Z.AI, developer of the GLM family of AI models, says it has completed construction of a large AI data center designed for model training and development. According to Bloomberg and TechNode, the facility houses only Chinese-made chips and is positioned to support Z.AI’s frontier-model work. TechNode characterizes the project as 1GW-class and reports that part of the site is already operating, while Bloomberg describes it as a “giant” data center and highlights its significance within Beijing’s broader push to reduce reliance on restricted Nvidia hardware for future AI development. Both accounts frame the center as a step toward scaling compute using domestically sourced components. While the reports differ slightly in emphasis—Bloomberg focusing on the policy context and TechNode on the scale and operational status—they agree on the core points: Z.AI has finished the data center, it is built around Chinese chips only, and it is intended for training and development of its AI models.