A newly released evaluation report places Beijing, Hangzhou, and Shenzhen as the top three Chinese cities for artificial intelligence computing power. The findings were unveiled at the AICC2026 Artificial Intelligence Computing Conference in Beijing on September 21.
The "2026 China Artificial Intelligence Computing Power Development Assessment Report" ranks Beijing first, followed by Hangzhou and Shenzhen. The remaining top ten positions are held by Shanghai, Guangzhou, Chengdu, Nanjing, Wuhan, Jinan, and Suzhou.
Beijing secures its leading position thanks to a comprehensive computing power supply chain and a strong cluster of model and AI application companies. Hangzhou takes second place, driven by its innovative AI enterprise ecosystem, robust internet industry base, and continued progress in areas such as large language models and embodied intelligence. Shenzhen ranks third as a key manufacturing hub for AI servers and computing chips, leveraging its deep industrial foundation in computing supply to seize opportunities presented by the surge in AI agents.
The report, co-published by International Data Corporation (IDC) and Inspur Information, highlights that AI agents are fundamentally reshaping computing power infrastructure. By creating a multiplier effect that dramatically increases token consumption, these agents have sharply raised the projected growth trajectory of China's intelligent computing capacity. The report elevates computing power to the status of a national strategic infrastructure, alongside water conservancy, power grids, and railways.
According to the report, China's intelligent computing scale is expected to reach 2576.5 EFLOPS (exaFLOPS) in 2026, representing a 87.9% year-on-year increase. This figure is projected to further expand to 10524.3 EFLOPS by 2030. In parallel, the market size for AI computing power in China is forecast to reach $51.5 billion in 2026, up 37% from the previous year, and is expected to grow to $150.1 billion by 2030.
The report also notes that in the AI agent era, infrastructure is shifting from pure scale expansion to competition focused on system efficiency. Through coordinated optimization across computing, storage, networking, data centers, terminals, models, scheduling, and token operations, the core competitive metric is moving from "how much computing power one owns" to "how efficiently one can continuously produce intelligence."
Comments