英伟达存储架构变化如何影响 NAND Flash 的需求测算 20260107 摘要 英伟达新架构通过将 KV Cache 数据卸载到 NAND Flash，解决 HBM 和 DDR 内存容量不足问题，更好地支持 GPU 工作。 英伟达提供每个 GPU 额外增加 16TB NAND Flash 的明确数据，使得 NAND Flash 需求测算更加清晰，改变了以往依赖 GPU 出货量估算 HBM 增速的模式。 假设英伟达全生命周期出货量达到 800 万张卡，对应的 NAND Flash 额 外需求将增加 136 亿 GB，相较于当前 925 亿 GB 总供给量增长约 15%，增量将在 2026-2027 年落地。 ASIC 卡也将参考类似比例关系进行测算，英伟达新架构为 NAND Flash 需求提供了更明确的参考依据。 AI 时代，NAND Flash 需求预计持续扩容，一方面通过 GPU 或 ASIC 与 NAND Flash 比例关系进行需求增长测算，另一方面多模态生成技术产 生大量温数据需要存储。 与大模型交互过程中用户历史信息也会记录在 NAND Flash 上，大模型 推理规模落地后，将进一步推动对 ...

Core Insights - Hunyuan Image 3.0 has achieved the top position in the global multimodal generation ranking on LMArena, indicating its leading status in the field [1][2][4] - The model was evaluated through a "blind test" mechanism, reflecting real user preferences and showcasing its strong performance compared to other models [4] Group 1: Model Performance - Hunyuan Image 3.0 scored 1167 points, leading the ranking among 26 models, surpassing competitors like Gemini 2.5 and Seedream 4 [2][3] - The model has been recognized as the best comprehensive and best open-source text-to-image model [2][4] Group 2: Model Features - Hunyuan Image 3.0 is the first open-source industrial-grade multimodal generation model, capable of generating high-quality images with accurate semantic understanding [4][6] - It supports both Chinese and English text generation, including long text rendering, and can generate images based on simple prompts [8][9] Group 3: Community Reception - The model quickly gained popularity, reaching the top of the Hugging Face open-source community model leaderboard shortly after its release [4] - Hunyuan Image 3.0 has been downloaded over 2.6 million times in the 3D model community, indicating its widespread acceptance and usage [15]

Group 1 - The recent financing round was led by Bohua Capital, with participation from Baidu's strategic investment, Beijing Artificial Intelligence Industry Investment Fund, Qiming Venture Partners, Datatech Capital, BV Baidu Ventures, and other existing shareholders [1] - Shengshu Technology was established in March 2023, with a core team comprising talents from top global universities and industry professionals, showcasing strong practical experience and global technology implementation capabilities [2] - The company plans to utilize the new funding for model research and technological innovation, aiming to explore the intelligent limits and application breadth of multimodal large models, while enhancing product expansion, user services, industry collaboration, and global business layout [2] Group 2 - Vidu, launched globally in July 2024, introduced the innovative "reference life" image/video concept, achieving rapid coverage of over 30 million users and 6,000 developers and enterprises across more than 200 countries and regions [2] - The total number of videos generated by Vidu has exceeded 400 million, with the number of reference life videos and images surpassing 100 million, of which over 50% are commercial content materials [2] - The CEO of Shengshu Technology, Dr. Luo Yihang, indicated that the commercialization of multimodal generation technology in the digital content industry is accelerating but remains in its early stages, with significant market space and global growth potential expected in the next three years [2]

Group 1 - A-shares indices opened higher on May 28, with the Computer ETF (159998) experiencing a slight decline of 0.47% and a trading volume exceeding 20 million yuan [1] - The Computer ETF (159998) has seen continuous capital inflow over the past 8 trading days, accumulating a net inflow of 112 million yuan, making it the top performer in its category [1] - The latest scale of the Computer ETF is 2.801 billion yuan, making it the largest computer ETF by on-market size [1] Group 2 - The Cloud Computing ETF (517390) has recorded net inflows on 4 out of the last 5 trading days, indicating strong investor interest [2] - The Computer ETF tracks the CSI Computer Index, which includes stocks from companies involved in information technology services, application software, system software, and computer hardware, with top holdings including Hikvision, iFlytek, Kingsoft Office, and others [2] - Huawei has launched the Ascend Super Node technology, which consists of 12 computing cabinets and 4 bus cabinets, achieving the industry's largest scale of 384-card high-speed bus interconnection [2] Group 3 - The AI sector is undergoing rapid evolution, shifting focus from model scale and benchmark performance to user experience and interaction innovation, leading to a new round of industry reshuffling [3] - Generative AI is transitioning from passive response to active execution of complex tasks, enhancing its practical application scenarios and commercialization potential [3] - Domestic AI computing power ecosystems, represented by Ascend, are improving through innovations in underlying architecture and tools, boosting the performance of related industries such as high-speed connectors and liquid cooling [3] Group 4 - As general reasoning capabilities advance, high-value applications in research and programming are expected to unlock first, benefiting software and internet sectors [3] - Hardware demand is anticipated to rise alongside advancements in multimodal technology, maintaining a positive outlook on investment opportunities in the AI computing power sector [3]