← The AI Systems Map
L2 · ChipsData center

Huawei CloudMatrix 384

Ascend supernode answering Nvidia NVL72 with scale-out brute force

CloudMatrix 384 (CM384) is Huawei's rack-scale AI supernode linking 384 Ascend 910C accelerators over an all-optical mesh, presented as China's system-level answer to Nvidia's GB200 NVL72. It trades per-chip efficiency and power for sheer scale-out, and was used to serve DeepSeek and other large models domestically.

Config

384x Ascend 910C

Rival

Nvidia GB200 NVL72

How it fits the stack

Huawei CloudMatrix 384 with what it depends on (above) and what it feeds (below). The figure renders as a crawlable diagram and upgrades to an interactive 3D graph as it scrolls into view.

used byhostscompetes withHuawei CloudMatrix 384ChipsHuawei Ascend(910B/910C)chokepointHuawei CloudNvidia Vera Rubinplatform
Huawei CloudMatrix 384Depends on ↑Related

Huawei CloudMatrix 384 in the AI stack. Huawei CloudMatrix 384 with its immediate upstream dependencies (top) and downstream dependents (bottom) in the AI value chain. Hover a node in 3D, or read the full relationships below.

Graph data (text) — 4 entities, 3 relationships
  • Huawei CloudMatrix 384used byHuawei Ascend (910B/910C)
  • Huawei CloudMatrix 384hostsHuawei Cloud
  • Huawei CloudMatrix 384competes withNvidia Vera Rubin platform