AMD MI250
vs
Nvidia RTX 5090
A data center GPU against a consumer card, one CDNA2 and one Blackwell: 128 GB HBM2e against 32 GB GDDR7, 79% more memory bandwidth on the MI250, and Infinity Fabric 400 GB/s on the MI250 where the RTX 5090 is PCIe only.
At a glance
Updated 7 minutes ago- Memory
- 128 GB
- HBM2e
- Memory bandwidth
- 3.20 TB/s
- From
- --
- Not sold on demand · 1 provider
- Memory
- 32 GB
- GDDR7
- Memory bandwidth
- 1.79 TB/s
- From
- $0.35 /GPU/hr
- 20 providers
Price comparison
Cheapest listed price per GPU per hour by billing type, and the median across providers.
| Billing | MI250 | RTX 5090 | Difference |
|---|---|---|---|
| On-demand | Not sold on demand | $0.35 | |
| Spot | No spot listings | $0.25 | |
| Reserved | $1.30 | $0.21 | 6.2x more expensive on MI250 |
| Median | $1.50 | $0.69 | 2.2x more expensive on MI250 |
Median: Median across providers, one vote each, so a large catalog does not outweigh a small one.
Price history
MI250 vs RTX 5090 price history
Weekly median on-demand price per GPU per hour
Specs comparison
| Spec | MI250 | RTX 5090 | Difference |
|---|---|---|---|
| Memory | |||
| Capacity | 128 GB HBM2e | 32 GB GDDR7 | 4x more on MI250 |
| Bandwidth | 3,200 GB/s | 1,792 GB/s | 1.8x higher on MI250 |
| Bus width | 8,192-bit | 512-bit | 16x wider on MI250 |
| Compute · dense | |||
| FP4 | Not supported | 1,676 TFLOPS | RTX 5090 only |
| FP8 | Not supported | 419 TFLOPS | RTX 5090 only |
| FP16 / BF16 | 362.1 TFLOPS | 209.5 TFLOPS | 1.7x higher on MI250 |
| INT8 | 362.1 TOPS | 838 TOPS | 2.3x higher on RTX 5090 |
| FP32 | 90.5 TFLOPS | 104.8 TFLOPS | 1.2x higher on RTX 5090 |
| FP64 | 45.3 TFLOPS | Not supported | MI250 only |
| Platform | |||
| Interconnect | Infinity Fabric 400 GB/s | PCIe only | Infinity Fabric on MI250 only |
| Board power | 500 W | 575 W | 1.1x higher on RTX 5090 |
| Released | Q4 2021 | Q1 2025 | 3 years newer on RTX 5090 |
| Precision | |||
| Supported formats | FP16BF16FP32FP64INT4INT8 | FP4FP8FP16BF16TF32FP32INT8 | |
Compute: Vendor peak figures, dense and per GPU. Not measured throughput.
Source: MI250 datasheet and RTX 5090 datasheet.
Where to rent them
Providers listing these GPUs, those with both first, so you can switch without moving clouds.
| Provider | MI250 /GPU/hr | RTX 5090 /GPU/hr |
|---|---|---|
Cirrascale
|
Reserved (12mo) $1.30 | -- |
HyperAI
|
-- | On-Demand $0.35 |
Salad
|
-- | On-Demand $0.45 |
GPUhub
|
-- | On-Demand $0.46 |
Nova Cloud
|
-- | On-Demand $0.63 |
Lium
|
-- | On-Demand $0.69 |
Theta EdgeCloud
|
-- | On-Demand $0.82 |
Vast.ai
|
-- | On-Demand $10.67 |
GPU.ai
|
-- | On-Demand $53.34 |
1 provider list the MI250 and 20 the RTX 5090. Every listing is on the MI250 page and the RTX 5090 page.
Similar GPUs
Other comparisons in the same class.
Our data for AMD MI250 was last updated on Sept. 7, 2026, and for Nvidia RTX 5090 on Sept. 7, 2026.