Funda AI:“我们的访谈表明,前沿实验室仍致力于在内部训练和提升模型能力,而额外的评估和修复措施可能只会减缓公开发布的速度。”
他们采访的一名员工,包括某实验室的销售主管表示:“旨在提高安全性的放缓实际上导致整体算力需求增加,而非减少。”$谷歌-A(GOOGL.US) $Meta(META.US) $微软(MSFT.US) $亚马逊(AMZN.US) $甲骨文(ORCL.US) $英伟达(NVDA.US) $AMD(AMD.US)与我们分享你的投资心得,或其他想分享的内容...
Funda AI:“我们的访谈表明,前沿实验室仍致力于在内部训练和提升模型能力,而额外的评估和修复措施可能只会减缓公开发布的速度。”
他们采访的一名员工,包括某实验室的销售主管表示:“旨在提高安全性的放缓实际上导致整体算力需求增加,而非减少。”$谷歌-A(GOOGL.US) $Meta(META.US) $微软(MSFT.US) $亚马逊(AMZN.US) $甲骨文(ORCL.US) $英伟达(NVDA.US) $AMD(AMD.US)$费城交易所 半导体 ETF - iShares(SOXX.US)
背景:随着美联储最近决定加息,美国市场在经历近期回调后显示出一些复苏迹象。AI 股票和科技 ETF 展现出一定的韧性,为半导体板块的长期积累创造了机会。
我的交易:在这次回调期间,我继续持有 SOXX ETF。SOXX 让我分散投资于英伟达、博通和 AMD 等顶级芯片制造商,从而降低单只股票的风险。
启示:如果我们对所持有的股票和 ETF 保持信心,坚持持有优质股票和 ETF 并度过波动期是值得的。
今天我在 AMD 上赚了钱 $AMD(AMD.US)
TrendForce:全球电力需求容量
> 预计全球数据中心电力需求容量将从 2025 年的 122.9 GW 增长至 2026 年的 161 GW。> AI 服务器的份额到 2027 年可能超过 40%,而通用服务器的份额预计将从历史上约 40% 下降至 2027 年的 25.6%。> 2025 年,AI 服务器约占数据中心总电力需求容量的 25%,预计 2026 年将上升至 33.4%。> AI 服务器的份额到 2027 年可能超过 40%,而通用服务器的份额预计将从历史上约 40% 下降至 2027 年的 25.6%。> 2030 年全球数据中心电力供需缺口估计为 268 GW。> 基于当前电网数据,预计到 2030 年美国数据中心电力供需缺口将超过 170 GW。$英伟达(NVDA.US) $GE Vernova(GEV.US) $AMD(AMD.US) $谷歌-A(GOOGL.US) $亚马逊(AMZN.US) $微软(MSFT.US) $戴尔科技-C(DELL.US)NBIS 将于 10 月 1 日起上调按需 GPU 费率。
H100:每台 GPU-小时 3.85 美元 → 4.50 美元
H200:每台 GPU-小时 4.50 美元 → 5.40 美元
B200:每台 GPU-小时 7.15 美元 → 8.50 美元
B300:每台 GPU-小时 7.85 美元 → 9.50 美元
AMD EPYC Genoa CPU 和内存费率也将上调。
Intel Ice Lake CPU 价格保持不变。
$AMD(AMD.US) 和 $迈威尔科技(MRVL.US) 的派息日已卖出,等待下一个派息日。我现在开始适应大幅波动。
来源:Sunrise Trader

0916 |海豚君重点关注:🐬 宏观/行业 1、香港特区发布《香港特别行政区经济和社会发展第一个五年规划(2026—2030)》,为回归后首份五年发展纲领,对接国家十五五规划,重点巩固国际金融、科创中心,加速北部都会区建设,同步完善民生配套。这份中长期规划有利于稳定外资预期,强化香港 “由治及兴” 的发展主线,短期利好港股金融、科创板块情绪,但产业落地需要持续跟踪政策配套与项目兑现节奏...
AMD 日线再次站上所有主要均线。
来源:Sunrise Trader
📢 𝗝𝗨𝗦𝗧 𝗜𝗡: $Astera Labs(ALAB.US) Astera Labs Launches New Leo Memory Solutions for Agentic AI Infrastructure
👉 𝗞𝗲𝘆 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀:➤ 𝗔𝘀𝘁𝗲𝗿𝗮 𝗟𝗮𝗯𝘀 launches new 𝗟𝗲𝗼 𝗫-𝗦𝗲𝗿𝗶𝗲𝘀 and 𝗟𝗲𝗼 𝟮 memory controllers.➤ Leo X-Series targets 𝗞𝗩-𝗰𝗮𝗰𝗵𝗲-intensive, long-context agentic AI inference workloads.➤ Leo X-Series delivers up to 𝟲𝟮% faster time-to-first-token.➤ Platform delivers up to 𝟮𝟮% more tokens per second.➤ Leo 2 doubles 𝗺𝗲𝗺𝗼𝗿𝘆 𝗯𝗮𝗻𝗱𝘄𝗶𝗱𝘁𝗵 and capacity versus previous generation.➤ New controllers support 𝗗𝗗𝗥𝟰, 𝗗𝗗𝗥𝟱, 𝗣𝗖𝗜𝗲 𝟲 and 𝗖𝗫𝗟 𝟯.𝟮.➤ Leo 2 enables memory expansion, pooling, sharing, and 𝗗𝗗𝗥𝟰 reuse.➤ Astera reports growing design wins across 𝗔𝗜 𝗹𝗮𝗯𝘀, hyperscalers, and neoclouds.➤ New Leo products are 𝘀𝗮𝗺𝗽𝗹𝗶𝗻𝗴 𝘁𝗼𝗱𝗮𝘆 with hyperscaler customers.➤ Development involved 𝗔𝗠𝗗, 𝗔𝗿𝗺, 𝗜𝗻𝘁𝗲𝗹 and major memory suppliers.👉 𝗪𝗵𝘆 𝗧𝗵𝗶𝘀 𝗠𝗮𝘁𝘁𝗲𝗿𝘀:➤ Agentic AI is rapidly increasing demand for 𝗺𝗲𝗺𝗼𝗿𝘆 𝗰𝗮𝗽𝗮𝗰𝗶𝘁𝘆 and bandwidth.➤ Faster KV-cache access can improve 𝗔𝗜 𝗶𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲 economics and responsiveness.➤ Memory pooling and reuse could reduce expensive 𝗗𝗥𝗔𝗠 over-provisioning.➤ Growing design wins could expand Astera's exposure to 𝗿𝗮𝗰𝗸-𝘀𝗰𝗮𝗹𝗲 𝗔𝗜 infrastructure.👉 𝗘𝘅𝗽𝗲𝗿𝘁 𝗦𝘁𝗮𝘁𝗲𝗺𝗲𝗻𝘁𝘀:𝗠𝗮𝘁𝘁 𝗞𝗶𝗺𝗯𝗮𝗹𝗹, Vice President and Principal Analyst at Moor Insights & Strategy:“Agentic AI adoption is outpacing the industry's ability to provision memory for it, and that gap is widening every quarter," said Matt Kimball, vice president and principal analyst at Moor Insights & Strategy. "That growth is landing at the worst possible time for memory supply, with DDR5 tight and pricing climbing, so every gigabyte already deployed has to work harder. The new Leo X-Series and Leo 2 E and P-Series Smart Memory Controllers immediately address KV-cache-intensive AI workloads, improve memory utilization, and reliably reuse previously deployed memory across both AI and general-purpose cloud infrastructure, instead of simply buying more of it."𝗧𝗵𝗮𝗱 𝗢𝗺𝘂𝗿𝗮, Senior Vice President, Compute Connectivity Group at Astera Labs:“Agentic AI is where the economics of AI infrastructure are being decided, and those economics depend on putting every usable gigabyte of memory to work,” said Thad Omura, senior vice president, Compute Connectivity Group at Astera Labs. “The enhanced Leo family gives infrastructure providers purpose-built ways to connect memory to accelerators, CPUs, and hosts across the rack—turning previously deployed and stranded capacity into a resource that new AI and cloud workloads can use. We're seeing that translate into broader customer engagement across AI labs, hyperscalers, and neoclouds.”𝗥𝗼𝗯𝗲𝗿𝘁 𝗛𝗼𝗿𝗺𝘂𝘁𝗵, Corporate Vice President, Architecture and Strategy at AMD:“Agentic AI is increasing demand for balanced systems that can move and process more data without sacrificing efficiency. AMD EPYC processors deliver leadership performance, exceptional memory bandwidth and capacity, and high-speed I/O across the broadly deployed x86 platform. Our collaboration with Astera Labs around CXL extends that foundation, giving customers greater flexibility to scale memory-intensive AI and cloud workloads and improve infrastructure utilization.”𝗘𝗱𝗱𝗶𝗲 𝗥𝗮𝗺𝗶𝗿𝗲𝘇, Vice President of Go-To-Market, Cloud AI, Arm:“As AI infrastructure evolves, agentic AI, reinforcement learning and database workloads are driving demand for higher-capacity memory. Arm and Astera Labs are working together to pair Arm AGI CPU with Leo CXL Smart Memory Controllers, helping customers meet these growing requirements while maintaining scalable performance.”𝗦𝗿𝗶𝗻𝗶 𝗞𝗿𝗶𝘀𝗵𝗻𝗮, Fellow at Intel Data Center Group:"CXL was created to bring greater flexibility and efficiency to memory across the data center. Technologies like Astera Labs' Leo Smart Memory Controllers help turn that vision into reality, demonstrating how ecosystem innovation can help customers get more out of their AI infrastructure and support the next generation of data-intensive workloads."𝗝𝗮𝗻𝗴𝘀𝗲𝗼𝗸 𝗖𝗵𝗼𝗶, Vice President of Product Planning Team at Samsung Electronics:"We are contributing to the evolution of AI infrastructure through the development of CXL memory solutions designed to help address the growing memory demands of modern data centers. We are excited to collaborate with Astera Labs to help ensure our DRAM solution offer strong interoperability within the evolving CXL ecosystem, and remain committed to supporting a scalable, high-performance memory architecture for the AI era."
对外销售的商用 GPU 市场目前主要是$英伟达(NVDA.US) 和 $AMD(AMD.US) 两家。MI300 时代 AMD 因为缺少系统级方案(所谓系统,核心就是互连方案)几乎没有吃到 AI 算力的红利。但是现在 Helios 来了,那么:1)AMD 是否真正具备了与英伟达同台掰手腕的能力?2)在互连方案的竞争与迭代中,到底有哪些产业链逻辑变化,谁受益谁吃亏?我们将从互连的角度来研究这一问题...