深圳商报·读创客户端记者 李薇 本次，证监会要求该公司补充具体业务模式、近三年技术出口业务开展情况及合规性，境外子公司涉及 的境外投资、外汇管理等监管程序履行情况，并出具合规性结论。 股权与激励方案是否存利益输送 该公司还需说明新增股东入股价格合理性及差异原因，是否存在利益输送；已实施的股权激励方案需披 露人员构成、价格公允性及合规性，是否存在关联关系或利益输送。 2025年10月，武汉聚芯向港交所递交招股书，启动港股IPO流程。 据招股书，武汉聚芯主要通过提供智能感知、机器视觉和影像技术产品实现收入，受下游行业需求增加 等因素的影响，近几年该公司的收入也有所增长。 2022年、2023年、2024年和2025年1-6月（报告期），公司的收入分别为1.27亿元、2.42亿元、6.67亿 元、4亿元；净利润分别为-1.37亿元、-1.1亿元、973万元、2072万元。 虽然2024年该公司净利润由负转正，但2023年公司毛利率下滑至23.6%，主要产品售价呈下降趋势。 业务模式与合规性需说明 武汉聚芯微电子股份有限公司（以下简称："武汉聚芯"）近期因港股IPO申请被中国证监会要求补充材 料，监管关注的问题涉及业 ...

近日，存储市场呈现显著结构性变化：DDR4内存芯片价格持续攀升并反超DDR5，出现罕见"价格倒 挂"现象。这一逆势涨价信号背后，是存储产业技术迭代的必然趋势，标志着DDR4退场序幕正式拉 开，DDR5时代全面开启，存储产业"换挡"提速进入新阶段。 在行业数字化领域，金融高频交易、互联网支付洪峰等场景对系统性能要求严苛，DDR4在延迟、带宽 等方面的瓶颈日益凸显。转向DDR5不仅是硬件更新，更是金融机构构筑核心竞争力的战略举措。结 合"十五五"规划中"高水平科技自立自强"的发展目标，DDR5的普及成为存储产业支撑数字经济发展的 关键抓手。 DDR5相较DDR4实现代际飞跃：带宽提升两倍，容量与能效显著优化，更集成片上ECC纠错功能，能 大幅降低数据中心宕机风险，为AI大模型训练等场景提供坚实支撑。尽管DDR4仍有市场惯性，但固守 旧技术将面临多重挑战。 推动DDR5普及是顺应技术规律、满足消费需求、夯实AI发展基础的战略性举措。 市场数据显示，自今年6月起，16GB容量的DDR4内存芯片价格开始反超同容量DDR5芯片。截至8月 末，DDR4单价从6月的7.01美元升至8.59美元，DDR5则从5.85美元涨至6 ...

17 Nov 2025 03:47:51 ET │ 13 pages Vi e w p o i n t | China Technology & Communications Key Takeaways from Citi 2025 China Conference – Memory Price Hike, Foldable iPhone, OpenAI AI Edge, AI-PCB, Smart Glasses CITI'S TAKE 26 China/HK tech companies joined Citi 2025 China Conference. Key takeaways: (i) supply chain believes memory price hike should impact low- to-mid end models but not high-end, while OEMs may raise ASP to mitigate impact; (ii) foldable flagship smartphone will drive UTG, hinge spindle, heat ...

23 November 2025 | 5:21PM EST Economics Research Global Views: Looking Beyond December 1. Though badly delayed, the September jobs report may have sealed a 25bp cut at the December 9-10 FOMC meeting. On Friday, New York Fed President (and FOMC Vice Chair) Williams argued for "a further adjustment in the near term" because "the downside risks to employment have increased as the labor market has cooled, while the upside risks to inflation have lessened somewhat." His view is likely consistent with that of Cha ...

Group 1 - Changxin Storage showcased its latest DDR5 and LPDDR5X product lines at the 22nd China International Semiconductor Expo, with DDR5 products achieving speeds up to 8000Mbps and capacities of 24Gb [1] - The company plans to cover various market segments with seven types of modules, including UDIMM, SODIMM, and RDIMM, targeting high-end markets in servers, workstations, and personal computers [1] - According to Trendforce, Changxin Storage's production capacity is expected to reach 300,000 wafers per month by the end of 2025, marking a nearly 50% year-on-year increase [1] Group 2 - According to predictions, Changxin Storage's market share for DDR5 is expected to rise from 1% at the beginning of the year to 7% by year-end, while LPDDR5's share is projected to increase from 0.5% in Q1 2025 to 9% by year-end [1] - Zhaoyi Innovation has invested 2.3 billion in Changxin Technology, acquiring a 1.88% stake, and is making progress in custom storage solutions across various sectors, anticipating a scale-up period by 2026 [2] - Shangluo Electronics has been collaborating with leading domestic storage brands for over five years and has recently secured agency qualifications for Kioxia storage, serving clients in servers, mobile phones, and various consumer IoT fields [2]

机器之心报道 编辑：泽南 一举解决算力资源浪费。 不论是英伟达 GPU 还是昇腾的 NPU，都可以「融为一体」，动态切分了。 11 月 21 日，华为正式发布了 AI 容器技术 ——Flex:ai，同时，华为联合上海交通大学、西安交通大学与厦门大学共同宣布，将此项产学合作成果向外界开源，助 力破解算力资源利用难题。 据介绍，Flex:ai 深度融合了上海交通大学、西安交通大学、厦门大学三大高校与华为的科研力量，形成了三大核心技术突破： 针对 AI 小模型训推场景中「一张卡跑一个任务」可能造成的资源浪费问题，华为与上海交通大学联合研发 XPU 池化框架 ，可将单张 GPU 或 NPU 算力卡切分为 多份虚拟算力单元，切分粒度精准至 10%。 华为公司副总裁、数据存储产品线总裁周跃峰博士在发布会上表示，当前，AI 产业高速发展催生海量算力需求，但全球算力资源利用率偏低的问题日益凸显， 「算力资源浪费」成为产业发展的关键桎梏：小模型任务独占整卡导致资源闲置，大模型任务单机算力不足难以支撑，大量缺乏 GPU/NPU 的通用服务器更是处于 算力「休眠」状态，供需错配造成严重的资源浪费。 本次发布并开源的 Flex:ai ...

Nvidia graphics processing units like these latest Blackwell GPUs, are inside server racks all over the world. Nvidia has catapulted from gaming giant, to the very core of generative AI, training the models, running the workloads and sending Nvidia's valuation soaring. With 6 million Blackwell GPUs shipped over the last year.This connects all 72 GPUs, allowing the act as a single GPU, to power the most advanced AI workloads. GPUs are the general purpose workhorse stars of Nvidia and its top competitor, AMD. ...

SUBSCRIBE | FastCo Works advertisement Try our mini crossword with a business twist: Play Fast Company Mini Crossword LOGIN BYÂ Jennifer Mattson Listen to this ArticleMore info 0:00 / 0:00 Is this the beginning of the end for Bitcoin? A Bitcoin (BTC) sell-off on November 20 sent the cryptocurrency down 3% to $86,410.50 in midafternoon trading, after a rally had it above $93,000 earlier in the day. It's now hit its lowest level since April. It's part of an overall decline in the crypto market that also saw c ...

点击 最 下方 "在看"和" "并分享，"关注"材料汇 添加 小编微信 ，遇见 志同道合 的你 正文 引言 当芯片制程的微缩逼近物理与经济的双重极限，集成电路产业的发展动能，正前所未有地从晶体管尺度 的缩小，转向系统级集成与架构的创新。在这一被称为"后摩尔时代"的产业变局中， 先进封装与测试 （简称"先进封测"） 已从配套辅助环节，跃升为提升芯片性能、优化系统功耗、控制整体成本的关键 支点。 本文旨在深入剖析全球及中国先进封测行业的 发展现状、技术脉络、市场格局与驱动因素 ，并前瞻性 地展望在人工智能、高性能计算等需求的强劲牵引下，该领域所孕育的 未来机遇与发展趋势 。 一、集成电路先进封测概况 1、集成电路制造概况 集成电路制造产业链主要包括芯片设计、晶圆制造、封装测试三个环节，具体如下： 封装测试包含封装和测试两个环节 ，其中，封装是指将集成电路与引脚相连接以达到连接电信号的目 的，并使用塑料、金属、陶瓷、玻璃等材料制作外壳保护集成电路免受外部环境的损伤； 测试包括进入封装前的晶圆测试（CP）和封装完成后的成品测试（FT），晶圆测试主要检验每个晶粒 的电性能，成品测试主要检验产品的电性能和功能，目的是将有 ...

Take a Trial Today Get instant access to exclusive stock lists, expert market analysis and powerful tools with 2 months of IBD Digital for only $20! Dow Jones futures rose modestly early Friday, along with S&P 500 futures and Nasdaq futures, on renewed Fed rate cut hopes. Bitcoin tumbled again overnight but is off its worst levels. The stock market suffered an ugly downside reversal Thursday as Nvidia (NVDA) gave up early post-earnings gains. Many leading stocks, especially in the broader AI space, also sho ...