智通财经APP获悉，东方证券发布研报称，由于传统横向供电模式存在不可避免的PDN压降与损耗，以 及AI算力功耗提升带来PDN损耗的同步增长，终端用户部署ASIC后的核心痛点是省电。金属软磁粉芯 是ASIC省电的必选项。未来随着DDR6标准对内存性能、能效与供电方案相较前代持续升级，电感材料 使用量有望进一步增长。随着明年海外AI巨头加速大规模部署自研ASIC芯片，芯片电感在AI服务器的 应用占比或将大幅提升。 东方证券主要观点如下： 随着海外AI巨头转向自研ASIC芯片，省电成了终端用户的核心痛点 铂科新材是气雾化制粉的工业化鼻祖，公司掌握的气雾化金属粉末制备技术相较水雾化粉末具有颗粒度 高、含氧量低适于压制成粉芯等优势，公司生产的常规与超细金属合金软磁粉末粒径较同业产品更细的 同时具备更低的含氧量，因此具备更强的产品竞争优势。产能方面，公司募投项目有望于2025年底前完 全建成，并于2026年开始产能爬坡，总产能有望达3亿片。随着产线效率提升，公司电感材料产能中期 或有进一步增长的空间。 风险提示 随着AI算力发展，以英伟达GB300为代表的AI算力芯片参数及性能达到全新高度的同时，功耗同样大幅 增长。由于 ...

AI 发展为何离不开金属软磁粉芯 ——AI 新材料行业深度 1 核心观点 投资建议与投资标的 有色、钢铁行业 行业研究 | 深度报告 | 迎接金铜非线性变化的新时代：——有色 | 2025-09-29 | | --- | --- | | 钢铁行业周观点（2025 年第 39 周） | | | 降息博弈已落地，有色钢铁再出发：—— | 2025-09-22 | | 有色钢铁行业周观点（2025 年第 38 周） | | | 关注低风险高分红的有色钢铁子版块：— | 2025-09-18 | | —有色钢铁行业周观点（2025 年第 37 | | | 周） | | 有关分析师的申明，见本报告最后部分。其他重要信息披露见分析师申明之后部分，或请与您的投资代表联系。并请阅读本证券研究报告最后一页的免责申明。 ⚫ 随着海外 AI 巨头转向自研 ASIC 芯片，省电成了终端用户的核心痛点。随着 AI 算 力发展，以英伟达 GB300 为代表的 AI 算力芯片参数及性能达到全新高度的同时， 功耗同样大幅增长。由于 ASIC（Application-Specific Integrated Circuit，专用集成 电路）为单 ...

其次是，2024年中国芯片出口量达到2981.1亿块，出口金额突破1万亿元人民币大关，达到1594.991亿美 元(约合人民币1.156万亿元)，同比增长18.7%。这不仅标志着中国芯片出口迈上新台阶，更使其超越服 装、纺织品和手机等传统强项，成为中国出口额最高的单一商品。 最后是，在关税制裁的和国家大基金扶持的大背景之下，倒闭国产替代加速，像国家大基金三期首次出 手晶圆制造、光刻机等卡脖子环节。 彻底瞒不住了。 芯片半导体有望接替人形机器人、深海经济成为主力炒作的下一个超级风口。 尤其是这家业绩暴增5310%，300家机构重仓、且在硅光芯片、服务器、数据中心掌握了核心技术的公 司，值得大家重点关注。 因为它接下来很有可能成为下一个十倍妖王。 不开玩笑，而是有事实依据在支撑。 首先是，芯片半导体经过两年的下行周期，现在已经在逐步复苏，尤其是在AI、人形机器人等前沿产 业快速发展的大背景之下，芯片半导体已经成为国民经济发展不可或缺的一部分。 可以预见的是，国产芯片在技术和需求的共振下，正在逐步构建起自主可控的半导体生态。 但大家必须要明白的是，在技术公关阶段，哪些蹭概念的只会成为过去式，只有业绩保持稳定、技术走 ...

前面几天的文章，我们写了Memory和电力，都是围绕AI基础设施，这篇文章看下OpenAI的AI基础 设施扩张，对亚洲供应链的影响，其实主要是大陆和台湾。本文的内容参考自JP Morgan最新的一 个分析。 这里再提一下笔者对Stargate项目的看法：理想很丰满，现实很骨感。 根据JPM的报告，目前已确认的7GW功率包括五个新数据中心站点，比如德克萨斯州的Abilene，还 有与Oracle、Softbank和CoreWeave的合作。其中，Oracle负责4.5GW的部分，涉及德克萨斯、新墨西 哥和中西部地区；Softbank负责1.5GW，在俄亥俄和德克萨斯；CoreWeave外包了0.4GW，总额达220 亿美元。这些项目的时间表很紧迫，大部分要在未来三年内落地。 为什么这个计划这么激进？因为CSP的需求很强劲，预计2025年增长55%，2026年再增25%，总资本 支出达到3450亿美元。OpenAI的额外需求相当于 8万多台GB200 NVL72机架（现在每周也就出1k左 右） ，机架级支出2500亿美元，如果算上数据中心建筑等全成本，可能到4000-5000亿美元。这会让 供应链更紧张，包括先进 ...

过去几年，包括市场领导者 Nvidia、英特尔和 AMD 在内的主要芯片制造商都在积极开发新的处理 器。然而，多项行业研究表明，对 AI 性能的要求可能会更高，这反过来又迫使供应商开发更多新产 品，并对系统设计提出更高的要求。 公众号记得加星标⭐️，第一时间看推送不会错过。 来源 ： 内容 编译自 designnews 。 特朗普总统最近宣布了一项雄心勃勃的计划，旨在使美国成为人工智能和机器学习领域的强大力量。 他取消了对竞相开发未来技术以处理大型语言模型和未来人工智能形式的美国企业的种种限制。美国 企业将摆脱繁琐的州级法规，也不再受到联邦层面的护栏和其他限制的束缚。这一消息虽然对美国电 子行业来说令人振奋，但也给下一代处理器的开发带来了更大的挑战。 虽然对创新处理器的需求表明人工智能芯片市场前景广阔，但考虑到市场将会成熟和整合，新参与者 的窗口可能不会永远敞开。 增长正在放缓吗？ Omdia 最近发布了一份名为《云计算和数据中心 AI 处理器预测》的报告，预测尽管 AI 数据中心芯 片市场将继续快速增长，但增长速度已开始放缓。Omdia 表示， 2022 年至 2024 年间， AI 基础设 施的年增长率超 ...

Core Viewpoint - The company is focusing on differentiated, customized, and diversified display module products, with a strong emphasis on complex modules and automotive electronic terminals for future growth [1][2][4]. Group 1: Product Offerings - The company started with black and white LCD products and has expanded to color screens, automotive displays, and complex modules, with revenue contributions of 64% from color LCD modules, 22% from monochrome LCD modules, 11% from displays, and 3% from touch screens [1]. - The company provides customized services to high-quality clients across various industries, including smart home, smart finance terminals, communication equipment, industrial control, energy, healthcare, automotive electronics, and consumer electronics [1]. Group 2: Financial Performance - The company has shown steady growth, with a 5-year revenue CAGR of approximately 24% and a net profit CAGR of about 33% [2]. - The average product price increased from 9.7 to 14.6 yuan per unit from 2021 to 2024, reflecting a CAGR of 14% [2]. Group 3: Future Growth Areas - The company is focusing on complex modules and automotive electronic terminals, having made progress in developing smart serial display modules and variable light rearview mirror modules [2]. - The ASIC market is expected to grow rapidly due to the increasing demand for AI applications, with major companies like Google, Intel, and Amazon leading the way [3]. - The enterprise SSD market is projected to reach $39.6 billion globally and $9.1 billion in China by 2029, driven by trends in AI, data centers, and cloud storage [3]. Group 4: Strategic Partnerships - The company is collaborating with Tianlianxin and Xincun Technology to establish a full-chain layout for ASIC and enterprise storage modules, focusing on chip research and development, module manufacturing, and application implementation [4]. - Tianlianxin is responsible for developing ASIC chips and enterprise storage products, while Xincun Technology focuses on PCM chip design and manufacturing [4].

公司研究 黄仁勋深度访谈：解读英伟达"护城河"新维度 黄仁勋深度访谈：解读英伟达"护城河"新维度 事件：近日，英伟达（NVDA.O）创始人兼首席执行官黄仁勋受邀 出席「Bg2 Pod」双周对话节目。其间，黄仁勋就当前 AI 行业的竞 争格局、英伟达的护城河以及与 OpenAI 的合作关系等进行了深度 解读。 看好 AI 资本开支增长 在本次访谈中，黄仁勋仍对全球人工智能资本开支持续增长持乐观态 度。他指出，若想实现人工智能为全球 GDP 贡献 10 万亿美元增量价值 的目标，全球每年需投入约 5 万亿美元的资本开支，以支撑能够产生相 应价值的海量 Token。这部分价值的创造机制已不再依赖传统软件，而 是依赖于 AI 系统持续运转、处理海量数据的能力，背后亟需能够承载 庞大算力需求的"AI 工厂"作为核心基础设施。目前，全球云厂商和 科技巨头正全力抢占 AI 基础设施（AI Infra）这一战略制高点。 英伟达的护城河：比三年前"更宽" 面对近期市场对英伟达高增长的可持续性、竞争格局及"循环营收"的 核心质疑，黄仁勋表示公司定位已从"芯片公司"彻底升维至"AI 基 础设施提供商"，其长期增长建立在三大核心基石 ...

Core Viewpoint - The company reported a revenue increase driven by strong downstream demand, particularly in the IoT and consumer electronics sectors, while also improving profitability through product high-endization [2][3]. Financial Performance - In the first half of 2025, the company achieved a revenue of 1.898 billion yuan, a year-on-year increase of 14.67%, while the net profit attributable to shareholders was a loss of 245 million yuan, showing a significant reduction in losses [1][2]. - The second quarter of 2025 saw a single-quarter revenue of 988 million yuan, reflecting a year-on-year increase of 19.76% and a quarter-on-quarter increase of 8.59% [1][2]. - The overall gross margin for the company's chip products reached 23.96%, an increase of 4.85 percentage points year-on-year [2]. Market Position and Product Development - The company’s cellular baseband chip business saw a sales volume increase of over 50% and revenue growth of over 30% in the first half of 2025, with a gross profit increase of over 60% [3]. - The company successfully entered new markets such as AI toys, smart wearables, and AI glasses, which are expected to support long-term growth [3]. - The company launched the world's first RedCap+Android smart chip platform, currently in the design-in phase, with expected product launches in Q4 2025 [3]. Custom Business and Future Growth - There has been a significant increase in demand for smart wearables, edge AI, and RISC-V chips, leading to a substantial expansion of the ASIC custom service market [4]. - The company has a full order book for ASIC chip design services, with projects covering both design and mass production for leading clients [4]. - Revenue projections for 2025 to 2027 are estimated at 4.438 billion, 6.316 billion, and 8.443 billion yuan, with corresponding price-to-sales ratios of 10X, 7X, and 5X [4].

Investment Rating - The report indicates a positive investment outlook for the ASIC industry, highlighting significant growth potential driven by increasing demand for AI applications and specialized chip designs [2]. Core Insights - The report emphasizes the distinct business models of ASIC and GPU, noting that ASICs are specialized chips tightly coupled with specific downstream applications, while GPUs are general-purpose chips [3][10]. - ASICs demonstrate superior cost-effectiveness and efficiency, with notable examples such as Google's TPU v5 achieving 1.46 times the energy efficiency of NVIDIA's H200, and Amazon's Trainium2 reducing training costs by 40% compared to GPU solutions [3][15]. - The report forecasts that the global AI ASIC market could reach $125 billion by 2028, with significant contributions from major players like Broadcom and Marvell [30]. Summary by Sections 1. AI Model Inference Driving ASIC Demand - The global AI chip market is projected to reach $500 billion by 2028-2030, with AI infrastructure spending expected to hit $3-4 trillion by 2030 [8]. - ASICs are recognized for their strong specialization, offering cost and efficiency advantages over GPUs, particularly in AI applications [9][14]. 2. High Complexity of ASIC Design and Value of Service Providers - ASIC design involves complex processes requiring specialized service providers, with Broadcom and Marvell being the leading companies in this space [41][42]. - The report highlights the importance of design service providers in optimizing performance and reducing time-to-market for ASIC products [55][60]. 3. Domestic Developments: Not Just Following Trends - Domestic cloud giants like Alibaba and Baidu have made significant strides in ASIC self-research, establishing independent ecosystems rather than merely following international trends [4][30]. - The report identifies key domestic design service providers such as Chipone, Aojie Technology, and Zhaoxin, which are well-positioned to benefit from the growing demand for ASICs [41]. 4. Key Trends in Domestic ASIC Development - The report identifies PD separation and supernode architectures as two core trends in domestic ASIC development, with companies like Huawei and Haiguang leading the way [4][30]. - These trends reflect a shift towards more flexible and efficient chip designs that cater to diverse industry needs [4]. 5. Valuation of Key Companies - The report includes a valuation table for key companies in the ASIC sector, indicating strong growth prospects and market positioning for firms like Broadcom and Marvell [5].

近期，英伟达（NVDA.US）投资"出手"频繁，先是宣布50亿美元投资英特尔，随后斥资至多1000亿美元投资 OpenAI，而受此前OpenAI与甲骨文的合作，市场均在股价层面给予了积极反馈。 但市场也出现了质疑声音——称英伟达、OpenAI与甲骨文存在"收入循环"，财务数字"操作"大于实际营收。 9月25日，在播客BG2最新一期节目中，BG2主播、Altimeter Capital创始人Brad Gerstner，Altimeter Capital合伙人Clark Tang与英伟达CEO黄仁勋展开了一次对话。黄仁勋在对话中回应了当下市场的关心的问题。 此外，黄仁勋也特别解释了为什么ASIC芯片并不完全和英伟达GPU是竞争关系——因为英伟达是AI基础设施提供商， 其提供的能力范围已经不仅仅是硬件和软件层面，也包括其不断迭代的速度、规模优势带来的可靠性，以及整体能源 效率等综合因素。 因此，黄仁勋认为英伟达目前的护城河比三年前"更宽"，而Brad Gerstner甚至认为，英伟达将是史上第一家达到十万 亿美元的公司。 以下为「明亮公司」编译的访谈正文（有删节）： Brad Gerstner：Jensen，再次欢 ...