China advances AI infrastructure, chips and agent deployment
China’s AI compute sector saw new infrastructure designs, expanded supercomputing services, policy support for semiconductor capabilities and higher planned equipment spending.
• Huawei unveiled what it called the world’s first 3D data center in Wuhu, Anhui, designed to improve safety for large AI computing centers and compatibility across generations of AI chips.
• China’s National Supercomputing Internet now covers more than 170 computing scenarios, with over 500 agents and more than 10,000 skills online, according to Sugon Intelligent Computing.
• Xingyun Technology plans to raise its 2026 server equipment and parts procurement budget to as much as 20 billion yuan, from a previous ceiling of 7 billion yuan.
• The Ministry of Industry and Information Technology and National Development and Reform Commission issued a five-year plan calling for stronger advanced-process capabilities, breakthroughs in core chips for high-end phones and high-performance PCs, and wider deployment of domestic operating systems.
• Lenovo ranked first worldwide by x86 server shipments in the second quarter of 2026, shipping 286,000 units, while its x86 server revenue rose 96% year on year to $8.26 billion, according to IDC data.
• China’s Internet Basic Corpus 4.0 was released with 120GB of data for large-model training and AI development through the Chinese Internet Corpus Resources Platform.
• DeepSeek engineer Liu Shengyu said he will continue developing operators while using AI agents to help write them, and Baidu reportedly recruited former Google DeepMind engineer Qiyin Wu to work on large-model pretraining.
• ByteDance said it will increase investment in the enterprise market after integrating Doubao, Feishu and Volcano Engine, while more than 90% of Feishu’s new customers in the first half of 2026 purchased its AI products.