AMAT investor QA: inference-specific accelerators may not need 12-hi HBM stacks
BenBajarin · x · 2026-10-14
Ben Bajarin shares commentary from Applied Materials' ($AMAT) investor Q&A on memory despeculation, finding it logical and consistent with other industry voices: training will still require 12-hi HBM, but inference-specific accelerators—depending on design—may not use more than 4 or 8 stacks, making 12-hi unnecessary. This points to a bifurcating HBM demand structure as inference workloads scale.
More from Infra
- Same model, 9x cost gap: Artificial Analysis publishes cache hit rates per provider — ArtificialAnlys · 2026-10-14
- Apple's Minions turns idle Mac GPUs into a pooled cloud for ML inference — MikePFrank · 2026-10-14
- AnythingLLM ditches Ollama for llama.cpp, with up to 2x faster inference — solyarisoftware · 2026-10-14
- LlamaForge: a browser panel for upstream llama.cpp with local agents that must quote their sources — Sleepybear2611 · 2026-10-14
- CoreWeave's GPU fleet is mostly older boxes, raising questions about depreciation math — ivan_bezdomny · 2026-10-14
- NVMe reads the 7GB checkpoint in 2s, so why does ComfyUI take 30s to load? — rguerraf · 2026-10-14