AMD Instinct MI300X 192GB vs NVIDIA GeForce RTX 5090 Founders / reference
Instinct MI300X 192GB has 500% more memory than GeForce RTX 5090 Founders / reference (192 GB vs 32 GB). Its memory bandwidth is 196% more (5.3 vs 1.792 TB/s). In the cloud, the median on-demand Instinct MI300X 192GB price is $2.99 per GPU-hour against $0.54 for the GeForce RTX 5090 Founders / reference. Per gigabyte of GPU memory, that is $0.016 vs $0.017 per hour.
Specifications
Manufacturer figures as recorded by Kovara. Higher memory and bandwidth are highlighted; neither is a benchmark of your workload.
| Instinct MI300X 192GB | GeForce RTX 5090 Founders / reference | |
|---|---|---|
| Manufacturer | AMD | NVIDIA |
| Architecture | CDNA 3 | Blackwell |
| Memory | 192 GB HBM3 | 32 GB GDDR7 |
| Memory bandwidth | 5.3 TB/s | 1.792 TB/s |
| Power (TDP) | 750 W | 575 W |
| Form factor | OAM Module | PCIe consumer |
Cloud prices today
Public offers Kovara tracks, normalised to US dollars per GPU-hour. Prices use on-demand terms where available; spot and reserved offers are on each GPU’s price pages. Lower is highlighted.
| Instinct MI300X 192GB | GeForce RTX 5090 Founders / reference | |
|---|---|---|
| Lowest price | $2.59 | $0.25 |
| Median price | $2.99 | $0.54 |
| Highest price | $3.45 | $0.99 |
| Median per GB of memory | $0.016 | $0.017 |
| Public offers (all terms) | 8 | 7 |
| Providers | 3 | 3 |
Where to rent each
Lowest tracked price per provider. Open a provider to see every offer, region and term.
Instinct MI300X 192GB
GeForce RTX 5090 Founders / reference
Questions
Which is cheaper to rent, the Instinct MI300X 192GB or the GeForce RTX 5090 Founders / reference?
Right now the cheapest on-demand Instinct MI300X 192GB Kovara tracks is $2.59 per GPU-hour and the cheapest GeForce RTX 5090 Founders / reference is $0.25. Medians are $2.99 and $0.54. The cheaper GPU per hour is not always cheaper per job: a faster or larger GPU can finish sooner or need fewer units.
Which has more memory?
The Instinct MI300X 192GB has more memory: 192 GB against 32 GB. Memory decides which models and batch sizes fit on one GPU.
Are these benchmark results?
No. This page compares manufacturer specifications and public prices. Real performance depends on the model, precision, software and interconnect. Kova can help you estimate cost for a specific workload.