Sugon
The Chinese Academy of Sciences supercomputing spin-off, Entity-Listed since 2019 and the precedent case for US HPC export controls. Ships Hygon-based systems today; documented for history and as a design reference, not as a procurable NVIDIA vendor.
- Spin-off of Chinese Academy of Sciences supercomputing research, with ongoing joint R&D programs
- Operates 3 intelligent-manufacturing production bases and 5 R&D centers
- Built China's earliest top-tier supercomputers in-house (Nebulae, briefly world's #2 in 2010) on its own blade-system engineering
- Co-develops CPU/DCU server platforms with affiliate Hygon, including cold-plate liquid cooling (PUE 1.08) and ScaleFabric interconnect
| Model | Platform | Form factor | Cooling |
|---|---|---|---|
| Sugon C8000 series Hygon C86-5G CPU (128-core) plus Hygon DCU accelerator platform | domestic accelerator | 2U | air or DLC (cold-plate, PUE 1.08) |
| Nebulae (Dawning TC3600) Historical 2010 system, Intel Xeon plus NVIDIA GPUs, briefly world's #2 supercomputer | PCIe | blade system | air |
- Historically built HPC systems on Intel Xeon plus NVIDIA GPUs (Nebulae, 2010)
- Entity-Listed since June 24, 2019: barred from unlicensed US-origin items including NVIDIA GPUs under presumption of denial
- Now ships systems on Hygon C86 CPUs and DCU accelerators positioned as NVIDIA A100-class substitutes
- Beijing (Zhongguancun HQ and R&D)
- 3 intelligent-manufacturing bases across China
- Added to the US Entity List June 24, 2019 (acknowledged military end-uses of its HPC systems); restriction remains in force
- The proposed ~$16.4B Hygon-Sugon merger was called off in December 2025; the two remain tightly linked via cross-shareholding
- Not procurable under normal export-control compliance; relevant as the entity-list precedent that shaped later restrictions
Is Sugon an OEM, ODM or integrator for NVIDIA AI servers?
Sugon sits in the "China domestic (export-restricted)" tier of our index. Spin-off of Chinese Academy of Sciences supercomputing research, with ongoing joint R&D programs
Does Sugon design its NVIDIA servers in-house?
The Chinese Academy of Sciences supercomputing spin-off, Entity-Listed since 2019 and the precedent case for US HPC export controls. Ships Hygon-based systems today; documented for history and as a design reference, not as a procurable NVIDIA vendor.
Which NVIDIA platforms does Sugon ship?
Sugon ships systems on domestic accelerator, PCIe. The systems table on this page lists the specific models.
Where does Sugon manufacture its AI servers?
Beijing (Zhongguancun HQ and R&D). 3 intelligent-manufacturing bases across China