The Huawei-DeepSeek Symbiosis
Right before China’s national day holidays, DeepSeek released Ascend support for TileLang, DeepGEMM, DeepEP, TileKernels, FlashMLA, and DeepSelect.
Proof1 independent outlet
Key points
- Although they have Pangu, Huawei has no frontier lab to prove their chip can train or infer models, DeepSeek has models but cannot buy the compute it needs.
- September 30’s release of Ascend support in DeepSeek repositories is an artifact of this trade, where the exchange runs through the kernel-authoring language.
- As we show here and in future articles, the Huawei ecosystem is already mature and well developed, with strong foundations from the Ascend time and DeepSeek engineers building further ontop.
- We deep-dived into CANN and the Ascend 950 architecture in our DeepSeek V4 Pro article.
Sources (1)
- [1]The Huawei-DeepSeek SymbiosisSemiAnalysis · Oct 21, 05:34 PM
Right before China’s national day holidays, DeepSeek released Ascend support for TileLang, DeepGEMM, DeepEP, TileKernels, FlashMLA, and DeepSelect.
Although they have Pangu, Huawei has no frontier lab to prove their chip can train or infer models, DeepSeek has models but cannot buy the compute it needs.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 21, 2026Granite Speech 5.0: Transcribe, Translate, and Attribute more Accurately and Faster
- Oct 20, 2026[AINews] DeepSeek DeepInfer: Open Source SOTA Inference running 155T toks/day/cluster making $22m/day at ~80% margin for a 81 day payback
- Oct 19, 2026How Deepgram optimized voice AI at scale with AWS Trainium
- Oct 19, 2026Push-based KV transfer to improve TTFT and TPOT for disaggregated inference in vLLM
- Oct 19, 2026jax-ml/jax jax-v0.12.0: JAX v0.12.0
- Oct 17, 2026deepseek-ai/DeepSeek-V4.1-Flash-1017
