Aggressive Infrastructure Scaling Targets
China has set a national target to expand its intelligent computing capacity to 9,800 eflops by 2030. The Ministry of Industry and Information Technology revealed this objective in a fresh five-year industrial blueprint released on Monday. To reach this benchmark, the nation must grow its current output by more than four times within five years. This policy shift underscores a massive commitment to the hardware that powers artificial intelligence across the country.
The government plans to commit 3.8 trillion yuan, or roughly US$532 billion, toward information infrastructure from 2026 through 2030. Officials are calling for the strategic build-out of computing clusters that house at least 10,000 accelerator cards, with larger installations expected to host 100,000 cards. These facilities will serve as the backbone for both training and inference workloads. Planners also emphasized the need to prioritize domestic chip compatibility within these new systems.
Historical Growth and Current Capacity
Expansion is already happening at a rapid pace. As of the end of June, China’s intelligent computing capacity stood at 2,185 eflops. This figure represents a 177 per cent increase from the previous year. Data from the National Data Administration shows that total capacity grew to approximately 2,450 eflops by the end of July. Much of this growth stems from the existing East Data, West Computing project, which redistributes processing loads to regions with lower energy costs.
Connectivity between these regional hubs remains a central priority. A government-backed platform has linked 155 companies to share resources across 578 distinct pools. This arrangement has successfully aggregated 316 eflops of computing power. Despite this effort, officials note that over 80 per cent of all capacity remains locked within eight specific national hubs. This concentration highlights the challenge of managing hardware across such a vast geography.
Surging Demand and Domestic Chip Adoption
Daily token usage provides a window into the intensity of current demand. Consumption exploded from roughly 100 billion tokens in early 2024 to 140 trillion by March of this year. This surge drives the need for more, faster, and more efficient hardware. Domestic chip makers are seeing the impact of this transition. Sinolink Securities reported that Chinese-designed processors accounted for 41 per cent of shipments during 2025. Still, compatibility issues and fragmented software ecosystems continue to slow down widespread adoption of these local alternatives.
Integration remains a bottleneck for the industry. Companies find that making home-grown hardware work with existing software stacks takes significant time. Beyond the computing clusters, the new policy outlines a clear path for future communications networks. The plan targets the launch of commercial 6G services at an opportune moment. Researchers are currently preparing for the Release 21 standards, which will govern the next generation of global mobile technology. The nation is also looking at space-based computing to extend reach beyond ground infrastructure.
Contextualizing the Infrastructure Push
This push follows years of intense investment in 5G, which left China with 5.1 million base stations by June. That total comprises 69 per cent of the world's 5G infrastructure. With the build-out stage mostly complete, mobile operators are transitioning toward network upgrades and the deployment of 5G-Advanced technology. China Mobile has already brought this faster service to over 330 cities. The move toward 6G trials in the 6GHz band marks the next stage of this shift. What remains certain is that the state intends to maintain its lead in network density while pivoting its primary capital spending toward the hardware required for high-end AI computation.