Core Viewpoint - The emergence of High Bandwidth Flash (HBF) technology is accelerating commercialization, aiming to fill the storage gap in AI inference, while existing technologies like HBM continue to evolve [1][2]. Group 1: HBF Technology Overview - HBF is not a new concept but is derived from the transition in the AI industry from training to large-scale inference, addressing the storage bottleneck in current architectures [2]. - HBF aims to provide a solution that combines high bandwidth, large capacity, and cost-effectiveness, bridging the gap between HBM's limited capacity and SSD's slower read/write speeds [2][3]. - HBF utilizes a restructured architecture and packaging optimization, incorporating 3D stacking technology from HBM and replacing storage media with NAND flash, achieving high-speed data transfer while retaining NAND's advantages [3]. Group 2: HBF and HBM Collaboration - HBF is designed to complement HBM rather than replace it, with HBM handling latency-sensitive tasks and HBF focusing on large-capacity sequential reads [4]. - The "H3 hybrid architecture" by SK Hynix demonstrates the synergy between HBM and HBF, showing significant performance improvements in various scenarios [4]. Group 3: Industry Response and Future Outlook - Major storage companies, including SK Hynix and SanDisk, are actively investing in HBF, with SanDisk being the pioneer in its commercialization [6]. - The establishment of a dedicated working group under the Open Compute Project (OCP) will facilitate the standardization of HBF technology, paving the way for its widespread adoption [6]. - HBF products are expected to be showcased by late 2026, with market demand projected to grow rapidly by 2030, driven by the expansion of AI inference applications [6]. Group 4: Challenges Ahead - Despite its promising outlook, HBF still faces inherent limitations related to NAND flash's slower write speeds, which could impact user experience in dynamic caching scenarios [7]. - Industry collaboration is necessary to address compatibility with existing GPU architectures, packaging complexity, and thermal management issues [7].

Group 1 - SK Hynix announced the launch of the "HBF Specification Standardization Alliance" in collaboration with SanDisk, introducing a global standardization strategy for the next-generation storage solution HBF (High Bandwidth Flash) aimed at the AI inference era [1] - HBF technology is designed to fill the storage tier gap between HBM and SSD, providing an innovative storage solution that meets the dual demands of capacity scalability and energy efficiency for AI inference scenarios [1] - The HBF technology is expected to be integrated into products from NVIDIA, AMD, and Google by 2027-2028, with the market size projected to surpass HBM by 2038 [1] Group 2 - The standardization of HBF technology marks a new phase in AI storage architecture, and companies that can offer a full-stack storage solution combining both HBM and HBF will gain strategic advantages as AI inference demand surges [1] - Related A-share concept stocks mentioned include Saiteng Co., Ltd. and Yake Technology [1]

AI Programming - Huawei Cloud's CodeArts code intelligent agent public beta was released on February 26, integrating code generation, research knowledge Q&A, unit test case generation, and more, enhancing developer efficiency [1] - Cinda Securities believes AI programming is reshaping core productivity, with large model technology empowering programming tools, and expects the AI code tool market to reach $26 billion by 2030 [1] Commercial Space - China Aerospace Science and Technology Corporation announced that its reusable liquid rocket, the LQ-2, is set for its maiden flight in late March, aiming for 13 launches in 2026 [2] - Blue Arrow Aerospace plans to conduct recovery tests for its reusable rocket, ZQ-3, in the second quarter of this year, with a goal of achieving its first recovery flight by the end of 2026 [2] - Zhongtai Securities notes that 2026 will be a pivotal year for China's commercial rockets, transitioning from technology breakthroughs to scalable, low-cost implementations [2] Artificial Intelligence - OpenRouter data shows that from February 9 to 15, 2026, China's AI model call volume reached 4.12 trillion tokens, surpassing the US for the first time [3] - The growth momentum of Chinese AI models is attributed to a cluster of manufacturers rather than a single product, with four Chinese models ranking among the top five globally [3] - Research institutions predict that enterprise-level large model call volumes will continue to rise, with a focus on head manufacturers capable of handling high concurrency and long-term scenarios [3] Computing Power - DeepSeek announced the upcoming launch of V4 "Sea Lion Lite," which supports a 1M ultra-long context window, significantly enhancing AI model application boundaries [4] - The new version is deeply adapted to Huawei's Ascend platform, achieving over 35 times speed improvement, marking a key step towards a closed-loop domestic AI industry chain [4] PCB - NVIDIA's CEO Jensen Huang revealed a non-exclusive technology licensing agreement with AI startup Groq, aiming to enhance NVIDIA's AI infrastructure [5] - Groq's LPU chip, with a speed advantage of 5-10 times over GPUs, is expected to play a crucial role in NVIDIA's strategy to capture the inference market [5] HBF - SK Hynix and SanDisk announced the launch of the "HBF Specification Standardization Alliance," introducing the next-generation storage solution HBF for AI inference [6] - HBF technology aims to bridge the gap between HBM and SSD, expected to be integrated into products from major companies by 2027-2028, with a market size projected to exceed HBM by 2038 [6] Photovoltaics - A research team from the Qingdao Institute of Bioenergy and Process Research achieved over 15% photoelectric conversion efficiency with a new solar cell material, CZTSSe, which has been internationally certified [7] - CZTSSe is considered a promising next-generation solar cell technology due to its abundant material sources, low cost, and environmental safety [7]

Group 1 - The chip sector experienced a counter-trend rise on February 26, with notable increases in stocks such as Xingchen Technology (up over 14%), Cambricon (up over 7%), and Changchuan Technology (up over 6%) [1] - The ETF Huatai-PineBridge (516920), which tracks the chip industry, rose over 1% as of 14:40, marking its fifth consecutive increase [1] - SK Hynix announced plans to invest 21.6 trillion KRW (approximately 150.7 billion USD) in new chip production lines in South Korea by 2030 to meet the growing semiconductor demand [3] Group 2 - SK Hynix reported a record operating profit of 47.2 trillion KRW (approximately 330.14 billion USD) for the year [3] - A domestic chip company announced a price increase of at least 10% for MOSFET products, continuing the trend of rising chip prices [3] - The semiconductor capital expenditure expectations are increasing due to rising prices in computing, storage, and display sectors, benefiting domestic semiconductor development [3] Group 3 - The Huatai-PineBridge ETF (516920) tracks the CSI Chip Industry Index, which includes 50 leading stocks across the entire chip supply chain, covering equipment materials, wafer foundry, design, and testing [4]