证券日报网讯德明利（001309）10月10日在互动平台回答投资者提问时表示，公司持续与包括长江存储 在内的核心供应商深化合作，积极沟通市场变化与技术创新等相关事宜，相关合作为供应链稳定性及成 本优化创造有利条件。AI基建浪潮下，服务器和数据中心领域的企业级存储需求不断增长，带动存储 行业景气度不断提升，公司密切关注大客户需求动态，持续推进相关业务对接，公司企业级存储产品已 经实现稳定出货，具体业绩影响敬请关注公司后续定期报告相关内容。与核心供应商的技术适配及大客 户的深度合作，有助于公司强化"硬件+技术+供应链"的定制化存储解决方案能力，持续积累AI存储及 国产化场景经验，强化以研发创新为核心的市场竞争力，为公司在相关领域的业务发展奠定基础。 ...

格隆汇10月10日丨德明利(001309.SZ)在投资者互动平台表示，公司持续与包括长江存储在内的核心供 应商深化合作，积极沟通市场变化与技术创新等相关事宜，相关合作为供应链稳定性及成本优化创造有 利条件。AI基建浪潮下，服务器和数据中心领域的企业级存储需求不断增长，带动存储行业景气度不 断提升，公司密切关注大客户需求动态，持续推进相关业务对接，公司企业级存储产品已经实现稳定出 货，具体业绩影响敬请关注公司后续定期报告相关内容。与核心供应商的技术适配及大客户的深度合 作，有助于公司强化"硬件+技术+供应链"的定制化存储解决方案能力，持续积累AI存储及国产化场景 经验，强化以研发创新为核心的市场竞争力，为公司在相关领域的业务发展奠定基础。 ...

Core Viewpoint - The rapid development of AI has made storage a critical component in the AI infrastructure, alongside computing power. The demand for storage is surging due to the increasing data volume and inference scenarios driven by large models and generative AI. Three storage technologies—HBM, HBF, and GDDR7—are redefining the future landscape of AI infrastructure [1]. Group 1: HBM (High Bandwidth Memory) - HBM has evolved from a high-performance AI chip component to a strategic point in the storage industry, significantly impacting AI chip performance limits. In less than three years, HBM has achieved over twofold capacity and approximately 2.5 times bandwidth increase [3]. - SK Hynix is leading the HBM market, currently in the final testing phase for the sixth generation (HBM4) and has announced readiness for mass production. In contrast, Samsung is facing challenges in HBM4 supply to Nvidia, with a two-month delay in testing [3][5]. - A notable trend is the customization of HBM, driven by cloud giants developing their AI chips. SK Hynix is shifting towards a fully customized HBM approach, collaborating closely with major clients [4]. Group 2: HBF (High Bandwidth Flash) - HBF aims to address the limitations of traditional storage by combining the capacity of NAND flash with the bandwidth of HBM. Sandisk is leading the development of HBF technology, which is expected to meet the growing storage demands of AI applications [8][9]. - HBF is seen as complementary to HBM, suitable for specific applications requiring large block storage units. It is particularly advantageous in scenarios demanding high capacity but with relatively relaxed bandwidth requirements [10][11]. Group 3: GDDR7 - Nvidia's introduction of the Rubin CPX GPU, utilizing GDDR7 instead of HBM4, reflects a new approach to AI inference architecture. This design optimizes resource allocation by separating the inference process into two stages, effectively utilizing GDDR7 for context building [13]. - The demand for GDDR7 is increasing, with Samsung successfully meeting Nvidia's orders. This flexibility positions Samsung favorably in the graphics DRAM market [14]. - GDDR7's cost-effectiveness may drive the widespread adoption of AI inference infrastructure, potentially increasing overall market demand for high-end HBM due to the proliferation of applications [15]. Group 4: Industry Trends and Future Outlook - The collaborative evolution of storage technologies is crucial for the AI industry's growth. HBM remains essential for high-end training and inference, while HBF and GDDR7 cater to diverse market needs [23]. - The ongoing innovation in storage technology will accelerate as AI applications expand across various sectors, providing tailored solutions for both performance-driven and cost-sensitive users [23].

又有超百亿资金跑步入场！宽基ETF获爆买 券商主题遭弃 记者据Wind数据统计，股票型ETF的净流入额已连续两日在100亿元以上。全市场1037只可统计的股票 型ETF基金，在9月26日、9月29日的合计净流入额分别为193.93亿元、122.69亿元。其中，宽基ETF已 连续两日包揽净流入额前三。在周一资金净流入前十的ETF中，更是有9只为宽基ETF，剩余1只行业主 题ETF则为电池板块。值得一提的是，在周一券商板块大幅拉升的情况下，不少资金选择获利了结，多 只券商ETF净流出额居前。不过整体来看，月内以来，券商、机器人、电池是最吸金的三大板块。（澎 湃） 香港隔夜Hibor升破5% 创下今年来首次 香港隔夜借贷成本今年来首次跃升至5%以上。香港隔夜银行同业拆息(Hibor)周二大涨130个基点至 5.018%，为去年12月以来的最高水准。这使得截至9月的三个月内，累计上涨近500个基点。（同花 顺） 硅片企业Q4确定减产 电池片整体库存水平降至3-4GW 部分硅片企业国庆节后才公布排产计划，激进预期下硅片企业10月减产幅度或很小，后续仍需观察原料 价格以及下游传导情况。整体Q4根据协会配额，硅片企业基本确 ...

Core Viewpoint - The "storage instead of computing" paradigm emerges as a disruptive technology to overcome computing bottlenecks and storage limitations in AI inference, significantly reducing latency and costs while enhancing throughput [1][2]. Development Background - AI inference has become a key measure of the commercial value of large models, facing challenges such as slow processing and high costs. The "storage instead of computing" technology addresses these issues by migrating vector data from expensive DRAM and HBM to cost-effective SSDs, enabling strategic expansion of storage layers [1]. Core Technology - The "storage instead of computing" Cached Attention technology caches historical KVCache data across HBM, DRAM, and SSD, achieving an 87% reduction in first token latency and a 7.8-fold increase in prefill throughput, leading to a 70% decrease in end-to-end inference costs [2]. Hardware Breakthroughs - Under the "storage instead of computing" paradigm, SSDs evolve from mere data storage to core components in AI inference, requiring high capacity, throughput, and low latency. The AISSD technology will develop in three directions: transitioning to QLC particles, adopting PCIe 5.0/6.0 interfaces with NVMe protocols, and upgrading functionalities towards intelligent solutions [4]. Industry Layout - Major industry players are actively engaging in the core practices of "storage instead of computing," with companies like Huawei and Inspur optimizing storage architectures and cache management for efficient AI inference [5][6]. International Developments - Global giants such as Kioxia, Micron, and Solidigm are pushing for technological iterations and product innovations in AISSD, with QLC+PCIe/NVMe+CXL expected to form the foundation for the next generation of AISSD, transforming SSDs into long-term memory carriers for AI inference [10]. Investment Recommendations - The AI storage revolution is underway, with "storage instead of computing" creating significant opportunities. Companies in storage module manufacturing, storage chips, and distribution/testing are recommended for attention, including Jiangbolong, Demingli, and Zhaoyi Innovation [10].

深信服（300454.SZ）9月18日在投资者互动平台表示，公司于2019年正式发布了企业级分布式存储产品 EDS。根据第三方咨询机构IDC的相关报告，2025年第一季度，公司EDS产品在"中国软件定义存储-文 件存储"市场占有率排名第五。根据内部统计数据，公司EDS产品累计客户已超4000家。2025年上半 年，公司EDS产品聚焦高价值场景、持续推进产品品质提升和方案优化，收入实现较快增长，但由于收 入基数还不大，对公司整体收入增长的拉动作用还不明显。 未来，公司EDS产品将重点聚焦在高性能 统一存储方向，持续推动EDS产品应用于大数据、AI推理等高价值业务场景。关于相关业务的具体进 展，敬请关注公司定期报告、"深信服科技"公众号及视频号所发布的信息。 （文章来源：每日经济新闻） 每经AI快讯，有投资者在投资者互动平台提问：请问公司是否有AI存储相关的业务，上半年是不是增 长迅猛？ ...

Group 1: SK Hynix and HBM4 Development - SK Hynix announced the successful development of HBM4, a high-performance memory for AI, and established a mass production system, leading to a 7% stock price increase and a new historical high [1] - HBM4 offers double the bandwidth of its predecessor and over 40% energy efficiency improvement, with 2048 input/output ports, expected to enhance AI service performance by 69% and reduce data center electricity costs [1] - SK Hynix's DRAM market share reached 36% in the first half of the year, surpassing Samsung's 33%, and is projected to maintain around 60% market share in the HBM segment by 2026 [1] Group 2: Industry Insights and Competitor Analysis - Industry experts suggest that SK Hynix's new AI storage chip aligns well with the demand for computational upgrades, but potential risks include production delays, yield rates, and cost pressures [2] - The high-end HBM market is currently dominated by Samsung, Micron, and SK Hynix, with Samsung planning initial HBM4 production in Q4 2025 and Micron's HBM4 samples undergoing customer validation [1][2] Group 3: Tesla Stock Performance and Business Developments - Tesla's stock surged over 13.8% in two days, driven by multiple factors including the approval of Robotaxi testing in Nevada, strong sales of new energy products, and Elon Musk's substantial compensation package [2] - Tesla's three main business areas—energy storage, Robotaxi, and Optimus—are seen as creating an ecosystem that supports profitability and growth, although long-term challenges include regulatory delays and competition [3] Group 4: Gold Price Forecast and Investment Potential - Goldman Sachs raised its long-term gold price forecast from $2850 to $3300 per ounce, with expectations for gold to reach $4000 per ounce by mid-2026, potentially nearing $5000 in extreme scenarios [4] - The profitability expansion of mid-sized and large gold mining companies is expected to drive stock performance, with a focus on the dynamics of central bank gold purchases and the Federal Reserve's interest rate decisions [4][5]

Core Viewpoint - The semiconductor storage market is expected to reach a historical high of $167 billion in 2024, driven by demand recovery in mobile phones, PCs, and servers, with NAND Flash and DRAM markets projected at $69.6 billion and $97.3 billion respectively [4][12]. Summary by Sections Overview of Storage Chips - Storage chips are essential components in modern electronic devices, categorized into volatile and non-volatile types. Volatile storage loses data when power is off, while non-volatile storage retains data [5]. Types of Volatile Storage - Static Random Access Memory (SRAM) is fast but costly, used in high-speed applications like CPU caches [6]. - Dynamic Random Access Memory (DRAM) is widely used in smartphones, PCs, and servers, requiring constant refreshing to maintain data [7]. - High Bandwidth Memory (HBM) offers high speed and bandwidth, suitable for AI accelerators, but is also expensive [7]. Non-Volatile Storage - NAND Flash is the mainstream large-capacity storage, known for its low cost and high capacity, but has slower write speeds and limited write cycles [8]. - NOR Flash is used for storing programmable code, offering fast random read speeds but with smaller capacity and higher costs [8]. AI Device Storage Requirements - AI devices require high-capacity, high-bandwidth, and low-power storage solutions, with LPDDR5 or LPDDR5X being the mainstream choices [9]. - The cost of storage in AI devices may account for 10-20% of overall hardware costs, reflecting its high priority in these applications [9]. Market Trends - The storage market experienced significant price increases in 2021, followed by a period of inventory digestion in 2023-2024, with prices expected to rebound starting late 2023 [12][14]. - HBM revenue is projected to double from $17 billion in 2024 to $34 billion in 2025, driven by strong demand [14]. 3D Stacking Technology - 3D stacking technology is crucial for meeting the high capacity, bandwidth, and low power requirements of AI storage chips, with ongoing developments in both packaging and wafer levels [19]. Industry Chain - The storage chip industry chain consists of upstream materials and equipment, midstream design and manufacturing, and downstream applications [20][23]. - The design segment has the highest profit margins due to high technical barriers, while packaging and testing have lower margins due to intense competition [23]. Recent Price Movements - Micron has paused pricing due to AI SSD demand shortages, with planned price increases of 20-30% for AI-related products [25].

Group 1 - The company, Demingli, plans to optimize its previous fundraising project by increasing investment in the "PCIe SSD storage control chip and module project" from 499 million to 743 million yuan, with a new implementation site in Shenzhen Guangming [1] - This strategic adjustment aims to align with the explosive demand in the AI server storage sector, ensuring the company builds a competitive advantage in the current AI industry opportunity [1][2] - The demand for AI computing power is surging, with major domestic cloud service providers expected to significantly increase capital expenditures, such as ByteDance's projected investment of 150 to 160 billion yuan in this area by 2025 [1][2] Group 2 - According to a report by China Telecom Research Institute, the AI industry is expected to contribute over 11 trillion yuan to China's GDP by 2035, driving a tenfold to hundredfold increase in computing power demand [2] - The adjustment in Demingli's fundraising project aligns with the urgent need for capacity upgrades in storage, which is a core component of computing infrastructure [2] - Innovations in AI storage technology, such as multi-level cache architecture and task-specific designs, are accelerating market expansion and require upgrades in data throughput, response speed, low power consumption, and overall cost [2] Group 3 - Demingli is leveraging its self-developed technology and a deep understanding of end-user applications to enhance storage performance and stability through a comprehensive customization capability [3] - The company is expanding its manufacturing capabilities with a new base in Shenzhen Guangming, focusing on automated production and digital management to meet high-quality, customized AI storage demands [3] - With the implementation of its fundraising project, Demingli aims to enhance its technological leadership and market coverage, promoting the penetration of domestic storage solutions in high-performance AI scenarios [3]

Core Insights - The article discusses Huawei's launch of AI SSDs designed to address the performance and capacity limitations of traditional storage systems in the AI era [1][3][8] - The new SSDs aim to enhance AI training and inference efficiency, breaking through existing bottlenecks in the industry [5][6][9] Group 1: Product Innovation - Huawei introduced AI SSDs, including the OceanDisk EX/SP/LC series, optimized for AI workloads, which can handle high-frequency tasks like training data feeding and vector retrieval [3][5] - The performance SSDs (OceanDisk EX and SP) significantly improve IOPS by over three times and reduce latency to microsecond levels, enhancing data transfer capabilities [5][6] - The capacity SSD (OceanDisk LC) offers options of 61TB, 122TB, and a record-breaking 245TB, drastically reducing storage costs and space requirements for enterprises [6][8] Group 2: Market Impact - The introduction of Huawei's AI SSDs is expected to lower the entry barriers for small and medium enterprises in the AI sector, enabling them to access high-quality AI storage solutions [8][9] - The shift in focus from mere computational power to a combination of storage and computing capabilities is seen as a critical evolution in the AI industry [8][9] Group 3: Technological Advancements - Huawei's DiskBooster software enhances memory capacity by 20 times and optimizes data management, allowing for larger model handling without additional hardware [5][9] - The development of core technologies like virtual pooling and intelligent data scheduling positions Huawei to lead in the storage sector, potentially benefiting other areas like cloud computing and big data [9]