Parts/Accelerators/AMD Instinct MI300X
Why this part
The discrete variant, aimed at language-model inference where 192GB against H100's 80GB means a model fits on fewer devices. It is AMD's strongest position in the AI market and its weakest in software, and that pairing is the whole story of the accelerator business: memory capacity can be bought, an ecosystem has to be grown.
Systems using this part
- Systems
- 2
- Rmax underneath
- 114.5 PFlop/s
- Units recorded
- -
0 of 2 edges disclose a quantity
| System | Role | Country | Units | System Rmax | Supplied by | Rank |
|---|---|---|---|---|---|---|
| MAXIMUS-384 2025-11 | Accelerator | USA | - | 114.5 PFlop/s | AMD | #24 → |
| Vultr AMD GPU Supercompute Cluster (Chicago) 2024-12 | Accelerator | USA | - | - | AMD | - |
“Units recorded” sums only the edges where a public source states a quantity. It is a floor, not a total, and should never be read as installed base.