Data center GPUs
NVIDIA B300 (HGX, Blackwell Ultra SXM)
Memory
288 GB
Bandwidth
8,000 GB/s
FP16
2,250 TFLOPS
FP8
4,500 TFLOPS
- Maker
- NVIDIA
- Model
- B300 (HGX, Blackwell Ultra SXM)
- Type
- Data center GPUs
- Memory
- 288 GB
- Memory type
- HBM3e
- Bandwidth
- 8,000 GB/s
- FP16
- 2,250 TFLOPS
- FP8
- 4,500 TFLOPS
- FP4
- 13,500 TFLOPS
- Power
- 1,400 W
- Form factor
- SXM
- Interface
- NVLink 5 1,800 GB/s; PCIe gen not stated per GPU
- Released
- 2025
Notes on these figures
Per-GPU compute derived as HGX B300 8-GPU totals / 8: FP16/BF16 36 PF and FP8 72 PF sparse (dense is half), FP4 144 sparse / 108 dense PF. NVIDIA chip-level blog (https://developer.nvidia.com/blog/inside-nvidia-blackwell-ultra-the-chip-powering-the-ai-factory-era/) lists Blackwell Ultra NVFP4 15 dense, FP8 5 dense PF and max TGP up to 1,400W (chip max, not HGX-specific). 288GB HBM3e up to 8TB/s from https://docs.nvidia.com/enterprise-reference-architectures/hgx-ai-factory/latest/components.html. Announced GTC March 2025.
Source: the maker's published specifications at nvidia.com. A missing figure means the maker does not publish it. Confirm current specifications with the maker or a dealer before you buy.
Comparable data center gpus
| Model | Memory | Memory type | Bandwidth | FP16 |
|---|---|---|---|---|
| AMD Instinct MI350X | 288 GB | HBM3E | 8,000 GB/s | 2,300 TFLOPS |
| AMD Instinct MI355X | 288 GB | HBM3E | 8,000 GB/s | 2,500 TFLOPS |
| AMD Instinct MI325X | 256 GB | HBM3E | 6,000 GB/s | 1,300 TFLOPS |
| AMD Instinct MI300X | 192 GB | HBM3 | 5,300 GB/s | 1,300 TFLOPS |
| NVIDIA B200 (HGX, SXM) | 180 GB | HBM3e | 8,000 GB/s | 2,250 TFLOPS |
| NVIDIA H200 NVL | 141 GB | HBM3e | 4,800 GB/s | 835.5 TFLOPS |