Recently, International Data Corporation (IDC) and Inspur Information jointly released the "2026 Assessment Report on the Development of Artificial Intelligence Computing Power in China" (the report). In the city rankings, Beijing, Hangzhou, Shenzhen, Shanghai and Guangzhou took the top five spots.

Guangzhou has become an important supporting hub for the development of the AI industry in the Guangdong-Hong Kong-Macao Greater Bay Area, with its computing power infrastructure continuing to expand. According to the report, Guangdong province, excluding Shenzhen, invested approximately $2.149 billion in AI hardware in 2025, with Guangzhou accounting for the largest share. The Guangzhou Intelligent Computing Operation Service Platform has integrated 21 intelligent computing centers, while 94 large AI models have completed national registration, ranking third nationwide by number. The city is also home to more than 2,200 AI-related companies.

This year, the report for the first time included AI agent deployment, the industrial layout of embodied AI, large AI model development, and inference computing infrastructure among the key indicators for assessing the level of AI computing power development in cities. In other words, a city's computing power is no longer measured simply by how many intelligent computing centers it has built, but also by whether AI agents are actually being put to work.

The driving force behind this shift in the assessment criteria is the explosive growth in computing demand brought about by AI agents. From 2026 to 2030, the number of active AI agents in China is expected to grow at a compound annual growth rate of 151.2%, while token consumption is projected to grow at 3,499.3%. Computing demand is rising accordingly. China's intelligent computing capacity is expected to reach 2,576.5 EFLOPS in 2026, up 87.9% year on year, and is projected to reach 10,524.3 EFLOPS by 2030. The report therefore positions computing power as national strategic infrastructure on a par with water conservancy, power grids and railways.

广深跻身AI算力城市前五

近日，IDC（国际数据公司）与浪潮信息联合发布《2026年中国人工智能计算力发展评估报告》（以下简称《报告》）。在城市排名中，北京、杭州、深圳、上海、广州位列前五。

广州成为粤港澳大湾区人工智能产业发展的重要支撑节点，算力基础设施持续扩容。《报告》显示，2025年广东省（不含深圳）人工智能硬件投入约21.49亿美元，其中广州占比最高；广州市智算运行服务平台已整合21家智算中心，94款大模型通过备案、数量居全国第三，人工智能相关企业超2200户。

今年《报告》首次将智能体应用落地、具身智能产业布局、大模型开发以及推理算力基础设施等纳入城市人工智能算力发展水平的关键评价指标。换句话说，城市算力比的不只是建了多少智算中心，还要看智能体有没有真正跑起来。

推动规则变化的是智能体带来的算力需求爆炸。2026年至2030年，中国活跃智能体数量复合增长151.2%，Token消耗量复合增长3499.3%。算力需求随之水涨船高，2026年中国智能算力规模预计达2576.5EFLOPS（算力性能单位），同比增长87.9%，2030年将达10524.3 EFLOPS；《报告》由此将算力定义为与水利、电网、铁路同级的国家战略基础设施。

文丨记者 沈钊
图丨受访者提供
翻译丨魏薇
审校丨曾敏