China is back on top of supercomputing — and the way it got there is the real story. A system named LineShine, installed at the National Supercomputing Centre in Shenzhen, debuted at No. 1 on the TOP500 — the authoritative global ranking of supercomputers — at the ISC 2026 conference, Network World reported. It is the first China-based machine to lead the list since 2017.
The numbers
LineShine clocked 2.198 exaflops on the High Performance Linpack (HPL) benchmark — the standard test the TOP500 uses — more than 20% ahead of the No. 2 system, the US Department of Energy's El Capitan at Lawrence Livermore National Laboratory. A quick translation: an exaflop is a quintillion (a billion billion) calculations per second; "exascale" computing was a frontier only crossed in the last few years. LineShine reached it with nearly 14 million processor cores.
The twist: no foreign chips, and no GPUs
What makes LineShine pointed, not just powerful, is what's inside it. The machine is built entirely on domestic Chinese hardware — custom LX2 processors, a proprietary interconnect and the Chinese-developed Kylin operating system — with no Nvidia chips and, strikingly, no GPUs at all.
That design is a direct response to export controls. Since 2022, the US has steadily restricted China's access to the most advanced chips — especially Nvidia's GPUs (graphics processing units), the accelerators that power cutting-edge AI — and the tools to make them. By building a chart-topping machine out of ordinary CPUs and sheer scale, China is signaling that it can reach the summit of traditional scientific computing without the hardware Washington is trying to withhold.
Why it matters
Supercomputers are strategic infrastructure: they run climate models, design drugs and weapons, simulate chips, and increasingly train artificial intelligence. Leadership in them is a proxy for the broader US-China technology contest. LineShine is a genuine milestone — independently ranked at the top of the list, not merely claimed.
But the win comes with an asterisk that matters. On benchmarks designed to mimic AI workloads — rather than the classic Linpack math test — LineShine reportedly ranks lower, behind several US systems, according to coverage of the results. That gap reflects a hard truth: for the GPU-heavy work that modern AI depends on, China's domestic chips still trail Nvidia and AMD by a meaningful margin, and the advanced manufacturing needed to close that gap remains squarely in the crosshairs of US controls.
The verdict on export controls
So LineShine cuts both ways for US policy. It shows that export controls have a real effect — they pushed China to design around the restrictions, accepting a CPU-only architecture rather than the GPU-rich machines America's labs favor. But it also shows the controls have limits: they can slow China at the AI frontier without stopping it from leading on raw, general-purpose computing power.
The bottom line is that China has reclaimed a prestigious crown and made a statement about self-reliance — while the part of the race that arguably matters most for the future, the AI-specific chips, is the part where the US still leads and is fighting hardest to keep its edge. As the Al Jazeera coverage framed it, the symbolism is large; whether it translates into AI dominance is the open question.
