China’s Five-Year Informatization Plan Pushes Advanced Chipmaking and Embodied AI

The Central Cyberspace Affairs Commission’s 2026–2030 plan sets 12 targets, including lifting smart terminal and AI agent adoption from 40% to over 90%.

WHY IT MATTERS

Chipmakers, data-center operators and AI device makers now have national 2030 benchmarks, including PUE below 1.2 for new large compute facilities and 3.6% R&D intensity in electronics manufacturing.

China’s Central Cyberspace Affairs Commission has issued its digital blueprint for 2026 to 2030, and the chip passage reaches beyond the mature-node capacity that Chinese fabs already run at scale. The National Informatization Plan for the 15th Five-Year Plan Period (《国家信息化“十五五”规划》), published on the CAC website on October 23, also lists compute-in-memory, 3D integration and optoelectronic fusion as technologies to break through and apply, and calls for faster industrialization of quantum technology, brain-computer interfaces, embodied AI and 6G.

The plan tells China’s chip industry to refine mature-process production and raise its capability to manufacture at advanced process nodes. It gives no node, capacity or spending figure for that goal. Its numbers sit instead in a table of 12 main indicators for 2030, every one labeled "expected" (预期性), meaning a guidance target rather than a binding quota. The most aggressive is the adoption rate of new-generation smart terminals and AI agents, which the plan expects to rise from 40% in 2025 to above 90%.

The release notice names the commission as sole issuer and describes the plan as the guide for informatization work by all regions and departments. That reach matters for anyone tracking where ministries and provinces will point subsidies and procurement over the next five years.

The plan sets 12 numbered targets for 2030

According to the full text attached to the CAC notice, the value added of the digital economy’s core industries should reach 12.5% of GDP by 2030, from 10.5% in 2024. Research and development spending intensity in computer, communications and other electronic equipment manufacturing is set to rise from 3.03% to 3.6%. New large and very large computing facilities should run at a power usage effectiveness, or PUE, below 1.2, against 1.25 in 2025; PUE compares a data center’s total power draw with the power its servers use, so lower means less energy lost to cooling and overhead.

Network targets include 50 5G base stations, including 5G-Advanced, per 10,000 people, up from 34, and internet penetration of 85%, up from 80.1%. The IPv6 share of network traffic is set at 42%, from 33.79%. Valid domestic invention patents in the digital economy’s core industries should grow from 2.375 million to 3.266 million, and cross-border e-commerce imports and exports from RMB2.84 trillion to RMB3.5 trillion.

The plan’s qualitative 2030 goals include a national blockchain network that is basically complete and a commercial 6G launch "at an appropriate time." Xinhua, the state news agency, led its English report with the blockchain target and the internet and 5G figures.

Chips sit inside a wider program that runs from AI chips to humanoid robots

The chipmaking sentence appears under the plan’s heading on breaking through key core technologies, next to a call to merge operating systems and software development toolchains with AI. Business daily 21jingji headlined the three chip technologies. Compute-in-memory processes data where it is stored, cutting the traffic between memory and processor that slows AI workloads. 3D integration stacks chips vertically. Optoelectronic fusion combines optical and electronic components, for example to move data with light.

A separate Cyberspace Science and Technology Innovation Breakthrough Project (网信科技创新突破工程) calls for developing high-performance AI chips and a highly available basic software stack, alongside programs for industrial software, quantum computing and communication, brain-computer interfaces, digital twins and bioinformatics. Embodied AI is described as unmanned vehicle terminals and humanoid robots.

On compute, the plan asks for faster construction of a national integrated computing network that coordinates computing power, electricity and networks, and for a better spread of general, intelligent and supercomputing capacity. Regional sections assign integrated circuits to Beijing-Tianjin-Hebei, the Yangtze River Delta, the Greater Bay Area and Chengdu-Chongqing, with the Yangtze River Delta also taking quantum and brain-computer interfaces.

Legislation and capital are the next markers

The plan calls for foundational laws on AI, the digital economy, telecommunications, cybercrime prevention and anti-cyber-violence, and for judicial interpretations on AI and personal information protection. It also asks regulators to improve assessment and filing for AI agents, on-device large models and anthropomorphic interactive services. It sets a separate goal for an AI agent technology maturity evaluation system and identity standards for agents.

On money, an official of the commission’s office said in a Q&A that various fiscal and industry funds will support the plan, and that China will develop "patient capital, long-term capital and strategic capital" and guide financial institutions toward the sector. No amounts were given. The plan arrived the same day as a MIIT-led five-year plan for industrial foundation capabilities, which also left its 2030 goals without published numbers.

The office said in the Q&A, which The Paper also carried, that it will monitor implementation and regularly publish national informatization development reports. Those reports are where the agent-adoption and PUE figures will first be measured against their 2030 marks, and where a draft AI law, if one advances, is likely to be tracked.