HBM2E
The refined HBM2 behind A100-80GB, Gaudi and Maia.
You're looking at a technology record: a building block of AI hardware — what it does, who makes it, and which chips and systems use it.
HBM2E pushed per-stack bandwidth toward ~460 GB/s and 16 GB capacities — the workhorse memory of the 2020–2023 accelerator class and still used by cost-focused designs.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Used as memory in
Succeeded by
Successor to
Sources
Record reviewed as of 2026-01-15 — see Methodology.