What changed
Qianwen Office launches Qwen3.8-Flash with 2x faster generation and lower token consumption. Zhipu GLM-5.3-Flash slashes price 97.5%—flash is the new price war.
Tsinghua-incubated Xijian Technology raises nearly 500M yuan in a year for brain-inspired chips. Neuromorphic silicon is still a Chinese priority.
Siemens and Doubao at Work both push enterprise agents beyond thin LLM wrappers. Industrial AI is selling workflow, not chat.
So what? Jensen grabs model distribution, Chinese labs race to near-free inference, and agents drive token demand. The moat is shifting to usage and cost.