Comprehensive comparison for AI, machine learning, and high-performance computing workloads.
Blackwell Architecture
Ampere Architecture
| Specification | NVIDIA GB200 | NVIDIA A40 |
|---|---|---|
| Architecture | Blackwell | Ampere |
| Release Year | 2024 | 2020 |
| VRAM | 192 GB+300.0% | 48 GB |
| Memory Type | HBM3e | GDDR6 |
| Memory Bandwidth | 8000 GB/s+1049.4% | 696 GB/s |
| FP32 Performance | 90 TFLOPS+140.6% | 37.4 TFLOPS |
| FP16 Performance | 180 TFLOPS+140.6% | 74.8 TFLOPS |
| INT8 Performance | 3600 TOPS+1104.0% | 299 TOPS |
| Tensor Cores | 18432 | 10752 |
| CUDA Cores | N/A | 10752 |
| TDP | 1000W | 300W |
| Form Factor | Superchip | PCIe |
| NVLink Support | Yes | No |
| Avg. Price/Hour | $3.75+400.0% | $0.75 |
Single-precision floating-point performance for general compute workloads
NVIDIA GB200 is 140.6% faster
Half-precision performance optimized for deep learning training
NVIDIA GB200 is 140.6% faster
Integer performance for efficient model inference and deployment
NVIDIA GB200 is 1104% faster
Data transfer speed between GPU and memory
NVIDIA GB200 is 1049.4% faster
NVIDIA GB200
NVIDIA GB200
NVIDIA A40
NVIDIA A40
Enterprise-grade infrastructure
Get a custom quote in 24 hours for reserved GPU clusters with high-speed interconnect, any region, any GPU model, and any number of GPUs you need.
Any GPU
Choose your hardware
Any Quantity
Scale as needed
Any Region
Global availability
Interconnect
High-speed networking
Go from comparison to running workload in under 60 seconds. No complex setup required.
Only pay for what you use. Stop instances anytime. No hidden fees or long-term commitments.
Enterprise-grade infrastructure with 99.9% uptime. Trusted by AI teams worldwide.
Explore more GPU comparisons to find the perfect match for your workload