Built without Nvidia chips, China's LineShine supercomputer just dethroned America's most powerful El Capitan

China has reclaimed the top position on the world's most authoritative supercomputer rankings, with a domestically built machine powered entirely by homegrown chips surpassing its American rival for the first time since 2017. The achievement signals a significant shift in the global technology rivalry between Beijing and Washington, and raises fresh questions about the effectiveness of US export restrictions on advanced semiconductors.

What Is the LineShine Supercomputer and Where Is It Located?

The machine at the centre of this milestone is called LineShine, housed at the National Supercomputing Centre in Shenzhen, China's southern technology hub. It has displaced El Capitan, located at the Lawrence Livermore National Laboratory in California, from the summit of the TOP500 list, the biannual global ranking that benchmarks the world's most powerful supercomputers.

According to the latest TOP500 results, released on Tuesday at an award ceremony in Hamburg, Germany, LineShine recorded a computing speed approximately 20 per cent faster than El Capitan. The ranking evaluates machines on their ability to handle complex, high-speed computations across scientific and engineering domains.

China Builds Its Supercomputer Without Nvidia GPUs

What distinguishes LineShine from most of its competitors is its architecture. The machine relies entirely on central processing units, or CPUs, the conventional computing chips found in everyday consumer electronics, rather than the graphics processing units, known as GPUs, that power the vast majority of today's leading supercomputers.

GPUs have become the dominant engine of modern high-performance computing and artificial intelligence, with American firms such as Nvidia holding a commanding position in their supply.

Since 2022, Washington has enacted successive rounds of export controls specifically targeting the transfer of advanced GPUs to China, seeking to slow Beijing's progress in artificial intelligence and military computing.

LineShine's chief designer, Lu Yutong, speaking at the TOP500 ceremony in Hamburg, said the machine broke away from the conventional hybrid architecture that combines CPUs and GPUs. The system, he said, leverages domestically developed, full-stack computing infrastructure, including CPUs and high-bandwidth memory, for scientific, engineering and AI workloads.

China's National Supercomputing Centre described the result in a statement as "the result of breakthroughs across a series of core technological bottlenecks." LineShine's achievements, the centre added, "mark a historic leap for China's supercomputing sector in overcoming foreign technology restrictions and building an independently controlled hardware and software ecosystem."

How US Export Controls Pushed China to Innovate

The development of LineShine fits into a broader pattern of Chinese technological adaptation in response to sustained American pressure. Since the first wave of chip export restrictions under the Trump administration, Beijing has directed considerable resources towards building domestic alternatives to foreign-sourced components.

That strategy produced a notable result in the AI sector last year, when the Chinese startup DeepSeek released a large language model that delivered performance close to industry-leading standards while using significantly fewer advanced chips, a development that caught many in Silicon Valley off guard.

LineShine represents a parallel trajectory in supercomputing. Rather than finding workarounds for restricted components, Chinese engineers appear to have constructed an entirely separate technological pathway, one that bypasses GPU dependency altogether.

Since coming online, LineShine has been used for a range of applications including climate modelling, engineering simulations, drug discovery, neuroscience research and AI development, according to the National Supercomputing Centre.

What the TOP500 Ranking Does and Does Not Measure

Despite the symbolic and technical weight of the result, experts caution against treating the TOP500 position as a direct indicator of overall AI capability or national technological strength.

Andrew Rohl, director at the National Computational Infrastructure in Australia, noted that the TOP500 benchmark was designed decades ago to measure traditional scientific computing workloads and was not built to assess modern AI performance, according to CNN report.

He also pointed out that many of the most powerful AI systems operated by US companies such as xAI and Google, as well as supercomputers run by top defence facilities, do not appear on the TOP500 list, either for reasons of sensitivity or because submitting offers no commercial benefit.

Where the US and Other Nations Stand in the New Rankings

El Capitan retains second place on the updated list. Three other American machines follow closely, with facilities in Tennessee and Illinois both featuring in the top tier, alongside one system based in Germany.

Beyond the leading positions, countries including Italy, Switzerland and Japan maintain a presence in the top ten, reflecting the continued spread of high-performance computing investment across multiple regions.

Supercomputers are used across government, industry and academia for tasks including drug discovery, climate forecasting, AI model training and complex physical simulations.