Ascend output could hit ~20M chips next year; memory and packaging, not logic, are the bottleneck
teortaxesTex · x · 2026-11-19
tphuang estimates Huawei Ascend production: each 950DT uses 2× 400-430mm² dies, the 960DT uses 4; a 12-inch wafer yields 130+ dies, at 60-70% yield 80-90 good dies. With 20k wpm on N+2 next year, that's 19.2-21.6M chips — 9.6-10.8M if all 950DT, 4.8-5.4M if all 960DT. The 960DT beats two 950DTs on interconnect, HBM and supernode size. 5-7nm capacity may ramp to 100k wpm next year. teortaxesTex adds that SMIC can saturate logic die supply with a fraction of throughput: the real constraint order is Memory >>> Packaging >>> Logic.
More from Infra
- AMD Loom boosts Qwen3.5-4B prefill up to 2.6x on Radeon 890M laptops — jfowers_amd · 2026-11-19
- Dev submits PR adding periodic checkpoints to GKE sandboxes so agents don't lose progress — yoimnotkesku · 2026-11-19
- AWS Lambda Adds GPU Support via Managed Instances for Inference and Media Workloads — DavidWells · 2026-11-19
- Nvidia Allegedly Made GPUs Artificially Scarce to Keep Prices Sky-High — GaryMarcus · 2026-11-19
- AWS adds private CA support to Bedrock AgentCore Identity for connecting agents to self-hosted IdPs — AWS ML Blog · 2026-11-19
- Leak Hints at Classified 1M-GPU Training Cluster in China That Skips HBM — teortaxesTex · 2026-11-19