…
Cloud rental price, compute per dollar and specs, side by side.
Open in the interactive comparer →| Spec / price | B200 SXM | Instinct MI355X |
|---|---|---|
| Memory | 180 GB HBM3e | 288 GB HBM3ebetter |
| Memory bandwidth | 7,700 GB/s | 8,000 GB/sbetter |
| FP16 / BF16 (dense) | 2.25 PF | 2.52 PFbetter |
| FP8 (dense) | 4.5 PF | 5.03 PFbetter |
| FP4 (dense) | 9 PF | 10.1 PFbetter |
| INT8 (dense) | 4.5 PF | 5.03 PFbetter |
| TF32 (dense) | 1.1 PF | — |
| FP32 (dense) | 0.075 PF | 0.157 PFbetter |
| FP64 (dense) | 0.037 PF | 0.0786 PFbetter |
| Power | 1000 W | 1400 W |
| Architecture | Blackwell (2024) | CDNA 4 (2025) |
| Lowest on-demand | $5.43better | $8.60 |
| Median on-demand | $8.10better | $8.60 |
| Lowest spot | $2.01better | $4.50 |
| FP16/BF16 $ per PFLOP·h | $2.41better | $3.42 |
| Providers | 16 | 3 |
The Instinct MI355X has 1.1× more dense FP16/BF16 compute than the B200 SXM. The Instinct MI355X has 1.0× more memory bandwidth than the B200 SXM. Real-world speedups depend on the workload: training and prefill scale with compute, while LLM decoding scales mostly with memory bandwidth.
Right now the B200 SXM is cheaper per unit of compute: $2.41 vs $3.42 per PFLOP-hour at the lowest on-demand prices.
Running one GPU around the clock at the lowest on-demand price costs about $3,966 per month for the B200 SXM and $6,278 for the Instinct MI355X.
The Instinct MI355X has more memory: 288 GB vs 180 GB.