Workflow
半导体存储
icon
Search documents
自研芯片再发力,江波龙兑现承诺
半导体行业观察· 2025-03-18 01:36
Core Viewpoint - Jiangbolong aims to break the "curse" of storage module operations by achieving over $2 billion in revenue and transitioning towards becoming a semiconductor storage brand [1][2]. Group 1: Self-Developed Chip Strategy - Jiangbolong has increased its investment in technology to support its transition to a semiconductor storage brand, focusing on algorithm and IP design [2]. - The company established a main control chip team in 2020 and launched eMMC and SD controllers last year, marking a significant milestone in its self-developed chip strategy [4][5]. - The self-developed main control chips allow for application innovation in storage technology, enabling high customization based on customer needs [4]. Group 2: New Product Launches - Jiangbolong introduced several self-developed main control chips at the flash memory summit, including WM7400 (UFS 4.1), WM7300 (UFS 3.1), and WM3000 (USB 3.2) [5][6]. - The UFS 4.1 main control chip WM7400 features advanced manufacturing processes and supports both TLC and QLC NAND Flash, achieving sequential read speeds of up to 4350MB/s [7][8]. - The eMMC Ultra product, equipped with the WM6000 main control chip, enhances bandwidth by 50% compared to the eMMC 5.1 standard, providing a theoretical speed of 600MB/s [9][10]. Group 3: Automotive Storage Solutions - Jiangbolong has developed a comprehensive automotive storage product matrix, including AEC-Q100 compliant UFS, eMMC, and LPDDR4x products [11]. - The company has established deep partnerships with over 20 OEMs and 50 Tier 1 automotive clients, with a projected market growth of nearly 100% in the automotive sector [12][18]. Group 4: Business Model Transformation - Jiangbolong is shifting its business model to include Technology Contract Manufacturing (TCM) and Product Technology Manufacturing (PTM) services, aiming to reduce costs and optimize inventory management [15][16]. - The PTM model provides comprehensive storage Foundry services, addressing industry homogenization and innovation bottlenecks [16][18]. - The company has successfully implemented the PTM model across various industries, including automotive, servers, and mobile devices, enhancing its market presence [18][19]. Group 5: Global Market Expansion - Jiangbolong is accelerating its global market strategy, particularly in the Americas and Europe, to deepen the implementation of TCM and PTM models [20][21]. - The company aims to leverage its self-developed chips and proprietary packaging to enhance its service capabilities and meet diverse customer needs [23][24].
都盯上了HBM
半导体行业观察· 2025-03-09 03:26
Core Viewpoint - Samsung plans to launch its first mobile product featuring LPW DRAM memory by 2028, which is expected to significantly enhance performance and reduce power consumption in mobile devices [1][15]. Group 1: HBM in Automotive Sector - The automotive industry is increasingly adopting HBM memory due to the rising demand for real-time data processing and high-resolution image handling in advanced driving assistance systems and smart cockpit systems [2][4]. - SK Hynix's HBM2E has already been integrated into Waymo's autonomous vehicles, marking a significant entry of HBM into the automotive sector [3]. - The global automotive storage chip market is projected to grow from $4.76 billion in 2023 to $10.25 billion by 2028, with demand for HBM in automotive applications expected to surpass that in data centers [5]. Group 2: HBM in Mobile Devices - The rapid advancement of AI and 5G technologies is driving the need for high-performance memory in mobile devices, as they require higher bandwidth and lower latency for complex applications [7][9]. - Mobile HBM, developed by Samsung and SK Hynix, utilizes advanced 3D stacking technology to significantly increase memory bandwidth, achieving data transfer rates of hundreds of GB/s [9][10]. - Samsung's LPW DRAM is expected to achieve a bandwidth of over 200GB/s, a 166% increase compared to existing LPDDR5x, while reducing power consumption to 1.9pJ/bit, which is 54% lower than LPDDR5x [16]. Group 3: Competitive Landscape - Samsung and SK Hynix are competing in the mobile HBM market, with Samsung focusing on high bandwidth designs and SK Hynix prioritizing low power consumption and cost-effectiveness [20][21]. - The development of mobile HBM is expected to shift from a supplier-centric model to a demand-driven approach, allowing for customized solutions tailored to specific customer needs [21]. - As both companies advance their technologies, the competition will likely influence the market dynamics for mobile HBM, impacting the performance of smartphones, PCs, and AR/VR devices [20][26].
新兴存储,最新预测
半导体行业观察· 2025-03-06 01:28
Core Viewpoint - The article discusses the evolution and future of alternative and persistent memory technologies, highlighting the competition among various types of memory to become mainstream in the semiconductor industry [2][3][6]. Group 1: Historical Context - For the past 40 years, semiconductor memory has evolved from SRAM, DRAM, EPROM, and EEPROM to include newer technologies like FRAM, MRAM, ReRAM, and PCM [2][3]. - NAND flash memory has been a cornerstone of non-volatile storage, but it faces limitations at the 15nm node, leading to the development of 3D NAND variants [7][8]. Group 2: Current Developments - Recent advancements in memory technologies include the emergence of microcontrollers utilizing MRAM and FRAM, with companies like NXP and Texas Instruments leading the way [6][10]. - The collaboration between NXP and TSMC aims to develop MRAM-based microcontrollers for the automotive market in 2023 [6]. Group 3: Challenges and Limitations - The transition to alternative memory technologies faces economic challenges, as the costs associated with these new technologies are currently higher than traditional NAND and DRAM [8][10]. - The integration of new memory types into existing systems is complicated by the need for additional on-chip SRAM, which increases costs and complexity [5][10]. Group 4: Future Predictions - Experts predict that it may take around ten years for alternative memory technologies to replace flash and SRAM in embedded applications due to slow development in microcontroller technology [10]. - The transition to alternative memory in external NAND flash chips and SDRAM is expected to be delayed, but once it begins, it may accelerate quickly [10].