For the first time since 2017, China takes the supercomputer crown from the US

China took back the coveted computing crown from the United States on Tuesday, reigniting a fierce technology competition that has ramifications for science, national security and geopolitics.

LineShine, a massive computing system in Shenzhen, China, has been declared the fastest in the world by a group of researchers using a set of standard tests for supercomputers. In addition to raw speed, the system excelled in using only standard microprocessors and not the special chips called graphics processing units that most high-end supercomputers rely on for heavy number crunching.

This basic proposal could point to a better way to connect artificial intelligence with traditional scientific tasks, said Jack Dongarra, organizer of the so-called Top 500 list of the most powerful supercomputers in the world.

Dr. Dongarra, a professor of computer science and electrical engineering at the University of Tennessee, recently toured the new machine at the Shenzhen Cloud Computing Center. LineShine’s test results were more than 20 percent faster than El Capitan, a system at Lawrence Livermore National Laboratory in California that has topped the biannual supercomputer performance rankings since November 2024. China has not topped the list since 2017.

“It’s an impressive system,” said Dr. Dongarra about LineShine. “They elevated us by developing a system that is not dependent on the GPU.”

A new supercomputer adds to the race between China and the United States for technological supremacy. American tech giants such as OpenAI, Anthropic and Google have developed leading AI models, while another American company, Nvidia, has become the world’s dominant supplier of AI chips. China has tried to innovate in a variety of ways, with Chinese startup DeepSeek last year releasing a cutting-edge AI model using only a tiny fraction of specialized AI chips.

To prevent China from catching up, President Trump has imposed tariffs and at times restricted exports of AI chips. However, China’s use of standard microprocessors, known as CPUs, rather than GPUs, to create an ultra-fast supercomputer suggests a potential way around these hurdles.

“The US government should have tighter controls on the export and production of CPUs for the Chinese market,” said Jimmy Goodrich, senior fellow at the California Institute on Global Conflict and Cooperation. “It’s a loophole in the current regulations.”

Supercomputers, the term for the largest machines dedicated to science, have been used since the 1960s for tasks such as creating climate models, breaking codes and designing nuclear weapons. They usually use high-precision math, expressing numbers using 64 bits of data.

Commercial AI systems from companies like Google and OpenAI, on the other hand, can be even faster. They can use approximations for tasks like identifying pictures or choosing the next word in a sentence, relying on so-called four-bit and eight-bit numbers that allow systems to perform many simpler calculations at once.

“It’s remarkable and impressive what China has done here, but they can’t hold a candle to these massive AI supercomputers that have been built by US AI labs” and others, Mr Goodrich said.

US national laboratories, which are major buyers of some of the largest supercomputers, are eager to use artificial intelligence to speed up aspects of their scientific work. So they accept more of these less accurate calculations along with the 64-bit variant.

Although American companies have historically dominated the list of largest supercomputers, foreign systems have sometimes risen to the top. The system in Japan, for example, topped the list from 2020 to 2022.

“There’s a lot of talk about America being the only country capable of these systems,” said Addison Snell, an analyst at Intersect360 Research, a firm that tracks the sector. “Then you find out that other companies have the capabilities, too.”

Powerful systems from China and Japan have regularly prompted the Department of Energy and other US agencies to push for more funding for supercomputers. In November, the Trump administration began Genesis missionwhich aims to use supercomputers at US national laboratories alongside private companies to support artificial intelligence and scientific research.

GPUs, primarily developed by Nvidia and Advanced Micro Devices, have been a critical weapon in the recent supercomputer race. These chips excel at performing many tasks simultaneously, including the so-called vector calculations used in science and the matrix multiplication at the heart of many artificial intelligence tasks.

When U.S. officials restricted China’s access to GPUs and other powerful chips, as well as limiting exports of some of the most advanced semiconductor machines, it caused it to “invest in developing architectures and technologies to effectively have supercomputers that are on par with the most powerful systems in the U.S.,” said Dr. Dongarra.

China’s LineShine system does not separate the traditional microprocessor and GPU tasks, as most high-end systems do. Instead, it embeds GPU-style workloads with specialized circuitry that accelerates matrix and vector calculations. This capability is built into chips that have a total of nearly 14 million computing cores, or tiny electronic brains, installed in 90 hardware boxes.

These chips are an original design based on an instruction set licensed by Arm Holdings, a British company controlled by the Japanese conglomerate SoftBank. Arm technology is best known for powering smartphones, but has recently been adapted by Nvidia, Amazon, Qualcomm and others for use in data centers.

Arm has been active in China for a long time. “Arm operates worldwide, including in China, in compliance with applicable export control laws and regulations,” a company spokeswoman said.

LineShine’s designers, who are veteran supercomputers in China, did not disclose details about which company made the chips or the level of chip-making technology used, said Dr. Dongarra.

He and other experts have long thought China has systems capable of ranking first, but labs there have not recently submitted test results.

“It doesn’t surprise me that there is a Chinese machine capable of being No. 1,” Mr Snell said. “The surprise is that they wanted confirmation.

Dr. Dongarra, who wrote a detailed report on the new system, was told during a visit to China that the system was created without government funding, so the designers felt it was permissible to submit tests for Top500 evaluation, he said.

The Shenzhen scientists also sought recognition for the new machine through 14 submissions to the Gordon Bell Prize, which supports solving sophisticated problems in science, said Dr. Dongarra. Three systems are finalists for this award and three for a related award in climate science.

According to a report by Dr. Dongarry has used LineShine for projects such as a sophisticated simulation of the Earth, including atmospheric, ocean, land and ice components, as well as a comprehensive simulation of the human brain.