Experts: China's new data group highlights AI infrastructure needs

China's establishment of a new state-owned data group highlights the importance of data infrastructure for artificial intelligence (AI) development, according to Zhang Xianghong, a member of a national advisory committee on data.

The National Data Group was established in Xiongan New Area, Hebei Province, on October 20. The central state-owned enterprise aims to develop and utilize data resources, unlock their value and support AI development. Its stated responsibilities include developing and utilizing these resources and advancing data infrastructure connectivity to support China's digital development. 

Unlocking data value for AI development

Zhang said the group would help integrate national-level data resources, promote data sharing across industries and regions, and break down data silos. These efforts could improve resource allocation and help build a unified national data market.

China established the National Data Administration in 2023. The establishment of the new state-owned data group represents another major development in the country's efforts to advance its data strategy.

Zhang said the two institutions will play complementary roles in promoting the development of the data sector. The National Data Administration is primarily responsible for developing foundational data systems and designing related policies, while China National Data Group will focus on the commercial development, utilization and operation of data resources.

The coordinated efforts of the government and the market are expected to create new opportunities for the data industry, paving the way for deeper changes in the way data resources are managed, shared and utilized.

With data infrastructure becoming increasingly interconnected, more market participants are expected to gain access to high-value data resources at lower costs, Zhang said.

Ensuring reliable AI services

Beyond data access, reliable AI services also depend on sufficient computing capacity, dependable infrastructure and effective system maintenance. These factors can help keep services running and reduce disruptions.

The importance of service reliability was underscored by a partial outage affecting DeepSeek's web chat and API services on October 15. The company's status page did not specify the cause, so the incident cannot be directly attributed to insufficient computing capacity.

Tian Feng, the founding dean of Thinking Fast and Slow Research Institute, an independent AI think tank, suggested that limited computing capacity, traffic-control strategies and resource scheduling might have contributed to operational pressures. He also stressed the importance of testing, staged deployment, traffic management and rollback mechanisms. These were industry assessments rather than an officially confirmed explanation of the outage.

The incident highlights the need for AI providers to strengthen system resilience through improved operations, fault isolation, traffic management and disaster recovery. Businesses may also consider integrating multiple model providers or deploying models locally to reduce their dependence on a single service provider.

Although data resource development and AI service reliability represent different aspects of the industry, both are essential to its sustainable growth. As China advances data market development and expands AI applications, closer coordination among data resources, computing power, engineering systems and real-world use cases will be crucial to translating technological innovation into reliable productivity gains.