China’s AI Infrastructure: Building the Digital Backbone

China's Ministry of Industry and Information Technology (MIIT) has released a new three-year plan for integrating artificial intelligence with the country's information and communications sector. Framed as the "AI + Information and Communication" initiative, the document lays out targets for intelligent network operations, expanded computing infrastructure, AI-enabled telecom services, and more than 30 high-value application scenarios by 2028.

Illustration by The Geostrata

AI THE INFRASTRUCTURE BENEATH

Telecommunications networks, computing systems, industrial applications, and digital services appear throughout the plan alongside AI development itself. The policy spends as much time discussing the movement of data, computing capacity, and network operations as it does discussing artificial intelligence.

AI is presented here as part of a larger digital system rather than a sector operating on its own. 

China's digital infrastructure already operates at considerable scale. The country now hosts more than five million 5G base stations, while intelligent computing capacity has reached 1,882 EFLOPS, a measure of the processing power available for training, deploying, and scaling AI systems across industries.

Existing networks have also deployed 400 Gbps long-distance transmission capabilities, creating the foundations required for more demanding AI applications. AI models occupy surprisingly little space in the document. Much of it is concerned with the systems underneath, such as networks, computing channels, transmission capacity, and scheduling. Much of the attention instead falls on the machinery that allows those applications to function in the first place. 

The document outlines seventeen policy tasks grouped under four headings: industry upgrading, AI development foundations, integrated applications, and industry governance. Similarly, Infrastructure construction, industrial applications, governance mechanisms, and technological development are not treated as separate policy tracks.

Instead, they are presented as mutually dependent parts of the same ecosystem. Progress in one area is expected to support progress in another. This reflects a planning approach that places coordination alongside innovation, particularly in sectors where multiple technologies and regulatory systems must evolve simultaneously.

The distribution here has no single section that dominates the document. Infrastructure, regulation, deployment, and technical development appear side by side, reflecting how policymakers are approaching AI as a systems problem rather than a standalone technology challenge. 

BRINGING COMPUTING CLOSER

Part of the effort centres on reducing the distance between computing power and users. Authorities aim for at least 75 per cent coverage of one-millisecond-latency access in metropolitan areas by 2028, alongside new investments in 400 Gbps and 800 Gbps transmission networks.

The objective is not simply faster connectivity. It is to make computing resources available where they are needed, without delays created by network congestion or physical distance between users and data centres.

The plan also envisages a layered computing system linking national hubs, regional centres, and edge nodes. This resonates with the efforts to build a nationally integrated computing network.

Large data centres and computing hubs remain important, but not every task requires processing in distant facilities. This also reflects an existing geographical imbalance in China's computing infrastructure, where much of the country's major data centre capacity is concentrated in the eastern provinces while demand for computing resources continues to expand nationwide.

Building additional computing capacity in the western regions and linking it through integrated computing networks is intended to ease pressure on eastern hubs while reducing latency and improving access for users across different parts of the country.

Some applications, specifically those involving industrial automation, autonomous systems, or real-time decision-making, benefit from computing resources located much closer to the point where data is generated. The layered structure attempts to balance efficiency, cost and response time across different levels of the network. Rather than concentrating capacity in a handful of locations, the intention is to distribute it across the network.

Processing power can therefore move closer to where data is generated, whether in factories, logistics systems, autonomous machines, or consumer devices. Such arrangements become more useful when computing tasks need to be handled close to where data is generated rather than routed through distant facilities. Research on AI-driven network architecture occupies a substantial section of the plan. Resource scheduling, multi-agent communication systems, and collaboration between large and small models all appear in the document.

These discussions imply that communications networks are expected to perform functions beyond simply carrying information. Networks have traditionally moved information from one point to another. Here, networks are expected to assume a greater role in resource allocation, traffic management, and operational optimisation.

FROM TELECOM NETWORKS TO FACTORY FLOORS

AI-powered mobile devices, intelligent assistants, smart home systems, wearable technologies, and industrial applications are among the policy's prominent target areas. It also encourages telecommunications firms to incorporate AI into traditional services.

The role assigned to telecommunications operators is notable because they already sit at the centre of China's digital infrastructure. Unlike many technology firms that focus on software products, telecom operators manage the networks through which data, cloud services, and digital applications move.

The document, therefore, treats them not simply as service providers but as participants in AI deployment.

Their existing infrastructure offers a ready channel through which AI-enabled services can reach households, businesses, and public-sector users. Machine-vision inspection and precision equipment control already feature in Chinese factories and industrial settings. Both are mentioned as examples of how AI can be used beyond chatbots, assistants, or consumer applications.

References to industrial applications, machine-vision inspection, and equipment control appear throughout discussions of AI deployment, suggesting that deployment in manufacturing remains a major area of attention.

Embodied intelligence also appears among the sectors identified for further development. In Chinese policy discussions, the term is commonly used for robots, autonomous equipment, and AI systems capable of interacting with physical environments.

Its inclusion alongside communications infrastructure is interesting because it places machines, sensors, networks, and computing resources within the same policy framework. The boundary between digital infrastructure and physical systems becomes less distinct in that arrangement. Several sections acknowledge that integration remains difficult. Ministry officials described the convergence of AI and communications technology as a highly complex and systematic process.

Challenges persist in key technological breakthroughs, integration pathways, and business model development. The document also devotes a separate section to governance, covering intelligent regulation, network security, anti-fraud systems, standards development, and public service capacity.

The policy, therefore, functions not only as a development plan but also as an attempt to coordinate different parts of the digital ecosystem around a shared direction. 

SETTING RULES BEYOND THE NETWORK

The policy also links domestic deployment with participation in international standard-setting. It calls for strengthening the formulation of "AI + Network" standards in areas such as 5G-A, 6G, and optical network technologies. The objective thus is not limited to expanding AI adoption at home; it also reflects an interest in shaping some of the technical rules and frameworks that may govern future communications infrastructure.

The timeline extends beyond 2028. By 2030, China aims to achieve breakthroughs in core technologies connecting AI and communications networks while improving integrated sensing, communications, computing, and intelligent service capabilities. The document also envisages a more complete industrial ecosystem supported by stronger collaboration between infrastructure providers, technology firms, telecommunications operators, and industry users. 

Discussions around artificial intelligence often revolve around models, algorithms, and headline-grabbing breakthroughs. Here, Beijing is placing equal emphasis on the foundations beneath AI as on the technology itself.

Networks, computing power, transmission capacity, intelligent scheduling, and deployment mechanisms occupy much of the document. Beijing appears less interested in treating AI as a separate industry and more interested in folding it into existing networks, services, and production systems.

The emphasis is not only on innovation but on the deployment of AI in most sectors. It suggests that China views the next phase of competition through infrastructure, connectivity, and the ability to move AI into the wider economy at scale. 

---

BY SIMRAN

COVERING PEOPLE'S REPUBLIC OF CHINA

TEAM GEOSTRATA

info@thegeostrata.com