News
ClustersLumen’s Rob Ward: 'The AI factory has no walls'NeocloudsAWS Lambda now supports GPU instances for Event Functions on Lambda Managed InstancesMinersBitcoin Miners Are Selling Their Stacks to Fund the AI BoomClustersDDN picks up AMPHIX edge AI Factory-in-a-Box winClustersmicroagi trains robots for European industry at Telekom’s AI factoryDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanSovereignNeysa Deploys AMD GPU Cluster in Mumbai for AI Training and InferenceInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersWhat Does Hewlett Packard Enterprise (HPE) Gaining A Top 14 AI Supercomputer Mean?ClustersLumen’s Rob Ward: 'The AI factory has no walls'NeocloudsAWS Lambda now supports GPU instances for Event Functions on Lambda Managed InstancesMinersBitcoin Miners Are Selling Their Stacks to Fund the AI BoomClustersDDN picks up AMPHIX edge AI Factory-in-a-Box winClustersmicroagi trains robots for European industry at Telekom’s AI factoryDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanSovereignNeysa Deploys AMD GPU Cluster in Mumbai for AI Training and InferenceInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersWhat Does Hewlett Packard Enterprise (HPE) Gaining A Top 14 AI Supercomputer Mean?
EveryNeocloud

🇨🇳 Providers of China

11 providers from China, best known first.

China built up supercomputing from the 1980s, reaching the top of the TOP500 list with Tianhe-1A in 2010 and later Sunway TaihuLight. Machines are concentrated in National Supercomputing Centres serving universities, government and firms. US export controls since 2021 have restricted access to advanced chips and technology.

  • Yinhe-1, China's first operating supercomputer, ran from 1983
  • Tianhe-1A topped the TOP500 list in November 2010
  • Sunway TaihuLight reached 93 petaflops peak in 2016
  • 109 Chinese systems on TOP500 in November 2015, up from 37 in May
  • Shenzhen Phase II adds 116,800 square metres, due 2025

AI compute in China

China's supercomputing effort centres on a network of National Supercomputing Centres that lend computing power to universities, government departments and technology companies. The Computer Network Information Center of the Chinese Academy of Sciences supports this at national level through the China National Grid, a network of linked supercomputers. Growth in the first two decades of the 2000s came largely from buying non-Chinese parts, though domestic processors such as Sunway's manycore chips later appeared in leading machines. Identified weak points include missing core technologies, thin application software, dependence on imported commercial packages, an unsustained development environment and a shortage of skilled people.

History

Work began in the 1980s, when planning and science commissions plus the World Bank backed a project covering networking and three supercomputer centres. Yinhe-1 ran from 1983. Rankings climbed steadily: 43rd in November 2002 with DeepComp 1800, then 11th, then 10th, then first in November 2010 with Tianhe-1A. Sunway TaihuLight took the top spot in 2016 at 93 petaflops. Early domestic chips included the Loongson CPU, a MIPS-style design licensed properly only after 2007.

Providers

Lenovo grew into a major international supplier of supercomputing systems. Tianhe-1 and Tianhe-2 were assembled with Intel CPUs and Nvidia GPUs, while Sunway TaihuLight relied on home-grown manycore processors. In November 2015, Chinese organisations supplied 88.1% of the TOP500-class machines installed in the country, yet 108 of 109 systems still ran American processor technology from Intel, AMD and IBM.

National programmes

The 863 Programme supported the Tianhe-1 project begun in 2008. National Supercomputing Centres were approved from May 2009 onward, backed by the Ministry of Science and Technology together with provincial and city governments. In May 2024 the Jinan centre launched the Shandong Computing Network, described as the first supercomputing internet project in China, spanning 16 cities.

Data centres and power

Shanghai Supercomputer Center, founded in 2000, was the country's first public high-performance computing facility and reached about 3.3 PFLOPS by 2020. Shenzhen's Phase I site covers 43,400 square metres; Phase II in Guangming Science City adds 116,800 square metres of floor area and is due to hold a 2E-level machine. Tianjin, Changsha, Wuxi, Jinan, Guangzhou, Zhengzhou, Kunshan and Chengdu host further national centres.

Today

Since 2022 Chinese institutions have cut back and then stopped submitting new systems to the TOP500 ranking, and their share of leading positions has dropped. Unlisted exaFLOPS-class machines include OceanLight and Tianhe-3, also called Xingyi. Tianhe Xingyi was unveiled in Guangzhou in December 2023, reported to beat Tianhe-2 on CPU power, networking, storage and applications. US export controls widened in March 2025 to more than 50 further China-based firms.

Where this page comes from

The facts in this overview come from the Wikipedia article "Supercomputing in China", written by its contributors and used under the CC BY-SA 4.0 licence. EveryNeocloud rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.

AI compute hubs in China

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.