Agent workloads are shifting the CPU-to-GPU ratio in AI servers from the training-era standard of 1:8 toward 1:1, driving a multiplicative expansion in host-side memory and storage demand. Constrained by HBM production bottlenecks and the high premiums on large-capacity DDR5 modules, the competitive focus in AI storage is moving from single-layer high-speed video memory to cross-tier coordination among HBM, host memory, CXL, and enterprise SSDs. Google has leveraged TPU Sync to establish low-los
November 12, 2026
China's AI storage industry shifted from disorderly competition to ordered innovation. At the 2026 Data Storage Industry Conference, China's Data Storage Committee, with Sugon and Huawei, released two AI storage group standards and began compiling a national standard. They unify terminology and evaluation, focusing on concurrency, metadata performance, scenario adaptation, continuity, accelerator direct pass, zero copy and inference cache sharing.
November 10, 2026
In January this year, reports emerged that ASUS would exit the smartphone market and lay off staff in its mobile division. ASUS subsequently responded, stating that while there are no plans to launch new smartphones this year, the operations of its mobile business unit remain unchanged. However, recent supply chain updates indicate that ASUS smartphone inventory was largely sold out by the fourth quarter. With zero inventory and no new models to "carry the torch," this appears to signal the deat
November 9, 2026
Abstract As large model parameters expand and long-context, multi-turn reasoning becomes widespread, the actual computational load and historical cache (KV Cache) of AI servers continue to grow, but data access exhibits a distinct "hot-cold differentiation." In a high-end AI accelerator card, HBM already accounts for nearly 60% of the total Bill of Materials (BOM). When AI vendors are no longer willing to pay HBM prices for every piece of historical data, server storage is shifting from simply s
November 5, 2026
As AI computing power demand continues to reshape the DRAM industry cycle, three major financial reports have provided answers. Recently, Samsung, SK hynix, and ChangXin Memory Technologies (CXMT) released their results, confirming the sustained high prosperity in the memory sector: SK Hynix's Q3 revenue rose 328% year-on-year; Samsung's memory business revenue surged 427% year-on-year; CXMT's year-on-year growth rate remained at a high level of 589% to 619%. Amid this boom, questions arise: How
October 30, 2026
Core Summary In Q4 2026, China's AI compute supply chain exhibited significant divergence in memory capacity structure. While contract prices for frontier-generation HBM3E/HBM4 stacks rose to over USD 200–300 per stack, procurement prices for mature-generation (HBM2/HBM2E class) stacks used in mainstream Chinese accelerator cards increased from an initial range of USD 30–40 to USD 95–120 per stack. Channel surveys indicate that driven by early-stage yield ramp-up costs (public reports cite appro
October 29, 2026
On the evening of October 28, BIWIN Storage released a performance report that seemed to stun the entire semiconductor industry. Net profit attributable to shareholders for the first three quarters reached CNY 8.768 billion, representing a year-on-year increase of 28,728.12%. Upon the release of BIWIN Storage's Q3 report, many focused on the year-on-year growth rates of CNY 21.7 billion in revenue and CNY 8.7 billion in net profit. However, the more critical figure was different: Q3 net profit w
October 29, 2026
Author: Peng Cheng Only a handful of companies globally have achieved mass production of MRAM. For a long time, this track has been dominated by a few top players: Everspin is the only publicly listed company in the world with MRAM as its core business; Samsung and TSMC integrate embedded MRAM into advanced logic processes; NXP applies MRAM to automotive-grade MCUs. Within this small circle, domestic manufacturers are rapidly gaining presence: HikStor Technology's MRAM products have entered mass
October 28, 2026
In recent years, discussions on AI computing power have focused almost exclusively on GPUs. However, a frequently overlooked fact is that as large models enter the phases of long context windows, complex reasoning, and multi-agent systems, actual GPU utilization increasingly depends on whether model weights, KV Cache, and MoE experts can reach the correct compute nodes at the right time. Storage is moving from being a "peripheral device" into the real-time main path of token generation, which is
October 27, 2026
Recently, market research firm Counterpoint released partial smartphone shipment data for the third quarter of this year. Simply put, whether in the global or domestic market, the smartphone industry's performance this quarter has been relatively sluggish. Under the pressure of storage costs, overall shipments have declined. Meanwhile, the smartphone market is experiencing further internal differentiation. In the past, several major leading brands captured most of the market share, presenting a
October 21, 2026
Author | Bao Yonggang, Editor | Liu Wei Leiphone.com has exclusively learned that MemSilicon has completed its Series A funding round of over CNY 400 million. The round was led by Boyuan Capital, with follow-on investments from Bosch Ventures, Chuangling Capital, miHoYo, Zhongguancun Science City Company, GigaDevice Semiconductor, and existing shareholders Furong Investment, CDB Venture Capital, Hengyu Capital, Jinqiu Fund, Shunxi Fund, Xiaomi Strategic Investment, and CAS Star continuing to inc
October 20, 2026
As AI shifts from training to inference and Agentic AI, KV cache, long context, and large-scale data lakes are shifting the pressure from the compute side to the memory and storage side. Relying solely on HBM (High Bandwidth Memory) to stack bandwidth is no longer sufficient; systems need to re-tier across bandwidth, capacity, latency, and energy efficiency. At the recently concluded OCP 2026 Global Summit, the latest showcases from Samsung, SK hynix, Marvell, and DapuStor perfectly outline this
October 20, 2026
Author | Yang Yiting, Editor | Bao Yonggang In 2025, the golden age of the memory industry is unfolding. TrendForce estimates that the global memory industry's output value will reach USD 889.3 billion this year, and will continue to grow by 53% in 2027, approaching USD 842.7 billion. Amid the boom, the supply side is further locked down: the three major memory manufacturers—Samsung, SK Hynix, and Micron—have already finalized the capacity allocation for DRAM (Dynamic Random Access Memory) and H
October 19, 2026
On October 12, major DRAM chip manufacturer Nanya Technology held an investor conference and officially released its Q3 2026 financial report. Not only did its revenue and net profit continue to achieve rapid quarter-on-quarter growth, but they also hit record highs. The gross profit margin reached a record high of 85.7%. Nanya Technology General Manager Li Peiying expects that the average DRAM price will continue to rise in Q3, and there is still room for the gross profit margin to increase. Av
October 15, 2026
Changxin Memory Technologies (CXMT) is expected to launch a new generation of DDR5 RDIMM products by the end of this year, a message revealed by Cao Kanyu, President of CXMT, at the recently held 4th Integrated Circuits and Chiplets Conference. Cao Kanyu delivered a speech at the conference, comprehensively introducing the technological exploration and accumulation of CXMT in DRAM devices, the 4F² architecture, and hybrid bonding, and officially announced that the new generation of DDR5 RDIMM pr
October 12, 2026
Author: Peng Cheng Industrial storage is a low-key yet critical niche market in the semiconductor industry. Unlike consumer-grade storage, which prioritizes capacity and cost, industrial storage focuses more on reliability, such as 24/7 continuous operation, wide-temperature environments from -40 to 125°C, strong vibration and electromagnetic interference, and frequent unexpected power outages. These operating conditions dictate that industrial equipment cannot directly adopt consumer-grade stor
October 12, 2026
Korea Semiconductor Industry Association Releases Report First Quantification of the Technology Gap Between China and South Korea Approximately 1-Year Gap in NAND Approximately 2-Year Gap in DRAM Approximately 3-Year Gap in HBM On September 15, a report released by the Korea Semiconductor Industry Association pointed out that the technology gap between China and South Korea in the memory chip field has narrowed from the previously assessed "over 5 years" to about "one generation." The gap in NAN
October 10, 2026
From its listing on the ChiNext board on April 16 to the announcement of planning an H-share issuance on the evening of August 24, only 130 days passed. The market's initial reaction was almost reflexive: Is it short of cash? But if you know that developing a complete enterprise SSD controller chip + firmware + product validation, from project initiation to mass production, takes over 5 years—you might change your perspective. What they are rushing for is not money, but time. 130 Days: The Marke
October 8, 2026
AI inference KV caches are now large enough to need dedicated storage, as seen with Huawei M900 and NVIDIA CMX. Offloading reusable KVs from DRAM to SSDs trades storage cost for less recomputation, but 1TB DRAM does not equal 1TB SSD; solutions may need 1:N SSD capacity. SSD standards remain fragmented. Externalization mainly eases DRAM, not HBM, while data movement costs persist; CIM may help Prefill. The real cost depends on KV lifetime, reuse, media, and retrieval.
October 8, 2026
Author: Peng Cheng Recently, frequent news has emerged from the NOR Flash market. Market research firm TrendForce released a forecast: with limited capacity expansion for NOR Flash, prices for high-capacity products are expected to rise by another 90% to 110% in the second half of the year, and prices for some models may directly double. The root cause of this round of price increases stems from the concentrated explosion of demand in the AI server and automotive electronics markets. At almost t
October 8, 2026