On March 6th, Southern Finance reported that news about Nvidia attempting to block third-party GPU companies in its CUDA software is attracting attention from the artificial intelligence and chip industries both domestically and internationally. CUDA is a software platform launched by Nvidia in 2006, specifically designed to accelerate GPU computing. Nvidia did not respond to any inquiries from reporters regarding this matter. In response to market rumors, domestic GPU companies have officially responded. On the evening of March 5th, GPU chip company Mole Thread issued a statement stating that Mole Thread MUSA/MUSIFY does not involve any terms related to NVIDIA EULA, and developers can use it with confidence. Moore Thread further emphasized that the MUSA architecture it launched has no dependency on CUDA. How to migrate CUDA applications to third-party hardware platforms has always been a challenge for Nvidia's competitors. A middle-level domestic AI chip company once told reporters that once users get used to CUDA, it is difficult for them to have the motivation to migrate again. Because migration almost means complete code rewriting, the work is heavy, time-consuming, and does not generate economic benefits. (Interface News)