According to multiple media reports, artificial intelligence company Zhipu AI is attempting to build a fully integrated compute ecosystem independent of Nvidia's ecosystem. Bloomberg's report, citing anonymous sources, reveals that Zhipu has completed construction of a data center with 1GW of power capacity and has already brought part of it online. The facility runs entirely on Chinese-made AI chips and is primarily used for research and development of its GLM series of large language models. A 1GW power capacity can roughly meet the electricity needs of 750,000 households. Multiple media outlets have confirmed that Zhipu has spent hundreds of millions of yuan to acquire Zhongke Jiahe, a domestic heterogeneous computing software company. Zhongke Jiahe originated from the Compiler Laboratory of the Institute of Computing Technology at the Chinese Academy of Sciences, and its core team has participated in compiler development for Chinese chips including Loongson, Sunway, Cambricon, and Huawei Ascend. Earlier, The Information reported that Zhipu is in contact with Chinese chip design companies to evaluate the possibility of co-developing custom AI chips, though the project remains in early-stage discussions.