随美光大举释出封测委外订单，让力成首度跨足HBM封装市场，预计最快将于今年下半年试产并陆续 产出，外界认为，力成后续相关业绩成长动能可期。 美光上季财报与本季展望出色，持续于高频宽记忆体（HBM）市场攻城掠地，对后段封测委外需求激 增，美光将旗下HBM2封装大幅外包，由台厂力成（6239）独家拿下相关大单。 美光持续投资HBM、产出续扬，推升后段封测外包订单暴增，力成跟随美光，将成为此波市况爆发的 大赢家。 供应链透露，美光为全力抢攻HBM3E及建置明年将步入市场主流的HBM4产能，预计将把中科先进封 装基地中的封装产能移转到这两项产品，并开始将HBM2封装大量委外。 目前美光已与力成敲定HBM2的封装订单委外协议，后者并为此添购设备，预计今年中左右逐步到位， 下半年开始验证生产，年底前将进入小量试产阶段，明年相关产能可望大量开出。 美光提到，其12层HBM3E的良率与产量提升进展很顺利，预期在下半年某个时间点时，其HBM市占率 将达与整体DRAM市占率相当的水准。美光正向四家客户大量出货HBM，且其12层HBM3E 36GB 日前 已被超微的Instinct MI355X GPU平台采用。 法人看好，力成下 ...

Core Insights - Oracle is leveraging its Oracle Cloud Infrastructure (OCI) to drive AI growth, with OCI consumption revenues increasing by 62% in Q4 fiscal 2025, reflecting a strong demand for high-performance computing, particularly for AI workloads [1][8] - The company reported total cloud revenues of $6.7 billion in Q4 fiscal 2025, indicating a year-over-year growth of 27% [1][8] - Oracle has partnered with AMD to integrate AMD Instinct MI355X GPUs into OCI, achieving over 2X price-performance compared to previous generations, enhancing efficiency for large-scale AI workloads [2][8] - Oracle is aggressively expanding its global data center presence, currently operating in 23 cloud regions with plans for an additional 47, expecting OCI revenues to grow over 70% in fiscal 2026 [3][8] - The company deployed $9.1 billion in capital expenditures in the reported quarter, totaling $21.2 billion for fiscal 2025, with plans to invest $25 billion in fiscal 2026 [3][8] Competitive Landscape - Oracle faces significant competition in the cloud computing space from Amazon Web Services (AWS) and Microsoft Azure, both of which have established strong market positions [4][5][6] - AWS commands approximately one-third of the global cloud infrastructure market, benefiting from a growing enterprise customer base and high-margin services [5] - Microsoft Azure is recognized for its seamless integration with enterprise applications and advanced AI capabilities embedded in widely used software [6] Financial Performance - Oracle's stock has increased by 26.6% year-to-date, outperforming the Zacks Computer - Software industry's growth of 11% [7] - The company trades at an EV/EBITDA multiple of 26.7x, higher than the industry average of 19.24x, indicating strong market expectations for future growth [9] - The Zacks Consensus Estimate for Oracle's fiscal 2026 revenues is $66.73 billion, reflecting a year-over-year growth of 16.25%, with earnings estimated at $6.68 per share, indicating a growth of 10.78% from fiscal 2025 [10]

Core Insights - Advanced Micro Devices (AMD) has announced its next-generation AI chips, the Instinct MI400 series, which are set to ship next year, aiming to enhance its competitive position against Nvidia in the AI chip market [1][2]. Company Developments - The MI400 chips can be assembled into a server rack called Helios, allowing thousands of chips to function as a unified system, which is crucial for AI applications [2][3]. - AMD's CEO Lisa Su highlighted that the Helios rack operates as a single massive compute engine, positioning it against Nvidia's upcoming Vera Rubin racks [4]. - OpenAI's CEO Sam Altman confirmed that OpenAI would utilize AMD's chips, indicating strong industry interest [2][3]. Competitive Landscape - AMD aims to compete with Nvidia's Blackwell chips by offering lower operational costs and aggressive pricing strategies [5][6]. - Currently, Nvidia holds over 90% of the data center GPU market, but AMD is focusing on capturing a share of the projected $500 billion AI chip market by 2028 [10][19]. - AMD's MI355X chip is reported to have seven times the computing power of its predecessor, enhancing its competitiveness against Nvidia's offerings [14][15]. Market Trends - Major cloud companies are expected to invest hundreds of billions in new data center clusters for AI, with $300 billion planned in capital expenditures this year alone [9]. - AMD's AI chip sales reached $5 billion in fiscal 2024, with expectations of 60% growth in the category this year [19]. Technological Advancements - AMD's chips are designed to excel in inference tasks, which are increasingly important for deploying AI applications, and they feature more high-speed memory compared to Nvidia's chips [13][14]. - The company is integrating its GPUs with CPUs and networking chips from its acquisition of Pensando to enhance the performance of its Helios racks [17]. Customer Adoption - AMD's Instinct chips have been adopted by seven of the ten largest AI customers, including OpenAI, Tesla, and Oracle, indicating strong market traction [15].

Core Insights - Crusoe has announced a collaboration with AMD to develop AI cloud solutions, integrating AMD Instinct™ MI355X GPUs with Crusoe's AI cloud platform [1][2][3] - The partnership aims to meet the increasing demand for scalable and reliable AI infrastructure, with plans to deploy the GPUs across Crusoe's global data center network [2][3] - The new Crusoe Cloud offering featuring AMD GPUs is expected to launch in Fall 2025, providing enterprises with enhanced performance for AI projects [3] Company Overview - Crusoe is focused on creating a sustainable AI infrastructure by utilizing clean energy and building AI-optimized data centers [5] - The Crusoe Cloud platform is designed to support next-generation AI workloads with high scalability and reliability, allowing users to concentrate on innovation rather than infrastructure [4] Industry Context - The collaboration between Crusoe and AMD highlights the growing need for advanced AI infrastructure solutions in the market, as enterprises seek competitive advantages in their AI initiatives [2][3] - AMD's role as a trusted provider of AI solutions is emphasized, with its GPUs and ROCm software offering the necessary performance and flexibility for diverse AI applications [3]

Core Insights - Oracle and AMD are collaborating to offer an AI supercomputer featuring AMD Instinct MI355X GPUs, enhancing AI training and inference capabilities for customers [1][2] - The new zettascale AI cluster will support up to 131,072 MI355X GPUs, providing over 2X better price-performance compared to previous generations [1][2] - The partnership aims to meet the growing demand for AI infrastructure, enabling customers to handle larger and more complex datasets effectively [2][3] Performance Enhancements - The AMD Instinct MI355X GPUs deliver nearly triple the compute power and a 50% increase in high-bandwidth memory compared to the previous generation [2] - Customers can expect up to 2.8X higher throughput for AI deployments, resulting in faster results and lower latency [4] - The new architecture supports 288 gigabytes of high-bandwidth memory 3 (HBM3) and up to eight terabytes per second of memory bandwidth, facilitating the execution of large models entirely in memory [4] Infrastructure and Design - The OCI Supercluster features a high-throughput, ultra-low latency RDMA cluster network architecture, optimized for demanding AI workloads [2] - A dense, liquid-cooled design allows for 125 kilowatts per rack, accommodating 64 GPUs per rack at 1,400 watts each, enhancing performance density [4] - The powerful head node, equipped with an AMD Turin high-frequency CPU, optimizes GPU performance through efficient job orchestration and data processing [4] Open-Source and Networking Innovations - The open-source stack, AMD ROCm, enables flexible architectures and easy migration of existing code, preventing vendor lock-in [4] - Oracle will be the first to deploy AMD Pollara AI NICs, providing advanced RoCE functionality for high-performance and low-latency networking [5]