迈威尔科技首席执行官兼董事长马特·墨菲表示："我们与英伟达扩大合作伙伴关系，反映了高速连接、 光学互连和加速基础设施在扩展AI方面日益增长的重要性。通过NVLink Fusion将迈威尔科技在高性能 模拟、光数字信号处理器、硅光子和定制芯片领域的领先地位与英伟达不断扩展的AI生态系统连接起 来，我们使客户能够构建可扩展、高效的AI基础设施。" 值得一提的是，对迈威尔科技20亿美元的投资是英伟达本月第三笔如此规模的投资。此前该公司已宣布 与Lumentum(LITE.US)和Coherent(COHR.US)达成交易。本月早些时候，英伟达宣布向Lumentum和 Coherent各投资20亿美元，旨在借助相关技术强化其数据中心芯片，更好地满足市场对更快AI处理器日 益增长的需求。上述合作包括英伟达提供数十亿美元的采购承诺，以及未来获得两家公司先进激光与光 通信产品的产能保障和技术使用权。这些投资将帮助相关企业支持研发、扩产与运营，同时推动其美国 制造能力建设。 智通财经APP获悉，英伟达(NVDA.US)宣布向迈威尔科技(MRVL.US)投资20亿美元，两家公司还公布了 一项战略合作伙伴关系，将通过英伟达的NV ...

Collaboration Delivers Greater Choice and Flexibility for Customers and Fully Compatible With NVIDIA AI InfrastructureSANTA CLARA, Calif., March 31, 2026 (GLOBE NEWSWIRE) -- NVIDIA and Marvell Technology, Inc. (NASDAQ: MRVL) today announced a strategic partnership to connect Marvell to the NVIDIA AI factory and AI-RAN ecosystem through NVIDIA NVLink Fusion™, offering customers building on NVIDIA architectures greater choice and flexibility in developing next-generation infrastructure. The companies will als ...

据伯恩斯坦称，埃隆·马斯克的 TeraFab 要每年生产 1 TW 的 AI 硅，每年需要使用 142 到 358 个晶 圆厂处理 2240 万片 Rubin Ultra GPU 晶圆、271.6 万片 Vera CPU 晶圆和 1582.4 万片 HBM4E 晶 圆。 该公司采用自上而下的方法，将机架级电力需求转化为所需的半导体制造产能，从而得出这些数据， 并 将 其 描 述 为 " 非 常 粗 略 的 晶 圆 产 能 估 算 " 。 分 析 师 利 用 机 架 级 系 统 的 功 耗 （ Rubin 为 120 kW ， Rubin Ultra 为 600 kW），将系统体积转换为芯片数量，然后根据芯片尺寸（例如 GPU 芯片约为 825 mm²，CPU 芯片约为 800 mm²）、HBM 堆叠数量和良率，最终得出晶圆需求。然而，伯恩斯坦 似乎高估了逻辑晶圆厂的典型产能（每月 50,000 片晶圆启动，而不是 20,000 片），低估了 DRAM 晶圆厂的产能（50,000 片晶圆启动，而不是 100,000 至 200,000 片），并且假设每个晶圆厂的价格 为 3500 万美元，即使数万亿美元的规模总 ...

Nvidia (NVDA +1.95%) stock has the potential to surge 150% over the next few years, but one big thing is going to have to happen.First and foremost, artificial intelligence (AI) infrastructure spending is going to have to continue to climb, and its customers are going to have to signal that this spending is sustainable. Without that happening, little else matters. That's why this is the one thing that 100% has to happen for Nvidia stock to see big gains from here. The good news is that data center AI spendi ...

Shares of Arm (ARM) rocketed more than 15% higher in early trading Wednesday, after the company unveiled its first production data center processor: the Arm AGI CPU (central processing unit). Arm has traditionally licensed its intellectual property to other companies to develop their own chips, including Apple and Nvidia (NVDA), which uses Arm’s capabilities in its Grace and Vera CPUs. Graphics processing units, or GPUs, have dominated data centers thanks to their ability to train and run AI models. Bu ...

For years, Arm (ARM) has played a key role in the development of processors for everything from the iPhone to data center chips. But now the company is expanding its reach with the debut of its first production data center processor: the Arm AGI CPU (central processing unit). Arm has traditionally licensed its intellectual property to other companies to develop their own chips, including Nvidia (NVDA), which uses Arm’s capabilities in its Grace and Vera CPUs. Graphics processing units, or GPUs, have do ...

For more than 35 years, Arm Holdings has licensed its instruction sets to the world's biggest chipmakers and collected royalties on every processor made with its designs. Now the U.K.-based company is making physical silicon of its own for the first time.Arm CEO Rene Haas unveiled his company's first in-house chip on Tuesday at an event in San Francisco. Arm is calling the new data center central processing unit the AGI CPU. It's a long-anticipated move that marks a major change for the so-called Switzerlan ...

全球科技硬件调研汇报 20260322 AI 算力需求从训练转向推理，预计 2027 年数据中心市场规模将突破 1 万亿美元，推理算力需求将达 ChatGPT 初期的 1 万倍。 英伟达 Rubin 平台通过增大 SRAM 使推理每瓦性能提升 10 倍；Vera CPU 专为 Agent 任务设计，实现工具负载从 GPU 卸载。 高端 PCB 迎来量价齐升：Rubin 平台全盲插设计使单 GPU 对应 PCB 价 值增超 30%；正交背板单柜价值达 12-16 万美元。 存储行业受限于无尘室空间缺口，2026-2027 年将维持供不应求； NAND 价格涨幅超 60%，产能优先向 HBM/DRAM 迁移。 CPO 技术明确在 Scale-up 场景放量，预计 2027 年后随英伟达 Fermion 平台落地；2030 年 CPO 市场预期上调至 150 亿美元。 Scale-across 架构驱动空心光纤与 Coherent Lite 技术需求，旨在解 决跨数据中心 AI 集群微秒级同步与带宽瓶颈。 单波 400G 时代薄膜铌酸锂异质集成成为确定性趋势，相比传统方案功 耗降低约 20%，天孚通信等在位企业具备 ...

英伟达在GTC大会中发表最新的LPU芯片Groq，主要就是强化推论效能，GPU芯片经过大量训练及搜寻结果后，LPU芯片再将适合的答案推论相匹配，使用 者更快得到运算结果，而非GPU独自完成这两项步骤，GPU、LPU芯片搭配可让AI运算效力大幅提升。 英伟达Vera Rubin平台还包括Rubin GPU芯片、Vera CPU、Spectrum以太网路、NVLink交换器（Switch）平台、BlueField数据处理芯片。执行长黄仁勋指 出，所有芯片相互整合效应让英伟达硬体运算效能达到最高。 这项订单也代表后续衍生的AI服务器代工将进入交货旺季。业界推估，Vera Rubin架构的AI服务器预计今年第3季开始逐步量产及交货，英伟达明年将推出 Rubin Ultra的强化版GPU运算芯片，预期亚马逊AWS本次与英伟达签订采购协议也包含此产品。 英伟达和AWS本周已表示达成采购协议，并未透露交易的具体日程。英伟达超大规模与高效能运算副总裁巴克（Ian Buck）告诉路透，将从今年开始出货， 并持续到2027年。英伟达和亚马逊均未揭露交易具体财务数字。巴克透露，除了100万颗GPU，交易还包含英伟达的多种芯片，包括 ...

Nvidia (NVDA) has always been at the forefront of innovation. Since inception, the tech giant has invested $58.2 billion in research and development. As of fiscal 2025, the company also had 27,100 out of 36,000 total employees engaged in R&D. Therefore, unsurprisingly, Nvidia is not only strengthening its moat but also pushing new frontiers. At GTC 2026, Nvidia recently announced the Vera Rubin Space-1 chip system, which could potentially be used for orbital AI data centers. To simplify things, the curre ...