AMD Instinct MI300X
AMD's serious challenge to Hopper — more memory than H100, adopted by Microsoft and Meta.
You're looking at a chip record: what this processor is, its key specs, who designs, manufactures and supplies its parts, and where it gets used.
The MI300X stacked 8 CDNA3 compute chiplets on 4 base dies — 153 billion transistors, more than any GPU before it — with 192 GB of HBM3 when the H100 had 80. That memory advantage won real deployments: Azure ran GPT-4-class serving on it, Meta served Llama on it, and it proved the ROCm software stack at scale. The knock: software maturity, not silicon, remained AMD's gap.
Specifications & key facts
Each figure is labeled by basis: vendor spec, vendor claim, reported, estimate, or official. "Not publicly disclosed" means exactly that — we never invent values.
Overview
Manufacturing
Memory
Compute
Power
Physical
Price
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it. +6 more groups below.
Designed by
Manufactured by
Process node
Packaging
Memory technology
Memory supplied by
Found inside
Supported by
Succeeded by
Successor to
Competes with
Related
Sources
Record reviewed as of 2026-01-15 — see Methodology.