Core Viewpoint - The company is a leading supplier of single crystal silicon materials for integrated circuit etching, holding a global market share of approximately 15%. It is expected to benefit from the overall supply chain improvement driven by the storage industry's growth cycle, leading to potential earnings growth exceeding expectations [1]. Demand Side Analysis - The demand for large-sized single crystal silicon materials used in etching electrodes is applicable to logic chip etching. If the semiconductor cycle does not recover, the company's growth may be adversely affected [2]. - Rapid growth in AI server demand and the quick iteration of AI chips necessitate corresponding increases in storage throughput. The importance of storage chips in AI training and inference is growing, which significantly boosts the company's core business [3]. Supply Side Analysis - There is a continuous tight demand for orders from overseas storage manufacturers, creating challenges for the supply of large-sized single crystal silicon materials and etching electrodes. Domestic material manufacturers may have the opportunity to take on some overflow orders [3]. - Local Japanese and Korean enterprises have reached their supply limits, increasing the company's earnings elasticity due to overflow orders [5]. Catalysts and Indicators - Key indicators include the rhythm of storage demand and discrepancies between actual performance and guidance [6]. - Catalysts for growth include additional overseas orders and the company's capacity expansion [7]. Research Value - In the context of significant demand for AI computing power driving the storage industry, the market has overlooked the upstream segment of the storage supply chain. The manufacturing of storage chips requires multiple etching processes, necessitating large amounts of single crystal silicon materials and etching electrodes. The company is a global leader in this field, with a market share of about 15%, and is expected to benefit significantly from the growth cycle in storage [8]. Profit Forecast and Valuation - Considering the company's leading position in silicon materials and the downstream storage sector's benefits from computing power development, revenue projections for the company from 2025 to 2027 are estimated at 470 million, 800 million, and 1.31 billion yuan, representing year-on-year growth of 55.6%, 70.5%, and 62.8%. Net profits attributable to the parent company are expected to be 110 million, 270 million, and 450 million yuan, with year-on-year growth of 168.9%, 141.7%, and 68.6%. Based on the closing price on November 7, 2025, the PE ratios are projected to be 78.5, 32.5, and 19.3 times, maintaining a "buy" rating [9].

Core Viewpoint - The computing power sector is experiencing a surge due to rising prices of HBM4 chips supplied by SK Hynix to Nvidia, indicating a broader trend of price increases in global memory chips, which is positively impacting cloud computing ETFs [1][10]. Group 1: Market Trends - The cloud computing ETF Huatai (159273) saw an increase of over 2% in early trading, with its latest scale exceeding 1.5 billion yuan, leading its peers [1][4]. - SK Hynix reported a 119% year-on-year increase in net profit for Q3 2025, with NAND ASP rising by 10%-15% and DRAM ASP increasing by 4%-6% [10][11]. Group 2: Company Performance - New Yi Sheng reported a Q3 2025 revenue of 6.07 billion yuan, a 5% decrease quarter-on-quarter, attributed to the rhythm of customer orders [3][7]. - Zhongji Xuchuang achieved a gross margin of 42.79% and a net margin of 32.57% in Q3 2025, benefiting from the ramp-up of silicon photonic modules and the release of 1.6T high-speed optical modules [7][9]. Group 3: Future Outlook - The demand for high-speed optical modules is expected to rise significantly due to increased capital expenditures in AI by global CSP manufacturers and accelerated iterations of overseas optical communication solutions [3][9]. - The AI server market is projected to grow by over 30% annually by 2026, driven by strong demand from CSPs and sovereign clouds, with AI server shipments expected to increase by over 20% annually [10][11].

Investment Rating - The report maintains a "Strong Buy" rating for the company, indicating an expectation to outperform the benchmark index by over 20% in the next six months [20]. Core Insights - The company reported a revenue of 9.276 billion yuan in Q3 2025, reflecting a year-over-year increase of 6.58% and a quarter-over-quarter increase of 0.64%. The net profit attributable to shareholders was 202 million yuan, down 3.11% year-over-year but up 42.81% quarter-over-quarter [1][5]. - The strong performance is attributed to a significant increase in storage prices driven by supply-demand dynamics, with expectations for continued price increases in Q4 2025 [5]. - The company has successfully developed a dual business model of "distribution + products," enhancing its competitive edge in the high-end storage market [5]. Financial Performance Summary - For 2025, the company is projected to achieve total revenue of 35.249 billion yuan, with a year-over-year growth rate of 45.2%. The net profit attributable to shareholders is expected to reach 641 million yuan, reflecting a growth rate of 142.9% [6]. - The earnings per share (EPS) is forecasted to be 1.38 yuan in 2025, with a price-to-earnings (P/E) ratio of 77 [6]. - The gross margin for Q3 2025 was reported at 4.03%, with a net margin of 2.11%, indicating improved profitability despite a slight year-over-year decline in margins [5]. Market Dynamics - The report highlights that the demand for storage solutions is increasing, with significant price hikes expected in both DRAM and NAND Flash products due to supply constraints and rising demand from various sectors [5]. - The company is positioned to benefit from the ongoing trend of domestic substitution in the storage market, which is anticipated to contribute positively to its performance [5]. Strategic Developments - The establishment of a joint venture with Wuxi Lingjing Cloud is expected to enhance the company's distribution and product capabilities, further solidifying its market position [5]. - The company has made significant progress in the development of enterprise-level SSD and RDIMM products, which are now entering mass production [5].

Core Viewpoint - The article emphasizes the critical role of storage power in the AI ecosystem, highlighting the need for innovative storage solutions to meet the diverse demands of AI throughout its lifecycle [1][7]. Group 1: Storage Power in AI - The demand for storage power is exponentially increasing as AI permeates all aspects of production and life, with storage being likened to the "blood system" supporting AI's "brain" [1]. - Different stages of the AI lifecycle have varying storage requirements: high bandwidth for data collection, low latency for data cleaning, energy efficiency for training, and cost-reliability balance for inference [1]. Group 2: Town Yue 510 SSD Controller Chip - The Town Yue 510 SSD controller chip from PingTouGe Semiconductor is positioned as a solution to the storage bottleneck in the AI ecosystem, achieving breakthroughs in architecture and algorithms [3][5]. - The chip is characterized as a "hexagonal warrior," excelling in low latency (4 microseconds for random writes), high energy efficiency (420K IO processing per watt), high bandwidth (≥3400K IOPS), high reliability (10^-18 UBER), low cost (supporting various NAND types), and high capacity (≥32TB) [4][6]. Group 3: Architectural and Algorithmic Innovations - The chip's architecture features a "bottom-up reconstruction" with six RISC-V cores forming a heterogeneous computing cluster, significantly reducing IO processing latency [5]. - Key algorithmic innovations include a combination of Bit-Flipping and Normalized Min-Sum algorithms for data error correction, achieving a UBER of 10^-18, placing it among the top tier in the industry [6]. Group 4: Ecosystem Development and Industry Collaboration - PingTouGe is actively building an ecosystem around the Town Yue 510 chip, with rapid productization and deployment in Alibaba Cloud's EBS services, achieving a 92% reduction in latency compared to competitors [9]. - Collaborations with companies like Yiheng Chuangyuan and Derui Lingxin have led to high-performance SSDs based on the Town Yue 510, enhancing capabilities for AI training and real-time data analysis [10][11]. Group 5: Future Outlook - The company is preparing for future advancements, focusing on the maturity of the ecosystem and planning the next generation of chips to further enhance reliability and reduce power consumption [14]. - The vision is to transform storage from a mere data container into a driving force for AI innovation, akin to essential utilities like water and electricity [14].

Core Viewpoint - The article emphasizes the importance of enhancing storage capabilities in the AI era, highlighting that while computing power has received significant attention, storage development has been relatively overlooked. The storage infrastructure is crucial for supporting AI applications, akin to how a reservoir supports hydropower generation [1][6]. Group 1: Development of Storage Technology - The Zhenyue 510 SSD controller, developed by PingTouGe, has been successfully commercialized and integrated into Alibaba Cloud's EBS, significantly improving system IOPS and throughput while optimizing IO latency, thus enabling cost reduction and efficiency for customers [2][3]. - The Zhenyue 510 has enabled faster response times for applications running on Alibaba Cloud EBS, enhancing user experience [3]. - Beyond cloud computing, PingTouGe is providing new options for SSD manufacturers, with products like the PBlaze7 7A40 series achieving 1 million IOPS and the D8436/D8456 series improving performance-to-power ratio by 70% compared to previous PCIe 4.0 products [4][5]. Group 2: Importance of SSD in AI Applications - SSDs play a critical role in AI applications by affecting computation, transmission, and storage processes. The speed of data supply directly impacts GPU utilization during model training, with SSD I/O latency being a bottleneck compared to DRAM [7]. - During inference, AI models require rapid decision-making, necessitating high data access speeds and low latency [8]. - High write throughput is essential for the checkpointing phase in AI model training and deployment, indicating the need for advanced storage solutions [9]. Group 3: Technological Innovations of Zhenyue 510 - The Zhenyue 510 can handle up to 3.4 million IOPS and achieve a data bandwidth of 14 GByte/s, outperforming traditional HDDs [18]. - It supports PCIe 5.0 and DDR5.0 interfaces, achieving ultra-low latency of 4μs, which is over 30% lower than industry standards, and significantly reduces long-tail latency in mixed read/write scenarios [18][22]. - The controller features a low UBER of 10^-18, indicating a high reliability level, and employs advanced error correction algorithms to enhance performance [21]. Group 4: Strategic Vision and Market Position - PingTouGe aims to leverage its infrastructure and application scenarios to validate its products and expand the storage ecosystem, targeting various industries with tailored solutions [23][24]. - The company positions the Zhenyue 510 as a versatile solution that balances capacity, latency, cost, energy efficiency, reliability, and bandwidth, adapting to different industry needs [24]. - With a strong backing from Alibaba and a keen understanding of industry trends, PingTouGe aspires to lead the upcoming wave of storage device upgrades [26].