公众号记得加星标⭐️，第一时间看推送不会错过。 来 源 ： 内容来自半导体行业观察综合 。 据报道，谷歌母公司Alphabet (正与Meta Platforms 等公司洽谈，希望它们能使用谷歌的Tensor AI 芯片，此举将加剧其与英伟达的竞争。谷歌及其AI芯片合作伙伴博通股价尾盘上涨，而英伟达和 AMD股价则下跌。 谷歌传统上将客户使用的张量处理单元（TPU）用于自己的数据中心，然后出租给客户。但据The Information周一晚间报道，谷歌现在开始向客户出售TPU，供其在自己的数据中心使用。 报 道 指 出 ， Meta Platforms 正 在 考 虑 从 2027 年 开 始 在 其 数 据 中 心 购 买 价 值 数 十 亿 美 元 的 谷 歌 TPU，同时最早从 2026 年就开始从谷歌云租用 TPU 容量。Meta 一直以来主要依靠英伟达图形处理 器 (GPU) 来满足其人工智能需求。 对于谷歌和博通（它们参与了Tensor AI芯片的设计）来说，这可能是一个巨大的新市场。但它也可 能对英伟达和AMD构成重大竞争，威胁到它们巨大的销售和定价权。 受The Information报道的影 ...

Core Insights - Google’s TPU v7 accelerators demonstrate significant performance improvements, with Ironwood being the most powerful TPU to date, achieving 10 times the performance of TPU v5p and 4 times that of TPU v6e [4][11] - The TPU v7 offers competitive performance against Nvidia's Blackwell GPUs, with Ironwood providing 4.6 petaFLOPS of dense FP8 performance, slightly surpassing Nvidia's B200 [3][4] - Google’s unique scaling approach allows for the connection of up to 9216 TPU chips, enabling massive computational capabilities and high bandwidth memory sharing [7][8] Performance Comparison - Ironwood TPU has a performance of 4.6 petaFLOPS, compared to Nvidia's B200 at 4.5 petaFLOPS and the more powerful GB200 and GB300 at 5 petaFLOPS [3] - Each Ironwood module can connect up to 9216 chips with a total bidirectional bandwidth of 9.6 Tbps, allowing for efficient data sharing [7][8] Architectural Innovations - Google employs a unique 3D toroidal topology for chip interconnects, which reduces latency compared to traditional high-performance packet switches used by competitors [8][9] - The optical circuit switching (OCS) technology enhances fault tolerance and allows for dynamic reconfiguration in case of component failures [9][10] Processor Development - In addition to TPU, Google is deploying its first general-purpose processor, Axion, based on the Armv9 architecture, aimed at improving performance and energy efficiency [11][12] - Axion is designed to handle various tasks such as data ingestion and application logic, complementing the TPU's role in AI model execution [12] Software Integration - Google emphasizes the importance of software tools in maximizing hardware performance, integrating Ironwood and Axion into an AI supercomputing system [14] - The introduction of intelligent scheduling and load balancing through software enhancements aims to optimize TPU utilization and reduce operational costs [14][15] Competitive Landscape - Google’s advancements in TPU technology are attracting attention from major model builders, including Anthropic, which plans to utilize a significant number of TPUs for its next-generation models [16][17] - The competition between Google and Nvidia is intensifying, with both companies focusing on enhancing their hardware capabilities and software ecosystems to maintain market leadership [17]

Core Insights - The article highlights NVIDIA's advancements in AI technology and its collaborations with major companies like Uber, Palantir, Amazon, and Microsoft, emphasizing the significance of domestic manufacturing in the U.S. [1][2] - NVIDIA's CEO Jensen Huang showcased the capabilities of the Blackwell GPU, which has seen substantial demand, with 6 million units shipped and 14 million units ordered, translating to potential sales of $500 billion [1][2] - Huang's remarks reflect a strategic push for U.S. manufacturing and a desire to reduce reliance on foreign products in the AI sector [2][3] Group 1 - Huang praised the Blackwell GPU's computational power and highlighted the integration of 72 GPUs in a single server rack, weighing 3,000 pounds and costing millions [1] - The company has begun mass production of Blackwell chips in Arizona, although not all production processes are completed in the U.S. [1][2] - Huang expressed concerns about the U.S. potentially losing its market in AI technology due to reliance on foreign products and called for solutions from government officials [2][3] Group 2 - NVIDIA's partnerships include collaborations with Lucid Motors for autonomous driving and Eli Lilly for supercomputing in drug development [3] - The company is investing $1 billion in Nokia to integrate AI into 6G wireless networks, showcasing its commitment to enhancing energy efficiency in data centers [3][4] - Huang indicated the intention to hold annual AI conferences in Washington, reflecting a growing influence in the tech policy landscape [5]

Group 1: Market Developments - The first batch of three newly registered companies in the Sci-Tech Innovation Board's growth layer officially listed today, marking a significant step in the establishment of this new segment just over four months after the announcement by the China Securities Regulatory Commission [1] - The three companies include two high-tech enterprises in the biopharmaceutical sector and one in the semiconductor materials field, all of which are currently unprofitable [1] Group 2: Stock Market Performance - The "Magnificent 7" index of major U.S. tech stocks rose by 2.40%, reaching a record high of 208.95 points, with a cumulative increase of 4.35% over the last three trading days [2] - Tesla shares increased by 4.31%, Google by 3.6%, and both Nvidia and Apple saw gains of over 2% [2][3] Group 3: Corporate News - Former Alibaba Group Vice President Peng Chao has launched a new company named "Yun Jue Technology," focusing on AI wearable devices and smart agents [5] - The first product from Yun Jue Technology will combine wearable hardware with an intelligent agent, aiming to enhance consumer experiences in high-frequency sports environments [5][6] - AMD has established a $1 billion partnership with the U.S. Department of Energy to build two supercomputers for various scientific challenges, including nuclear energy and cancer treatment [10] Group 4: Regulatory Issues - Apple has been fined €48 million (approximately $55.9 million) by a French court for unfair marketing practices related to its iPhone sales contracts with mobile operators [8] - The penalties include €8 million in fines and compensation to several mobile operators, with Bouygues Telecom receiving €16 million, Iliad €15 million, and SFR €7.7 million [8]

首台超算名为Lux，基于AMD的MI355X AI芯片打造，计划未来六个月内建成并投入使用。第二台超算Discovery将采用AMD的MI430系列芯片，预计 2028年交付，2029年投入运营。美国能源部长称，这些系统将"极大加速"核能和聚变能源、国防与国家安全技术以及药物研发等领域的进展。 媒体报道称，AMD与美国能源部达成10亿美元的人工智能（AI）合作，协议包含打造两台超级计算机，用于核能、癌症治疗及国家安全等领域的科学 研究。 美国能源部长克里斯·赖特和AMD CEO 苏姿丰向媒体透露了上述合作计划。美国正通过建造新的超算确保拥有足够的算力运行日益复杂、需要处理海 量数据的实验。 赖特表示，这些系统将"极大加速"核能和聚变能源、国防与国家安全技术以及药物研发等领域的进展。他预计AI超算的计算能力将在未来两到三年内为 掌握聚变能源提供实际路径。 首台超算计划六个月内上线 继联手英伟达打造超算后，能源部与AMD也要展开合作。这是美国政府加速关键科技领域计算能力布局的最新举措。 报道传出后，美东时间27日周一，美股早盘小幅转跌的AMD股价迅速转涨，日内涨幅曾超过1%，此后保持涨势，尾盘涨幅扩大，收涨将近2 ...

【环球时报记者 刘文璋 环球时报驻日本特约记者 潘小多 严格】编者的话：日本总务省7月发布2025年 版《信息通信白皮书》后，在这个曾以"技术立国"的国家，社会各界近来对本国人工智能（AI）研发和 应用落后于美国、中国等世界领先国家的现状展开了深刻反思。事实上，日本在AI研究方面的起步并 不晚，甚至被日媒形容为上世纪50至70年代第一次AI热潮中的研究"先驱"，发挥了"奠基作用"。那么， 为什么日本在之后的全球AI竞赛中逐渐跟不上步伐？这背后有哪些系统性制约因素？日本各界看待本 国未来AI发展的心态又如何呢？ " 数字劳动力 " 不足、法律壁垒高、资本保守 "哆啦A梦的故乡在理论上拥抱新技术，但在实践中却未能真正落实。"英国《经济学人》杂志报道称， 日本不少领域其实都有应用AI，比如建筑公司用它提高施工现场效率；便利店AI店员相继上岗；科技 投资者对AI与硬件交叉领域充满兴趣，而日企在后者上具有传统优势。但软银集团董事长、日本科技 投资者孙正义去年10月批评说，日本企业只专注于小规模系统，而没有努力打造AI巨头，这使得日本 没能抓住AI的发展机遇。 上世纪 90 年代后，日本 AI 研究进入空白期 根据日本文 ...

英伟达将与日本理研将共同开发AI用图形处理器（GPU）。这是海外企业首次参与日本核心超级计算 机的开发。将凝聚日美的最先进技术，力争实现性能达到全球最高水平的超级计算机…… 在22日的记者会上，理化学研究所计算科学研究中心主任松冈聪表示："将与富士通、英伟达共同设计 AI时代的全球最高水平CPU和GPU，希望通过开发富岳NEXT，在全球确立日本作为AI先进国家的稳固 地位"。 英伟达副总裁伊恩・巴克（右）在"富岳NEXT"开发体制启动纪念仪式上发言（8月22日上午，东京都中央区） 富岳的CPU内部具备提升AI性能的功能，但并未配备GPU。GPU适合进行AI计算，以美国为中心，计 算能力强大的最先进超级计算机都利用GPU提升了AI性能。富岳NEXT也将通过配备英伟达的GPU来提 高计算能力。 英伟达副总裁伊恩·巴克（Ian Buck）表示："我们将通过使用英伟达的技术来解决社会问题，让科学家 能够扩大可能性的界限。希望为实现新的科学突破贡献一份力量"。 在面向AI的计算能力方面，目标是实现设计上的性能（峰值性能）达到每秒可进行10垓（1垓为10的20 次方）次运算的"Zetta级"。在高级模拟计算方面，也会将能 ...

每经AI快讯，有投资者在投资者互动平台提问：董秘您好！贵司从今年二季度开始供应Dojo项目的PCB 产品。据我所知有两个完全不同领域的Dojo项目。一个是特斯拉的超级计算机项目，另一个是区块链游 戏引擎项目。请问贵司生产的PCB产品是用于其中哪一个领域的？或者是这西个领域都有？贵司今年 Dojo项目的PCB产品的生产量销售量分别是多少？该类产品今年能够实现销售收入是多少？相应的利润 贡献是多少？ （文章来源：每日经济新闻） 世运电路（603920.SH）8月6日在投资者互动平台表示，公司与特斯拉合作的Dojo项目，PCB产品主要 用于其超级计算机领域。关于产销经营数据，请您关注公司后续发布的相关公告。 ...

特斯拉(TSLA.O)预计将在明年某个时候实现超级计算机机房的规模化运行。 ...

Core Insights - IBM and RIKEN researchers successfully simulated quantum behaviors of various molecules by combining quantum and supercomputing technologies, paving new paths for chemical and pharmaceutical research [1][2] Group 1: Research Collaboration - IBM's "Eagle" quantum computer collaborated with Japan's "Fugaku" supercomputer to simulate the quantum states of nitrogen molecules and two iron-sulfur compounds using a spatial quantum dynamics algorithm [1] - The quantum computer performed core calculations while the supercomputer acted as a "quality inspector," correcting errors in real-time [1] Group 2: Performance and Future Prospects - Although the hybrid model has not yet surpassed the independent performance of supercomputers, it has demonstrated competitive capabilities in computational chemistry [2] - Kenneth M. Moritz from Cleveland Clinic noted that the improved spatial quantum dynamics algorithm can simulate molecular behavior in solution, making chemical modeling more realistic [2] - There is optimism that the combination of quantum and classical computing could show significant advantages within a year after further algorithm optimization [2] Group 3: Industry Developments - NVIDIA has developed a software platform supporting hybrid computing, while Microsoft emphasizes the transformative potential of the combination of quantum computing, supercomputing, and artificial intelligence in chemistry and materials science [2] - Caution is expressed by Markus Reiher from ETH Zurich, who highlights the need to observe the method's performance in large molecule simulations before it can be considered the preferred approach for quantum chemical calculations [2] Group 4: Ongoing Improvements - The RIKEN laboratory has upgraded to a new IBM quantum processor with lower error rates, and the research team is working on optimizing the spatial quantum dynamics algorithm and enhancing the collaboration efficiency between "Eagle" and "Fugaku" [2]