科大讯飞刘庆峰：坚持国产算力自主可控，有实力做也必须做

Core Viewpoint - The chairman of iFlytek, Liu Qingfeng, highlighted the challenges of conducting large model training and algorithm innovation based on domestic computing power, emphasizing the need for additional steps and computational resources [1] Group 1: Domestic Computing Power Challenges - Liu Qingfeng pointed out that developing new algorithms based on domestic computing power will be three months slower compared to using overseas computing power [1] - The company insists on self-controllability as a necessary choice to avoid being constrained by external factors [1]