Z.ai powers 1GW AI data center entirely on Chinese chips, zero Nvidia

Chinese AI developer Z.ai (formerly Zhipu) has completed construction of a 1-gigawatt data center stocked exclusively with domestically made chips, switching part of it on. The facility will train the GLM model family. Bloomberg reported that Z.ai has built or operates multiple computing clusters holding more than 10,000 Ascend chips each. A gigawatt supplies enough power for roughly 750,000 homes, placing the site among the largest stood up by a Chinese AI lab.

The chip supplier was not named but Z.ai's recent training history points to Huawei. The company released GLM-5.2 in June, trained entirely on Huawei Ascend accelerators with no Nvidia hardware, and it topped open-weight leaderboards within a week. Z.ai was added to the U.S. Commerce Department's entity list in January 2025, cutting off legal access to Nvidia silicon and leaving domestic parts as its only supply line. Z.ai is on track for $1 billion in annual recurring revenue after hitting its 2026 sales target in July.

Huawei's Ascend accelerators trail Nvidia's current Blackwell parts on performance per watt, meaning a gigawatt of domestic silicon delivers less usable training compute than equivalent Nvidia systems. However, the buildout signals China's ability to scale AI infrastructure without U.S. silicon amid export controls. Beijing plans 2 trillion yuan ($295 billion) over five years on a nationwide grid of AI data centers, with at least 80% domestically sourced.