[DeepSeek V5 Expected to Complete Training Based on Ascend 950 Next Year] According to sources familiar with the industry chain, DeepSeek's next-generation large-scale model, DeepSeek V5, plans to use Huawei's Ascend 950 chip as its core training computing power, with the project projected to launch next year. Currently, DeepSeek's core R&D team is based in Hangzhou, and the two sides have completed extensive adaptation and customization of underlying operators in the V4 phase. According to the cooperation path disclosed by the sources, in the previous generation of DeepSeek, Ascend computing power was only used for inference tasks and had not yet entered the training phase; V5 will be its first large-scale model natively trained using the Ascend 950. (Financial News)