China's Ministry of Industry and Information Technology (MIIT) has unveiled a sweeping five-year plan targeting 9,800 eflops of intelligent computing capacity by 2030, backed by a cumulative information infrastructure investment of 3.8 trillion yuan (US$532 billion) — signalling Beijing's most ambitious AI infrastructure push to date. The plan, released on Monday, 8 September 2026, calls for the deployment of massive AI computing clusters and a sharper focus on domestically produced chips.

What the plan mandates

The MIIT blueprint calls for the 'orderly deployment' of intelligent computing clusters equipped with 10,000 graphics processing cards, as well as larger installations using 100,000 or more accelerator cards. It also mandates the build-out of inference computing facilities tailored to specific application scenarios. Critically, the plan pushes for greater adaptation of computing infrastructure to home-grown chips, a direct response to ongoing US export restrictions on advanced semiconductors.

The 3.8 trillion yuan investment envelope covers the full 2026–2030 period and spans broader information infrastructure, not just AI compute. This positions the plan as one of the largest state-directed technology infrastructure programmes in history.

Where China stands today

According to MIIT, China's intelligent computing capacity reached 2,185 eflops by the end of June 2026, representing a 177 per cent year-on-year increase. The country had already built 52 intelligent computing facilities each equipped with more than 10,000 accelerator cards. By the end of July 2026, capacity had climbed further to approximately 2,450 eflops, according to the National Data Administration, an agency overseen by the National Development and Reform Commission.

To meet the 2030 target of 9,800 eflops, China would need to more than quadruple its June 2026 capacity level — a formidable but not unprecedented pace given the 177 per cent annual growth rate already recorded.

Building on East Data, West Computing

The new targets extend the momentum of the 'East Data, West Computing' project launched in 2022, which redirects power-intensive computing workloads from densely populated eastern regions to western areas offering cheaper land and abundant energy. China has since built a national network anchored by eight national computing hubs, 10 data-centre clusters, and three regions dedicated to coordinating computing resources with power supplies.

The western pivot is strategic: it reduces pressure on eastern power grids while enabling large-scale, energy-hungry GPU clusters to operate at lower cost — a necessary condition for the 100,000-card deployments the plan envisions.

Why it matters

The plan arrives as China races to close the gap with the United States in AI infrastructure, even as chip export controls limit access to the most advanced foreign accelerators. By explicitly calling for infrastructure adaptation to domestic chips, Beijing is effectively mandating a parallel compute ecosystem — one that could insulate its AI sector from future supply shocks. Industry analysts note that the 9,800 eflops target, if achieved, would represent one of the largest concentrations of AI compute capacity anywhere in the world.

What's next

Execution will hinge on the pace of domestic chip development, grid expansion in western provinces, and whether private cloud and telecom operators — including China Mobile — align their own capex cycles with state targets. The trajectory of US–China semiconductor restrictions will also determine how much of the new capacity can be built with leading-edge hardware versus domestically engineered alternatives. Investors and policymakers alike will be watching quarterly capacity updates from the National Data Administration as the clearest near-term signal of whether the 2030 goal remains on track.