No. of Recommendations: 10
https://www.nytimes.com/2026/06/23/technology/chin...
China Takes Supercomputer Crown From U.S. for First Time Since 2017
A supercomputer in Shenzhen was declared the world’s fastest. It uses only standard microprocessors and not the special-purpose chips called graphics processing units.
By Don Clark, The New York Times, June 23, 2026
China took back a coveted computing crown from the United States on Tuesday, ratcheting up a fierce technological competition that has implications for science, national security and geopolitics.
LineShine, a massive computing system in Shenzhen, China, was declared the world’s fastest by a group of researchers using a set of standard tests for supercomputers. Besides raw speed, the system stood out because it uses only standard microprocessors and not the special-purpose chips called graphics processing units, which most high-end supercomputers rely on for heavy number crunching…
China’s LineShine system does not separate the traditional jobs of microprocessors and GPUs, as most high-end systems do. Instead, it builds in GPU-style tasks with specialized circuitry that accelerates matrix and vector calculations. That ability is embedded in chips that have a total of nearly 14 million computing cores, or tiny electronic brains, installed in 90 hardware cabinets.
These chips are an original design based on a set of instructions licensed by Arm Holdings, a British company that is controlled by the Japanese conglomerate SoftBank. Arm’s technology is best known for powering smartphones but has lately been adapted by Nvidia, Amazon, Qualcomm and others for use in data centers… [end quote]
What were the steps in designing and building this supercomputer?
1. Who designed the specialized CPU chips? The custom CPU is a Chinese design named the LX2 processor running on a platform called “LingKun.” The design incorporates Armv9 architecture coupled with proprietary Chinese engineering. It features an unprecedented 304 cores per chip, allowing the system to achieve a massive total of nearly 14 million computing cores.
2. The instructions were licensed by Arm Holdings, a British company.
3. Arm Holdings is controlled by the Japanese conglomerate SoftBank.
4. LineShine’s designers, who are supercomputer veterans in China, have not disclosed details about which company manufactured the chips or the level of chip production technology used. Were they made by TSMC? Does China have a fab that can produce an unlimited number of these specialized CPUs?
5. The LineShine system was supposedly made without government funding. Who owns it? It is owned and operated by regional public/state-affiliated computing centers in Shenzhen.
6. How do the costs of this system compare with a comparable supercomputer with GPUs? While exact dollar figures remain undisclosed, a CPU-only supercomputer of this scale is incredibly expensive to build and operate compared to a GPU-reliant one.
7. What are the implications for NVIDIA and other U.S. chip makers? For generative AI, NVIDIA’s specialized Tensor cores are still vastly superior. LineShine shows that China can build massive machines, but they are specialized for engineering/physics calculations rather than modern AI training.
8. What are the implications for U.S. computer manufacturers? LineShine’s success will force U.S. manufacturers and the Department of Energy to rethink architectural diversity and ensure that massive, CPU-heavy parallel architectures are not entirely ignored.
9. What are the implications for U.S. national security? Supercomputers are heavily utilized for simulating nuclear weapons stockpiles, cryptanalysis, aerospace design, and hypersonic weapon modeling. LineShine proves that China possesses the independent, domestic computing power to run these advanced national security simulations at speeds faster than any U.S. lab.
10. Which is more useful for practical purposes, generative AI or focused AI like the Chinese system?
If your practical goal is to write a piece of software, draft a business strategy, or summarize data, generative AI is infinitely more useful.
However, if your goal is to design a breakthrough jet engine, predict a hurricane’s path, or simulate a nuclear reaction, generative AI is useless because it “hallucinates” and guesses numbers. For those tasks, the brute-force, high-precision mathematical certainty of a CPU-heavy supercomputer is the only tool that works.
Wendy
