A staff member at a data center details information about the construction of the provinces' data center cluster, in Qingyang, Northwest China's Gansu province, on Jan 6. [Photo/Xinhua]
The Gansu provincial government has pledged to accelerate the launch and operational efficiency of key projects in the Qingyang data center cluster in 2026, with total computing power capacity expected to exceed 150,000 petaflops, according to the government work report delivered by provincial governor Ren Zhenhe at the opening meeting of the fourth session of the 14th Gansu Provincial People's Congress on Tuesday.
Qingyang, a city in Northwest China's Gansu province, is rapidly emerging as a national front-runner in computing power development, with its intelligent computing capacity surpassing 100,000 petaflops by the end of 2025 — an important milestone under China's "eastern data, western computing" initiative.
Since the national initiative was launched in February 2024, Qingyang, as one of eight national computing hubs, has leveraged its energy resources, geographic advantages and coordinated policy support to accelerate the construction of large-scale computing infrastructure and foster a complete digital industry chain.
"As one of China's eight national computing power hubs, Qingyang has taken on a leading role in the integrated national computing network by advancing large-scale capacity deployment, green power integration and pilot reforms in computing power monitoring, dispatch and security," said Li Jianguo, deputy head of the department of digital technology and infrastructure development at the National Data Administration at 2025 China Computing Valley Digital Industry Development Conference held in Qingyang on Dec 6.
Industrial clustering has followed. More than 500 digital enterprises have established operations in Qingyang, spanning cloud services, large language models, chip development and AI applications. By deploying training and inference workloads locally, these firms are forming a closed-loop ecosystem that links computing resources, data, algorithms and real-world applications.
"By deploying our core computing power at the Qingyang data center cluster, Capital Online helped to turn the city into a key hub for our global AI services, enabling low-latency cross-border scheduling and allowing China's computing power to support AI applications worldwide," said Yao Wei, CEO of Capital Online Data Service Co Ltd.
The spillover effects of computing power are also extending to cultural and creative industries."With large-scale computing power based in Qingyang, content creators can significantly reduce production costs and break content homogeneity, enabling film, television and digital media industries to move toward data-driven creation and scalable growth," said Wang Ruibing, secretary-general of the Shaanxi Radio and Television Association.
Qingyang is also strengthening regional collaboration to unlock new growth potential. The city has partnered with Ganzhou in Jiangxi province to launch a "digital computing long march-red corpus" initiative, a cross-regional project aimed at digitizing red cultural resources and integrating them into AI training systems.
The initiative seeks to revive historical memory through digital technologies while creating new commercial applications for red culture.
"We're linking Qingyang's computing power with Ganzhou's rich red cultural resources to explore how AI can drive digital transformation in heritage preservation, cultural tourism and content dissemination," said Wang Jiandong, vice-mayor of Ganzhou.
Qingyang's clean energy base, with over 10 million kilowatts of installed capacity, allows data centers to run on more than 90 percent renewable power, supporting a zero-carbon computing ecosystem.
With 13 data centers under construction and nearly 390,000 server racks planned, including a fully homegrown 100,000-card integrated training-and-inference cluster worth 36.6 billion yuan ($5.3 billion), Qingyang is steadily building a high-level digital economy chain, strengthening China's self-reliant computing ecosystem and industrial supply chains.
