GDDR7
Graphics memory conscripted for cost-efficient inference.
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.
GDDR7 trades HBM's bandwidth for dramatically lower cost and no advanced packaging. NVIDIA's Rubin CPX built a whole product philosophy on it: use cheap GDDR7 for the prefill phase of inference and save HBM for token generation.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Related
Sources
Record reviewed as of 2026-01-15 — see Methodology.