软件基建决定应用边界，低费率云计算ETF华夏（516630）涨超5%，资金连续两日抢筹

Group 1 - The core viewpoint of the article highlights the strong performance of AI applications and software, with platforms like Kimi, Xiaohongshu, Douyin Doubao, and Chinese corpus leading the market surge, while Yidian Tianxia has seen consecutive trading limit increases [1] - The AI market trend continues, shifting focus towards mid and downstream sectors, with Guojin Securities emphasizing that software infrastructure determines application boundaries and strengthens the industrial foundation [1] - AI infrastructure, while not directly producing models, is crucial for determining the "cost curve" and "capability ceiling" of the AI ecosystem, reflected in four main aspects: computing power scheduling, data throughput, full-stack support, and resource virtualization [1][2][3] Group 2 - Computing power scheduling serves as the neural hub of clusters, focusing on efficiently orchestrating large-scale heterogeneous processor tasks and ensuring training continuity through self-healing mechanisms [1] - Data throughput aims to eliminate I/O bottlenecks to maximize GPU utilization during training, while optimizing operators to handle high concurrency challenges and significantly reduce latency during inference [1] - Full-stack support provides high-performance training and inference frameworks, accelerating model convergence and enhancing response speed [2] - Resource virtualization abstracts hardware differences, enabling software-defined computing power and significantly lowering development and deployment barriers [3] Group 3 - The related ETF mentioned is the Huaxia CSI Cloud Computing and Big Data Theme ETF, with off-market connections (Class A: 019868; Class C: 019869), which has seen over 70 million yuan in net inflows over two consecutive trading days [1][4]