UBS Lead Time Map: Foundries Take 3-4 Years While GPUs Ship in 6-12 Months
AccBalanced · x · 2026-10-04
A UBS research note circulating on X lays out lead times across the AI compute supply chain:
- Foundry: 3-4 years
- Semicap equipment: 1-2 years
- Advanced packaging/substrates: 1-1.5 years
- Storage systems: 1-2 years
- GPUs/AI accelerators: 6-12 months
- Memory (HBM/DRAM): 6-12 months
- Optical transceivers/networking: 3-9 months
- CPUs: 3-6 months
- Servers: 1-2 months
The takeaway: the deepest bottlenecks sit upstream in foundry capacity and equipment, while GPUs and servers are the most elastic supply segments in the near term.
More from Infra
- From 100KB to 2.5TB: a developer maps the full AI model spectrum and what each tier costs to run — abhishekkumar333 · 2026-10-04
- China's aggressive AI push runs into a new problem: too much usage, NYT reports — TMWNN · 2026-10-04
- DeepSeek V4.1-Flash hits 5,800 TPS per Ascend 950DT card, but critics call throughput underwhelming — teortaxesTex · 2026-10-04
- NVIDIA NIM API unreliable despite decent speed, users seek alternatives — Professional_Log1367 · 2026-10-04
- LLMxRay: Open-Source Local Observability for LLM Traffic, Launches With One Command — GuruCsharp · 2026-10-04
- Datacenter buildout would have been less aggressive in a world where everyone gets AGI at once — willcb · 2026-10-04