Analyst argues labs already run Blackwell stacks, with 3-6x smuggling premium
teortaxesTex · x · 2026-11-13
teortaxesTex infers that leading labs already have functioning Blackwell training stacks — citing DeepSeek's TileLang/DeepInfer releases, which explicitly support Blackwell and Ascend but not Hopper.
He believes the 'multi-billion expenditures' are real but carry a 3-6x smuggling premium; without it, spending would already be in the $200B active regime. He finds the effective scale disappointing relative to headline numbers.
More from Infra
- Perplexity's Mac app shows a local models option, hinting at offline capability — dnlkwk · 2026-11-13
- 2-bit DeepSeek V4.1 Flash runs a coding-agent seat on two 2016 V100s: a 12-hour real-work trial — a4mula · 2026-11-13
- IBM QDC 2026 Unveils Nighthawk r2 Quantum Chip and Fault-Tolerant Dev Tooling — MJBiercuk · 2026-11-13
- Why Jensen Huang Splits From US Frontier Labs on Open Source and Export Controls — morganlai · 2026-11-13
- SGLang-Neuron v0.1.0 open-sourced: SGLang serving stack now runs on AWS Trainium — ying11231 · 2026-11-13
- Used GPU prices on eBay ranked by $/GB: P100 at $10.9/GB, V100 servers nearly doubled in weeks — MzCWzL · 2026-11-13