On April 24th, the preview version of the new DeepSeek-V4 model from DeepSeek was officially launched and open-sourced. The new version achieves a leading level in China and the open-source domain in terms of Agent capability, world knowledge, and logical reasoning performance. The model adopts a MoE architecture, supports 1 million Token native context, and has increased inference speed by 35 times compared to V3, with a cost reduction to 1/70 of GPT-4. This model iteration has catalyzed the development of two industries. On the computing power end, DeepSeek-V4 has completed deep integration with domestic chips, boosting the purchase of domestic AI chips and upstream processes such as wafer foundry and advanced packaging. On the application end, DeepSeek-V4 has significantly reduced the cost of inference, promoting the commercialization of scenarios such as cross-document analysis and intelligent agent workflows, opening up growth opportunities for cloud computing and enterprise-level AI services. In this context, two ETFs under E Fund are worth paying attention to. E Fund AI ETF (03489) closely tracks the FTSE Custom Global Artificial Intelligence Select Index (with a 13.25% increase in April), covering domestic computing power leaders such as SMIC and Hua Hong Semiconductor, and including core AI application targets like Tencent and Alibaba as well as overseas tech giants like Nvidia, Microsoft, and Broadcom. E Fund Asia Semiconductor ETF (03486) focuses on core Asian semiconductor companies, including TSMC and Hua Hong Semiconductor as industry benchmarks. The index it tracks saw a 26.05% increase in April, benefiting from both the surge in domestic computing power and the global semiconductor cycle.