A Five-Year Plan With an Exact Number Attached
China's Ministry of Industry and Information Technology posted its 15th Five-Year Plan for the information and communications industry publicly on September 7, 2026, a document dated August 12 and covering the 2026 to 2030 period. It sets a target of 9,800 exaflops of intelligent computing capacity by 2030, alongside 3.8 trillion yuan, roughly 532 billion dollars, in cumulative information infrastructure investment over the same five years. That is not a research ambition or a slogan. It is a number written into a government plan, with a spending figure attached to it.
The starting point makes the target legible. China's intelligent computing capacity had already reached 2,185 exaflops at FP16 precision by the end of June 2026, up 177 percent from a year earlier, with rack utilization across national computing facilities running at 71.4 percent. Hitting 9,800 exaflops means roughly quadrupling that figure inside four years.
| Metric | June 2026 | 2030 target |
|---|---|---|
| Intelligent computing capacity (FP16 exaflops) | 2,185 | 9,800 |
| Rack utilization, national computing facilities | 71.4 percent | not specified |
| Cumulative infrastructure investment, 2026-2030 | - | CNY 3.8 trillion (about USD 532 billion) |
Betting on Volume, Not on Matching Nvidia Chip for Chip
The hardware meant to carry that growth is domestic. Huawei is building on its existing Atlas 900 A3 SuperPod with a new Atlas 950 SuperPod, linking 8,192 of its own Ascend chips to deliver 8 exaflops of FP8 performance, backed by 1,152 terabytes of memory and 16.3 petabytes per second of interconnect bandwidth. Read against the 9,800 exaflop target, that is the shape of the plan: scale one homegrown chip family fast and repeatedly, rather than prove a next-generation process node matches Nvidia's roadmap on a chip-for-chip basis. It is a volume bet, not a technology-leadership bet, and it tells anyone tracking the AI hardware race that cost curves, not raw benchmark scores, are what China is optimizing to win.
What This Means Outside China
For a business with any China-facing supply chain or data center footprint, the practical read is that two AI hardware ecosystems are now hardening on separate five-year clocks: one built around Nvidia and TSMC's roadmap, the other around Huawei's Ascend line and China's own foundry capacity. The European Union's own Chips Act carries a comparable end-of-decade marker, a target of 20 percent of global semiconductor production by 2030, which gives European planners a rough timeline to weigh against Beijing's. Anyone budgeting AI infrastructure into the next decade should plan for those two ecosystems to keep diverging rather than reconverge.
Servola Journal
We do this for everyone trying to keep up with what technology is doing to our lives. The people who build it, and the people it happens to. The Servola Journal exists so that what we learn belongs to all of them.
Nobody pays us for this. No ads, no paywall, free to everyone. We just believe that understanding what's happening to all of us shouldn't depend on who can afford to pay for it.
If it gave you something today, tell us to keep going. Follow us, leave a like, or write a positive comment. We read every one, and they are what keeps us going.
Read next: Brazil Will Not Pick a Side in the AI Chip War | How 74 Banned AI Servers Reached China Anyway
