DeepSeek clusters 160,000+ Huawei Ascend 950DT chips for inference, still trains on NVIDIA
heypearlai · x · 2026-09-04
DeepSeek is reportedly assembling one of the largest known Huawei AI chip clusters ever, with 160,000+ Ascend 950DT units heading to a new data center in Inner Mongolia.
- The 950DT was built specifically for AI training, yet DeepSeek only uses it for inference — training still runs on NVIDIA
- Huawei's production is a bottleneck: memory shortages cap total output at a few hundred thousand chips this year across all customers
- Filling DeepSeek's order alone could take over a year
A telling signal about top Chinese AI labs' confidence in domestic training silicon.
More from Infra
- Nvidia guides 70% growth citing unmet AI demand, while media panics over a $500M Korean neocloud IPO — firstadopter · 2026-10-21
- Pichai: everyone is massively short on AI compute — 'everyone wishes you had two times what you had planned' — firstadopter · 2026-10-21
- Stanford CS336 Lecture 6: fused Triton kernels cut softmax memory traffic ~4x — stanfordnlp · 2026-10-21
- General Compute partners with Positron AI: Atlas memory-centric hardware ships Q1 2027 — aftahi_ai · 2026-10-21
- AI bond issuance hits $419B in 2026, up 90% from all of 2025 — SumitGup · 2026-10-21
- AI Obsession Is Making Enterprises Neglect Cloud FinOps, Expert Warns — DavidLinthicum · 2026-10-21