On July 20, at the 2026 World Artificial Intelligence Conference, Xia Lixue, co-founder and CEO of Wuwen Xinqiong, proposed that the current artificial intelligence infrastructure “must serve a larger scale with extreme efficiency.” As of July, the number of daily token calls on its platform had increased 40 times compared to December of last year. According to reports, Wuyuan Xinqiong has driven a 10-fold reduction in inference costs through a series of technological innovations. Xia Lixue judged that with the large-scale implementation of heterogeneous PD separation architectures across clusters, there is still room for inference costs to drop tenfold.

On July 20, at the 2026 World Artificial Intelligence Conference, Xia Lixue, co-founder and CEO of Wuwen Xinqiong, proposed that the current artificial intelligence infrastructure “must serve a larger scale with extreme efficiency.” As of July, the number of daily token calls on its platform had increased 40 times compared to December of last year. According to reports, Wuyuan Xinqiong has driven a 10-fold reduction in inference costs through a series of technological innovations. Xia Lixue judged that with the large-scale implementation of heterogeneous PD separation architectures across clusters, there is still room for inference costs to drop tenfold.