Cambricon joins PyTorch Foundation as platinum member, first Chinese chipmaker on its board
新智元 · wechat · 2026-09-08
Cambricon has become a platinum member of the PyTorch Foundation, joining its governing board after Alibaba Cloud did so in May — the first Chinese AI chipmaker in the decision layer of the Linux Foundation-hosted body that governs PyTorch, vLLM, DeepSpeed and Ray.
Cambricon pursues an Upstream First strategy, landing code in seven PyTorch core modules (torch.compile, eager ops, device runtime, distributed, AMP, Dataloader, Profiler) and achieving Day0 support for models like DeepSeek-V4 and GLM-5 on vLLM. Its seat-holders are engineers, not execs: AI framework senior director Wang Jin joins the board, and core maintainer Zhu Jing joins the Technical Advisory Council.
The bet: a unified framework interface lowers migration cost, so developers can switch accelerators without rewriting code — shifting Cambricon from adapting others' rules to helping set them.
Related event: Cambricon Joins PyTorch Foundation as Platinum Member(3 posts)→
More from Infra
- FT: CXMT and YMTC stockpiled enough ASML DUV tools for three years of expansion — basedjensen · 2026-09-08
- Nvidia's $59.7B quarterly profit tops 12 iconic companies combined ($57.9B) — FinanceYF5 · 2026-09-08
- Consolidating four small models into one inference server: a doc-QA agent's ops tradeoffs — Sad-Razzmatazz-7657 · 2026-09-08
- Huawei Ascend takes PyTorch China stage: from hardware adaptation to joint standards research — PyTorch · 2026-09-08
- AUR llama.cpp-cuda removal caused 10x slowdown; manual rebuild restored 1800 t/s prefill — MrHall · 2026-09-08
- CPO laser series part 3: MOPA approach vs Lumentum's single-cavity route — vikramskr · 2026-09-08