Chinese AI lab Moonshot reportedly has a compute capacity agreement with Alibaba Cloud to the tune of 20,000 Nvidia GPUs. As reported by Bloomberg and citing people familiar with both companies' operations, the Nvidia GPUs make up a large proportion of Moonshot's total compute capacity used for the training and running of the Kimi models.

Chinese tech giant says its latest open weight model is built for complex real-world tasks and long running AI agents.

Alibaba says its latest model can compete with Anthropic’s Claude Fable 5.

(Yicai) Aug. 4 -- With domestic artificial intelligence firms such as DeepSeek having shown that an open-source approach can carve out a distinct market niche, Chinese internet giant Alibaba has announced that its flagship Qwen-Max model will also be made open-source for the first time.

Alibaba says Qwen3.8-Max uses 2.4 trillion parameters for long-running AI tasks, with open weights planned for developers and enterprises.

Alibaba stock climbed after the company launched its powerful Qwen3.8-Max AI model, boosting investor confidence with advanced features and open-source plans.

Key takeaways ・Moonshot AI reportedly accesses roughly 20,000 Nvidia chips through a computing agreement with Alibaba Group. ・The Hopper-generation cluster provides a significant portion of the computing capacity behind Moonshot’s Kimi models.