China launches largest AI4S computing cluster with 60,000 chips

A new supercomputing cluster capable of handling the most demanding scientific AI tasks has gone live in Central China's Henan province, marking a major step forward in the country's push to integrate artificial intelligence with scientific research.

The cluster, announced by Dawning Information Industry, also known as Sugon, consists of 60,000 specialized processors designed to accelerate AI calculations. It is currently operating at the national supercomputing internet core node in Zhengzhou.

This is the largest facility of its kind in China dedicated specifically to "AI for Science", or AI4S — a field where AI models assist scientists in solving complex problems in areas like drug discovery, new materials and clean energy.

The new cluster does not operate in isolation. It is connected to China's national supercomputing network, a nationwide network that links supercomputers across the country, removing technical hurdles that used to make it difficult to share computing resources.

Together, the national supercomputing network now provides access to over 3 million traditional computing cores (CPUs) and more than 200,000 GPUs — the largest AI4S infrastructure in China.

The national supercomputing network was constructed under the guidance of the Ministry of Science and Technology, granting access to university researchers, institutes, and even small companies to tap into world-class computing power without having to build their own expensive data centers.

One of the most significant achievements, according to Sugon, is a homegrown high-speed network technology called scaleFabric. It is China's first domestically developed "RDMA" network — a technical term that simply means data can move between processors extremely fast, with very little delay. Before this, only a few foreign companies dominated this critical technology.

The cluster uses six different types of self-developed core chips, and the company says its overall performance is on par with the best in the world.

This year is seen as the turning point for AI4S — moving from lab experiments to real-world industrial applications. China's 2026 government work report listed ultra-large smart computing clusters as a national priority under the "AI plus" initiative.

Industry data shows the potential is enormous. According to QYResearch, the global AI for Science market was worth about $4.54 billion in 2025 and could reach $26.23 billion by 2032. More strikingly, AI4S can potentially support nearly $11 trillion worth of downstream industries — including chemicals, medicine, new energy, alloys, displays and semiconductors.