NVIDIA GB300 NVL72 GPU 288GB vs NVIDIA B30A
Side-by-side specifications, performance, and rental pricing for datacenter AI workloads.
GB300: Blackwell Ultra, 2025 B30A: Blackwell Ultra, 2025
NVIDIA GB300 NVL72 GPU 288GB
Full specs →Architecture
Blackwell Ultra
Year
2025
VRAM
288 GB
NVIDIA B30A
Full specs →Architecture
Blackwell Ultra
Year
2025
Specifications
GB300 B30A
Architecture
Blackwell UltraBlackwell Ultra
Launch Year
20252025
Form Factor
SXM—
VRAM
288 GB—
Memory Bandwidth
8.0 TB/s—
TDP
——
Process Node
4nm—
Spec GB300 B30A
Architecture Blackwell UltraBlackwell Ultra
Launch Year 20252025
Form Factor SXM—
VRAM 288 GB—
Memory Bandwidth 8.0 TB/s—
TDP ——
Process Node 4nm—
Performance (TFLOPS)
GB300 B30A
FP64
1.4 TFlops No verified data available
FP32
83.3 TFlops No verified data available
TF32
1250.0 TFlops
2500.0
TFLOPS sparse
No verified data availableBF16
2500.0 TFlops
5000.0
TFLOPS sparse
No verified data availableFP16
2500.0 TFlops
5000.0
TFLOPS sparse
No verified data availableFP8
5000.0 TFlops
10000.0
TFLOPS sparse
No verified data availableFP6
5000.0 TFlops
10000.0
TFLOPS sparse
No verified data availableFP4
15000.0 TFlops
20000.0
TFLOPS sparse
No verified data availableINT8
165.0 TFlops
330.0
TOPS sparse
No verified data availablePrecision GB300 B30A
FP64 1.4 TFlops No verified data available
FP32 83.3 TFlops No verified data available
TF32 1250.0 TFlops
2500.0
TFLOPS sparse
No verified data availableBF16 2500.0 TFlops
5000.0
TFLOPS sparse
No verified data availableFP16 2500.0 TFlops
5000.0
TFLOPS sparse
No verified data availableFP8 5000.0 TFlops
10000.0
TFLOPS sparse
No verified data availableFP6 5000.0 TFlops
10000.0
TFLOPS sparse
No verified data availableFP4 15000.0 TFlops
20000.0
TFLOPS sparse
No verified data availableINT8 165.0 TFlops
330.0
TOPS sparse
No verified data availableFLOPS by Precision
What actually differs
Both GPUs launched in 2025: the GB300 on Blackwell Ultra and the B30A on Blackwell Ultra.
For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.
Get Comparison Updates
New GPUs added weekly. Be the first to see how they compare.