI正在从"调用一 次"变为"时刻可用"。 为此，SoC不仅要能跑得快，更要 让AI跑得久、跑得稳 ，甚至在用户毫无察觉的情况下完成实时响应。 天玑9500 围绕这一目标重构芯片底座：首发双NPU架构，结合存算一体、硬件压缩等多项关键技术，在ETHZ苏黎世移动SoC AI榜单中蝉联 榜首，相比上一代跑分翻倍。 天玑9500正在让手机的AI变得更快、更聪明，也更贴近你的使用节奏。 写文案、整理想法、 ...

Investment Rating - The industry investment rating is "Positive (Maintain)" [5] Core Viewpoints - The smart terminal market is expected to flourish, driven by the recovery in iPhone sales and the entry of large model manufacturers like Open AI, which will accelerate innovation in smart terminal categories and forms [3][8] - The iPhone 17 series has shown strong sales performance, with a fourfold increase in trade-in sales compared to the previous year, indicating a positive outlook for the supply chain [8] - Open AI's entry into AI hardware terminals is expected to enrich the product ecosystem, benefiting existing supply chain companies like Luxshare Precision and GoerTek [8] Summary by Sections Investment Suggestions and Targets - The report recommends buying shares of several companies in the AI supply chain, including Luxshare Precision (002475), Lens Technology (300433), and GoerTek (002241) among others [3] Industry Dynamics - The report highlights the potential for rapid growth in the AI hardware terminal market, with Open AI planning to launch its first device by late 2026 or early 2027, which will include various innovative products [8] - The report notes that the iPhone 17 series has prompted Apple to increase production capacity, reflecting strong market demand and potential for supply chain growth [8] - Meta's recent launch of multiple AI & AR glasses indicates a growing market for AR technology, suggesting a shift from accessory to independent smart terminal status for wearable devices [8]

Core Insights - The consumer electronics industry is undergoing a significant transformation driven by artificial intelligence (AI) technology, with a projected market size of $1.29 trillion by 2025, where emerging markets contribute 40% of growth [2] - Traditional product categories are experiencing slow growth, while new categories like AI hardware, XR devices, and smart home products are growing at over 25% annually, indicating a strategic shift from scale expansion to quality enhancement [2] Market Performance - The domestic smartphone market in China is expected to see a shipment volume of 141 million units in the first half of 2025, a year-on-year decline of 3.9% due to a lack of breakthrough features [3] - The global PC market is also under pressure, with a 7.2% year-on-year decline in shipments in Q2 2025, marking three consecutive quarters of decline [4] Emerging Opportunities - Despite challenges, new growth opportunities are emerging. The smart glasses market in China is projected to ship 494,000 units in Q1 2025, a 116.1% increase year-on-year, with total annual shipments expected to reach 2.907 million units, up 121.1% [5] - AR/VR device shipments are anticipated to reach 742,000 units, reflecting a year-on-year growth of 38.1% [5] Technological Innovations - Key technological breakthroughs are reshaping the consumer electronics landscape, including: 1. Maturity of AI large models redefining human-computer interaction [9] 2. Advances in display technology enhancing the practicality of XR devices [9] 3. Semiconductor advancements increasing mobile device computing power by over 30% annually [9] AI Hardware Revolution - 2025 is being referred to as the "AI hardware year," with various companies launching AI-native devices that interact with users through natural language rather than traditional app ecosystems [10] - Major chip manufacturers are integrating dedicated AI processing units into their SoCs, with Qualcomm's Snapdragon 8 Gen 3 achieving 45 TOPS of AI performance and MediaTek's Dimensity 9300 reaching 50 TOPS [11] Competitive Landscape - Major companies are adjusting strategies to build competitive advantages, with Apple focusing on integrated hardware-software ecosystems and Huawei leveraging its HarmonyOS for seamless device collaboration [13] - Emerging companies are also finding success through differentiation in niche markets, demonstrating that precise positioning and innovation can lead to market success [14] Investment Opportunities - Investment focus areas include companies specializing in innovative categories such as XR, smart home, and wearable devices, with significant growth potential in these sectors [17] - The consumer electronics industry is cyclical, currently in a recovery phase after inventory adjustments, with high-end markets expected to recover faster than mid-to-low-end markets [18] Future Trends - The demand for AI computing power is driving upgrades in chips, with a focus on SoC, storage, and sensor segments [19] - New interaction methods such as voice recognition and gesture control are becoming increasingly important in the consumer electronics space [19] Conclusion - The consumer electronics industry is experiencing profound changes, with AI technology redefining product forms and human-computer interaction. Companies that can accurately grasp technological trends and continuously innovate will stand out in the new cycle [21]

Core Viewpoint - MediaTek has launched its flagship 5G AI chip, Dimensity 9500, which significantly enhances on-device AI capabilities, moving from experimentation to practical applications [2][12]. Group 1: AI Capabilities and Performance - The Dimensity 9500 can process long texts up to 128K characters in just two seconds, summarizing meeting notes and correcting typos automatically [3]. - Image generation on mobile devices has improved, with the Dimensity 9500 producing detailed images in 10 seconds, compared to 30 seconds with previous models [7]. - The chip supports 4K quality image generation, allowing users to create images from simple prompts in under 10 seconds [9]. - The AI applications running on the Dimensity 9500 are designed for real-world scenarios, operating locally without cloud data uploads, and consuming 50% less power than its predecessor, the Dimensity 9400 [11]. Group 2: Technological Advancements - The Dimensity 9500 features a new architecture built on a 3nm process, integrating over 30 billion transistors, resulting in a 111% increase in NPU peak performance while reducing power consumption by 56% [18][22]. - It achieved a score of 15015 on the AI Benchmark platform, nearly doubling the performance of the previous generation [19]. - The chip employs a dual NPU architecture, enhancing both performance and efficiency, and introduces a new BitNet 1.58-bit quantization framework, reducing power consumption by 50% compared to the previous model [25][28]. Group 3: Developer Support and Ecosystem - MediaTek has introduced the Dimensity AI development kit, which supports key technologies for AI model development, enabling the execution of 7 billion parameter AI models on-device [30][33]. - The company is focused on providing a standardized AI development paradigm, which is expanding the ecosystem for native AI applications [33]. Group 4: Industry Trends and Future Outlook - Major smartphone manufacturers like vivo and OPPO are set to launch devices powered by the Dimensity 9500, showcasing a shift towards advanced AI capabilities in mobile technology [36]. - The upcoming devices will feature personalized AI functionalities and enhanced performance for complex tasks, indicating a trend towards more intelligent and responsive mobile devices [39][40].

Core Viewpoint - DJI's rise in machine vision technology has transformed consumer drones from niche products to widely used devices, showcasing significant advancements in autonomous capabilities and AI integration [3][28]. Group 1: Development of Visual Tracking Technology - In 2014, DJI's founder Frank envisioned giving drones "eyes" and "brains" for autonomous navigation and obstacle avoidance, which seemed nearly impossible at the time [3][4]. - A small team led by Zhao Cong faced numerous challenges, including a lack of resources and internal skepticism, as they aimed to develop visual tracking capabilities that surpassed GPS limitations [7][11]. - The breakthrough came in mid-2015 when the team demonstrated successful autonomous tracking, leading to the inclusion of this technology in the Phantom 4 development plan [14][15]. Group 2: Launch of Phantom 4 - The Phantom 4 was officially launched in March 2016, featuring advanced visual tracking and obstacle avoidance, which captivated audiences and set a new standard in the industry [25][28]. - The product achieved rapid sales, exceeding one million units and generating over 10 billion RMB in revenue, solidifying DJI's position as a leader in the drone market [27][28]. Group 3: Advancements in AI and Chip Technology - The Mavic Pro, launched in 2016, marked a significant leap in integrating AI capabilities into consumer drones, showcasing real-time processing on limited hardware [36][37]. - The team optimized algorithms to run complex AI models on a chip with only 1.2GFlops of processing power, enabling features like gesture control and precise landing [37][39]. - Mavic Pro's success not only redefined consumer expectations but also prompted chip manufacturers to recognize the potential for AI applications in consumer hardware [38]. Group 4: Continuous Innovation and Market Expansion - Following the success of Phantom 4 and Mavic Pro, DJI continued to innovate with products like Spark and Mavic Air, which further simplified user interaction and expanded the application of drones [57][58]. - The Mavic Air introduced full gesture control, allowing users to operate the drone without a remote, demonstrating a significant advancement in autonomous interaction [50][52]. - DJI's visual navigation team played a crucial role in the company's transformation from a "flying camera" to a "flying robot," leading to breakthroughs across various product lines [59]. Group 5: Future Directions and Challenges - In 2019, Zhao Cong left DJI to address the challenges of AI commercialization across industries, founding a new company focused on scalable AI solutions [61][62]. - The new venture aims to bridge the gap between AI demand and talent shortages, emphasizing the need for innovative approaches to AI deployment in various sectors [61][63].

Core Insights - The application of artificial intelligence (AI) is penetrating various industries and is set to reshape the cloud ecosystem, entering a new era of Agent AI systems that can operate autonomously and respond in real-time [1][3] - The demand for computing power and memory bandwidth is experiencing exponential growth as AI model parameters scale from millions to trillions, posing significant challenges in power consumption and heat dissipation, especially for mobile devices [3][4] - Arm has launched the new Arm Lumex CSS platform, designed to accelerate AI experiences in flagship smartphones and next-generation personal computers, achieving up to five times AI performance improvement [3][5] Industry Transformation - AI is driving a deep transformation in global industry structures, with traditional data centers evolving into "AI data centers" to optimize model training and large-scale inference [4] - The automotive sector is transitioning from "software-defined" to "AI-defined," enhancing perception, automation, and personalized experiences [4] - Smart devices like smartphones and PCs are becoming "AI companions" or "AI workstations," integrating into users' thinking, work, and creative processes [4] Product Development - Arm is continuously upgrading its product ecosystem to meet evolving AI computing demands, focusing on a "platform-first" strategy that integrates high-performance CPU/GPU hardware, software tools, and libraries [4][5] - The new Arm Lumex CSS platform incorporates the second-generation Scalable Matrix Extension (SME2) technology, significantly enhancing device responsiveness and operational efficiency [5][6] - By 2030, SME and SME2 technologies are expected to add over 10 billion TOPS of computing power to more than 3 billion devices, leading to an exponential increase in edge AI performance [6] Market Opportunities in China - China possesses a vast market size and diverse application scenarios for AI, maintaining pace with global technological innovations and demonstrating strong commercialization capabilities [7] - The AI smartphone market in China is projected to reach 118 million units by 2025, with a year-on-year growth of 59.8%, capturing 40.7% of the overall market [7][8] - Arm is deepening collaborations with leading manufacturers like vivo to accelerate the deployment of edge AI applications and enhance user experiences [7][8] Future Outlook - Despite a slowdown in smartphone industry growth, with a projected 1.0% year-on-year increase in global smartphone shipments by Q2 2025, there is confidence in the ongoing partnerships with Chinese manufacturers [8] - The increasing consumer spending on advanced smartphone features reflects a growing recognition of the value of computing devices in daily life, driven by AI [8] - Arm aims to continue close cooperation with Chinese ecosystem partners to support and expand AI technology deployment across various sectors, including smart terminals, PCs, infrastructure, automotive, and IoT [8]

Core Viewpoint - Lixun Precision has triggered a surge in the consumer electronics sector, driven by rumors of collaboration with OpenAI and increased production demands from Apple for the iPhone 17 [1][5]. Group 1: Market Performance - Lixun Precision experienced a rare limit-up in early trading, with the maximum order amount exceeding 4.6 billion yuan [2]. - Apple concept stocks collectively surged, with notable increases such as Hongfu Chao up 20% and Zhixin Precision up over 12% [4]. Group 2: Key Developments - Reports indicate that OpenAI has signed an agreement with Lixun Precision to develop a consumer-grade device, currently in the prototype stage, expected to collaborate closely with OpenAI's AI models [5]. - Following strong pre-sale performance of the iPhone 17, Apple has reportedly requested Lixun Precision and Foxconn to increase the daily production of the standard iPhone 17 by approximately 40% [5]. Group 3: Future Trends - Analysts suggest that the consumer electronics sector may be entering a new trend, with OpenAI potentially developing a series of edge AI devices expected to launch between late 2026 and early 2027 [6]. - The hardware aspect is seen as a crucial part of OpenAI's strategy, with Chinese suppliers playing a significant role in global production resources [6]. - Galaxy Securities anticipates that the iPhone 17 series sales will exceed previous expectations, with a new innovation cycle for Apple expected to begin in 2025 [7].

Group 1 - OpenAI is developing a new consumer-oriented device, currently in the prototype stage, with potential involvement from companies like Luxshare Precision, GoerTek, and Lens Technology [1][2] - The pre-sale performance of the iPhone 17 standard version has exceeded expectations, prompting Apple to increase the daily production of this model by approximately 40% and notify some non-electronic component suppliers to raise their daily supply by about 30% [1][3] - The iPhone 17 standard version is expected to account for 25% of total production in early 2025, with the Pro series at 65% and the iPhone Air at 10%, indicating a consumer preference for lower-priced models due to the high cost-performance ratio of the standard version [3] Group 2 - OpenAI's entry into AI hardware may leverage Apple's supply chain, with plans to release a range of AI devices, including a "screenless smart speaker," by late 2026 or early 2027 [2] - The iPhone 17 standard version features high-end specifications, such as ProMotion high refresh rate display and advanced camera systems, enhancing its value proposition at a price of 5999 yuan [3] - The anticipated product innovation cycle from Apple in 2026-2027, including foldable phones and AI glasses, is expected to drive growth for the supply chain [4]

Summary of OpenAI's Hardware Ambitions and Market Dynamics Company Overview - **Company**: OpenAI - **Industry**: Artificial Intelligence and Hardware Key Points and Arguments OpenAI's Hardware Strategy - OpenAI is actively expanding into the hardware sector, aiming to launch hardware products by 2026, with projected sales of 100 million units and revenue exceeding $10 billion by 2027-2028, contingent on overcoming technical challenges related to integrating large models with end devices [1][4] - The company plans to achieve profitability by 2029, targeting $100 billion in revenue from various sources, including ChatGPT subscriptions, APIs, and other products [1][4] Revenue Projections - By the end of 2025, OpenAI anticipates that revenue from products other than APIs will surpass $2 billion, indicating a shift in revenue sources [4] - Hardware products are expected to become a significant growth driver, with a launch planned for 2026 [1][4] Subscription Model - OpenAI's core subscription model revolves around ChatGPT, featuring a multi-tiered payment system that includes various service levels [5] - The integration of self-developed chips and devices is intended to create a closed-loop business model, enhancing user engagement and monetization [5][6] Hardware Development Plans - OpenAI is developing a range of hardware products, including headphones, glasses, watches, and smart home devices, designed to deliver core AI subscription functionalities [8][10] - The first product is expected to be a pocket-sized companion assistant, emphasizing portability and environmental awareness [10] Talent Acquisition and Production Challenges - To address its lack of hardware production experience, OpenAI has recruited over 20 hardware experts from Apple and acquired the IO company, which was founded by former Apple employees [9] - This strategy aims to enhance the development of consumer-grade products that will coexist with existing devices like smartphones and Macs [9] Market Opportunities and Challenges - The smart terminal market poses challenges for traditional internet and software companies, but OpenAI's collaboration with Jony Ive and reliance on Chinese suppliers may provide a competitive edge [3] - Domestic companies such as Luxshare Precision, Goertek, and others are expected to benefit from the anticipated 100 million unit shipment, leading to increased revenue and profit growth [3][15] AI Deployment and User Engagement - OpenAI plans to deploy AI capabilities at the edge to gather user data, enabling personalized applications while reducing computational costs [11] - The strategy includes rapidly launching low-cost hardware to build user engagement and facilitate data collection for model iteration [11][13] Competitive Landscape - OpenAI competes with major players like Google and Meta in the C-end market, where it has quickly gained traction with nearly 1 billion monthly active users [26] - The competition between hardware and software companies is intensifying, with both sectors seeking to integrate their offerings [27][28] Future Trends and Considerations - The development of edge AI will significantly impact domestic SOC manufacturers, necessitating advancements in computational power, communication, and storage solutions [16][24] - Companies like Rockchip and Amlogic are positioning themselves to capitalize on the growing demand for edge AI solutions, with projections for substantial revenue growth in the coming years [18][19] Investment Insights - Investors should monitor upcoming quarterly earnings reports and developments in AI from major tech companies, particularly focusing on Google's comprehensive strategy and Apple's efforts to enhance its model capabilities [39] Additional Important Content - OpenAI's new hardware products are designed to be discreet and user-friendly, allowing for natural interaction without screens [10] - The integration of AI into everyday life is expected to deepen, with models capable of processing vast amounts of personal data to enhance user experiences [6][7]

据The Information9月19日消息，近日，OpenAI已与苹果公司的设备组装商立讯精密签署协议，共同打 造一款消费级设备。据悉，该款设备目前仍处于原型开发阶段，预计能与OpenAI的人工智能模型深度 协作。 值得注意的是，立讯精密可能并非唯一参与其中的苹果供应商。同样为苹果代工AirPods、HomePod及 Apple Watch组件的歌尔股份，或被邀请为OpenAI的首款消费级硬件产品提供扬声器模块。 长城证券则认为，AI与物联网的结合使得下游行业能够提供更加个性化的服务和体验。Counterpoint的 数据显示，到2030年，AI嵌入式蜂窝模组预计将占所有物联网模组出货量的25%，复合年增长率为 35%。ABIResearch预测，至2029年蜂窝模组全球出货量可达7.05亿片，收入可达92.39亿美元。在端侧 AI等新技术趋势下，构建多元化市场布局、提升本地化服务能力，将成为决定企业能否在全球竞争中 站稳脚跟的关键。 公司方面，除立讯精密、歌尔股份外，华泰证券表示SoC芯片恒玄科技、瑞芯微、晶晨股份、乐鑫科技 等也是智能硬件核心。 *免责声明：文章内容仅供参考，不构成投资建议 *风险提示 ...