As the supply of computing power moved from a single card to 10,000 card clusters, and computing power network scheduling moved from networking to cross-domain collaboration, the cost issue that the industry is most concerned about also surfaced. Many brokerage firms believe that the ultimate industrial dividend of integrated computing power grid scheduling is to continuously reduce the unit cost of large-scale model training and inference by revitalizing idle computing power, optimizing task allocation, and superimposing the advantages of green power and heterogeneous technology. The core is reflected in the sharp decline in AI Token production costs. At present, operator computing power network scheduling has formed a mature cost reduction paradigm. In addition to operators' scheduling practices, deep collaboration between domestic computing power software and hardware vendors has also been implemented. Not long ago, Chujing Technology reached a strategic cooperation with Moore Thread to build a localized high-quality AI token factory. The relevant plan has already been officially put into operation. Liu Xianhe, general manager of the Token Division of Chujing Technology, said that as large-scale model applications enter the large-scale commercial implementation stage, infrastructure competition is shifting from single-card performance to system-level token production efficiency.

Disclaimer:Webull uses external vendor Google Translation Service for news translations where we endeavour to ensure these are correct, however, we recommend that you please double-check this information accordingly. Webull is not responsible for translation errors or issues.