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 semiengineering 。 过去十年，数据中心处理器市场经历了两次重大转变。 过去，所有数据中心的计算都是 x86 架构，其中超过 90% 都来自英特尔。GPU 于 2016 年首次出现在数据中心（Pascal GPU）。 现在，大多数计算都在 GPU 上完成。AMD 正努力在 x86 市场超越英特尔，而基于 Arm 的 CPU 则在 NVIDIA 和超大规 模厂商的推动下快速增长。 AMD 正在数据中心 CPU 领域取代英特尔 AMD 正在迅速抢占英特尔在 x86 数据中心处理器领域的份额。 如上图所示，AMD 的份额目前为 40%，而 2018 年几乎为零。这是因为： Arm 高级副总裁兼数据中心基础设施总经理 Mohamed Awad 在 3 月份表示，到 2024 年底，基于 Arm 的 CPU 在数据中 心的份额约为 15%。但他预计，到 2025 年底，Arm 的 CPU 份额将增长到接近 50%——这大概是出货量份额。（下文将 介绍 Arm 的份额可能达到 50%，但要到 2020 年后期）。 这是因为另外两个重大转变。 ...

嘉宾：亚马逊全球副总裁、亚马逊云科技大中华区总裁 储瑞松 "AI的发展已经来到一个拐点，我们正处在Agentic AI爆发的前夜。" 文 丨 新浪科技 周文猛 "AI Agent只是单一的某个智能体，Agentic AI 才能比较清楚地描述我们的现状。" 期数：No.76 "今年亚马逊资本支出，将主要用于云和AI基础设施建设。" "在大环境的不确定性下，出海选择对的云要考虑业务策略灵活性。" …… 这一系列的观点，是亚马逊全球副总裁、亚马逊云科技大中华区总裁储瑞松在与新浪科技《科创100 人》独家沟通中，对当前AI发展及企业出海的洞见。 在国内热衷于谈论AI Agent的当下，作为亚马逊云科技大中华区一把手的他却有着更多的克制与理性， 认为Agentic AI 才更能够描述我们当前所处的AI发展环境。 但克制的背后，2025年亚马逊将投资1000亿美元用于云和AI基础设施建设的豪气，也在用实际行动， 为其带领的亚马逊云科技在中国乃至于全球市场发展提供着充足的弹药。 伴随大模型能力的快速提升和使用成本急剧下降，被模型上下文协议（MCP）极大简化了交互及协作 难度的智能体将如何重塑生产力？而储瑞松带领的亚马逊云科 ...

Investment Rating - The report maintains an investment rating of "Outperform the Market" for the telecommunications industry [1]. Core Insights - The NTN (Non-Terrestrial Network) industry chain is maturing rapidly, with ground infrastructure poised for significant development. The MWC Shanghai 2025 highlighted satellite internet as a key focus, presenting at least three catalysts and investment opportunities for the sector [2][15]. - The next-generation low Earth orbit (LEO) constellation design is nearly mature, with domestic projects like Qianfan, State Grid, Hongyan, and Tianqi starting to form. The NTN architecture is becoming mainstream, benefiting major 5G equipment manufacturers such as Huawei, ZTE, and others [3][16]. - As the trial commercial use of low Earth orbit constellations approaches, ground infrastructure will see accelerated growth. The low Earth satellite communication system in China is expected to enter public testing this year, with consumer-level satellite networking anticipated in the second half of the year [4][17]. Summary by Sections 1. Weekly Viewpoint and Investment Suggestions - The NTN industry is expected to benefit from the upcoming commercial trials of low Earth orbit satellites, with significant opportunities in ground infrastructure and satellite payloads [15][17]. - Key companies to watch include: - Low Earth Satellite Payload: Xinke Mobile, Fenghuo Communication, ZTE, Tongyu Communication, Jingxin Communication, and others [20]. - Low Earth Ground Infrastructure: Changjiang Communication, Zhenyou Technology, Xinke Mobile, China Tower, and others [20]. - Super Node Communication: Zhongji Xuchuang, Xinyi Sheng, Wolong Nuclear Materials, and others [20]. 2. Market Overview - The telecommunications index rose by 1.58% during the week of June 16-20, 2025, while major indices like the Shanghai Composite Index and Shenzhen Component Index saw declines [8][18]. - The top-performing sectors included optical modules (+9.6%), optical cables (+8.2%), and liquid cooling (+3.8%) [8][18]. 3. Company Performance - Notable stock performances included Longfly Fiber (+15.10%), Yuanjie Technology (+12.06%), and others leading in gains, while Zhenyou Technology and others faced significant declines [34][36].

Core Insights - The Amazon Cloud Technology China Summit highlighted the emergence of Agentic AI as a focal point for innovation and business transformation in the current uncertain era [3][4] - Amazon Cloud Technology aims to assist Chinese enterprises in expanding globally while leveraging local cloud services to drive business growth and AI innovation [4][11] Group 1: Agentic AI and Business Transformation - The development of AI has reached a turning point, with Agentic AI poised to significantly enhance customer experience, innovate business models, and improve operational efficiency [3][6] - Companies must prepare both management and technology aspects to seize the opportunities presented by the Agentic AI revolution [3][7] - Agentic AI is seen as a key engine for enterprise transformation, enhancing employee productivity and driving business model innovation [6][12] Group 2: Strategic Framework and Implementation - Companies should establish a clear cognitive framework and top-level planning while optimizing organizational processes and upgrading talent structures [7] - Four foundational pillars are essential for companies: security compliance, system resilience, architectural scalability, and technological foresight [7] - A pragmatic strategy for implementation is crucial, including setting realistic expectations and building a robust partner ecosystem [7] Group 3: Infrastructure and Technological Advancements - Amazon Cloud Technology has made significant investments in infrastructure, including the Graviton4 processor, which improves database application performance by 40% and large Java application performance by 45% [8][10] - The company has built a global infrastructure network covering 245 countries and regions, offering over 240 full-stack cloud services [10] - Amazon Cloud Technology provides a leading pre-trained model library and a comprehensive development toolchain to lower the barriers to AI innovation [10] Group 4: Globalization and Local Innovation - Amazon Cloud Technology's "three horizontal and one vertical" service architecture supports Chinese enterprises in navigating compliance risks and technological pressures in global markets [11] - The newly released Agentic AI practice guide offers a comprehensive methodology to help enterprises overcome AI application development bottlenecks [11][12] - The combination of technological empowerment and strategic consulting is driving the evolution of China's AI innovation ecosystem towards greater resilience and sustainability [12]

Market Overview - As of June 18, US stock index futures showed mixed results, with Dow futures down 0.02%, S&P 500 futures up 0.05%, and Nasdaq futures up 0.15% [1] - European indices also experienced declines, with Germany's DAX down 0.52%, UK's FTSE 100 down 0.06%, France's CAC 40 down 0.30%, and the Euro Stoxx 50 down 0.38% [2][3] Commodity Prices - WTI crude oil increased by 0.41% to $73.57 per barrel, while Brent crude oil rose by 0.46% to $76.80 per barrel [3][4] Federal Reserve and Economic Outlook - The Federal Reserve is expected to maintain interest rates during the upcoming meeting, awaiting clearer insights into the economic impact of government policy changes [5] - Market expectations suggest at least one rate cut by the Fed this year, with a focus on the potential for a 25 basis point reduction [6][7] Legislative Developments - The US Senate passed a stablecoin bill with a vote of 68 to 30, establishing regulatory rules for cryptocurrencies pegged to the dollar, marking a significant victory for the crypto industry and former President Trump [8] Corporate Developments - JPMorgan Chase is launching a pilot project for a digital deposit token named JPMD on Coinbase, indicating a significant move into the blockchain space by traditional financial institutions [9] - Disney and Amazon have formed a partnership to enhance targeted advertising capabilities for streaming services, allowing more advertisers to access Disney's content inventory [10] - Amazon's AWS is set to update its Graviton4 chip, aiming to compete with Nvidia in the AI infrastructure market [11] - Amazon CEO Andy Jassy warned that generative AI could lead to a reduction in the company's workforce, as automation reshapes job structures [12]

Group 1 - Amazon Web Services (AWS) is set to announce an upgrade to its Graviton 4 chip, increasing network bandwidth to 600 Gbps, which AWS claims is the highest specification in the public cloud [2] - Graviton 4, designed by Amazon's Annapurna Labs, is part of its custom chip strategy aimed at competing with traditional semiconductor giants like Intel and AMD [2] - The real competition lies in the artificial intelligence infrastructure sector, where AWS is directly challenging Nvidia [2] Group 2 - AWS has invested $8 billion in Project Rainier, an AI supercomputer built for the startup Anthropic, which utilizes over 500,000 Trainium chips [3] - Although Nvidia's Blackwell chip outperforms Trainium 2, AWS claims its chips offer better cost-performance ratios [3] - AWS's supply capacity is strong, but demand for these chips currently exceeds supply, indicating a robust market interest [3] Group 3 - With the upcoming Graviton 4 upgrade and the Trainium chips used in Project Rainier, AWS aims to control the entire technology stack of AI infrastructure, from network architecture to training and inference [3] - The success of mainstream AI models like Claude 4 being trained on non-Nvidia chips raises the question of how much market share AWS can capture from Nvidia [3] - The release schedule for the Graviton 4 upgrade will be announced by the end of June [4]

Core Insights - Amazon's AWS is set to update its Graviton4 chip, achieving a network bandwidth of 600 Gbps, the highest in public cloud [1] - The Graviton4 chip is part of Amazon's custom strategy to compete with traditional semiconductor companies like Intel and AMD [1] - The real competition lies in the AI infrastructure sector, particularly against Nvidia [1] Group 1: AWS Developments - AWS has invested $8 billion to support the AI supercomputer Project Rainier for startup Anthropic [1][2] - The Claude Opus 4 AI model from Anthropic is launched on the Trainium2 GPU, with Project Rainier powered by over 500,000 chips [2] - AWS aims to reduce AI training costs and provide alternatives to Nvidia's expensive GPUs [1][2] Group 2: Chip Performance and Market Position - AWS claims that while Nvidia's Blackwell chip outperforms Trainium2, AWS chips offer better cost-performance [2] - Trainium3 is expected to launch this year, doubling the performance of Trainium2 while saving 50% energy [2] - Demand for these chips has exceeded supply, indicating strong market interest [2] Group 3: Competitive Landscape - AWS is demonstrating its ambition to control the entire AI infrastructure stack, from networking to training and inference [2] - The success of mainstream AI models on non-Nvidia hardware raises questions about AWS's potential market share [2] - The timeline for the Graviton4 update announcement is expected by the end of June [2]

This is where Amazon Web Services chips are born. I'm Christina Partzine at Annaperna Labs in Austin, Texas, a lab owned by AWS. And today they're announcing that their Graviton 4 CPU chip will offer 600 gigabytes per second of networking bandwidth, which would be potentially the largest in the public cloud.If you're confused what that means, an executive told me it's like running 100 CDs per second. That's how quickly data is being transferred. and really shows that their custom chip chip strategy is uh ga ...

>> Amazon Web Services unveiling a new chip intended to take on Nvidia. Of course, in those cloud wars. Kristina Partsinevelos joins us now.She's in Austin, Texas and has more for us this morning. Christina. >> David, this is where AWS chips are born. This is Annapurna Labs.It's owned by AWS. And like you said today they are announcing their new graviton four chip will offer 600GB per second of networking bandwidth. So this is a CPU and really shows how their custom chip strategy is competing against, you k ...

Amazon's AWS unveiling a new chip to take on Nvidia in the cloud wars and Christina Partzene is live uh in Austin, Texas. Good morning. Good morning.This is where Amazon Web Services chips are born. This is Anna Pernal Labs. It used to be once very top secret, but now they're letting you know media journalists like me inside.And today they're announcing that their latest Graviton 4 chip, that would be their CPU, is now offering 600 gigabytes per second of networking bandwidth, which would be potentially the ...