Blackwell Ultra
NVIDIA GB300 SXM6 288GB
wins
4
The GB300 SXM6 (Grace Blackwell Ultra) is a CPU+GPU superchip with 288 GB HBM3e and 8 TB/s memory bandwidth.
Hopper
NVIDIA H100 PCIe 80GB
wins
1
The H100 PCIe is the more affordable form factor of the H100.
Performance comparison
Visual bars — winner highlighted
NVIDIA GB300 SXM6 288GB
Metric
NVIDIA H100 PCIe 80GB
FP16 Tensor TFLOPS (dense)
Bandwidth
VRAM
TDP (lower=better)
TFLOPS/watt
GB/s per watt
Full specification table
| Spec | NVIDIA GB300 SXM6 288GB | NVIDIA H100 PCIe 80GB |
|---|---|---|
| Architecture | Blackwell Ultra | Hopper |
| VRAM | 288 GB HBM3e | 80 GB HBM2e |
| VRAM type | HBM3e | HBM2e |
| Memory bandwidth | 8,000 GB/s | 2,000 GB/s |
| FP16 Tensor TFLOPS (dense) | 2,800 TFLOPS | 756 TFLOPS |
| TDP | 1,200 W | 350 W |
| TFLOPS/watt | 2.33 T/W | 2.16 T/W |
| GB/s per watt | 6.67 GB/s/W | 5.71 GB/s/W |
| NVLink | Yes | No |
| Release year | 2025 | 2022 |
Live cloud pricing
On-demand pricing across providers — updated in real time
Power efficiency analysis
TFLOPS/watt and GB/s/watt — critical for data center TCO
NVIDIA GB300 SXM6 288GB
NVIDIA H100 PCIe 80GB
TFLOPS/watt measures compute efficiency — how much AI throughput you get per watt of power consumed. For data centers with PUE of 1.2–1.5, a 10% improvement in TFLOPS/watt translates directly to lower electricity costs and cooling requirements. GB/s/watt measures memory bandwidth efficiency, which is the binding constraint for memory-bound LLM inference workloads.
When to choose each GPU
Choose NVIDIA GB300 SXM6 288GB for:
- Frontier AI training
- Inference at scale
- Memory-bound LLM serving
Weaknesses / Limitations
- Very limited cloud availability — still ramping supply
- Extremely high cost — enterprise/hyperscaler pricing only
- Requires specialized NVLink-C2C infrastructure
Choose NVIDIA H100 PCIe 80GB for:
- Budget Hopper inference
- PCIe server deployments
- Cost-sensitive LLM serving
Weaknesses / Limitations
- Lower FP16 TFLOPS (756) and bandwidth (2,000 GB/s) than H100 SXM5
- No NVLink — cannot scale across cards for very large models
- Higher cost than A100 with only moderate performance advantage
Popular GPU comparisons