ultraData Center
A100 PCIe 80GB GPU
The 80 GB version of the Ampere A100 as a PCIe card, with HBM2e memory at 1,935 GB/s and a 300 W power limit. It is the A100 in Azure NC A100 v4 instances and in the single-card and small multi-GPU A100 offerings of many GPU clouds.
VRAM 80GB
CUDA Cores 6,912
Tensor Cores 432
TDP 300W
Process 7nm
From
$1.12/hr
across 8 providers
Cloud Pricing
Cheapest on Seeweb — 39% below avg| Provider | Price / hr |
|---|---|
$1.12/hr 1× | |
$1.19/hr 1× | |
$1.35/hr 1×2×4× | |
$1.35/hr 2× | |
$1.35/hr 1×2×4×8× | |
$1.49/hr 1×2×4× | |
$1.59/hr 1×2×3×4×5× | |
$2.00/hr 1×2×4× | |
$2.00/hr 1× | |
$2.23/hr 2×4× | |
$2.23/hr 8× | |
$2.23/hr 1× | |
$3.67/hr 1×2×4× |
Prices updated daily. Last check: Sep 30, 2026
A100 PCIe 80GB pricing by provider
Every configuration, price rank, and alternative for one provider at a time.
Performance
FP16
312 TFLOPS
FP32
19.5 TFLOPS
BF16
312 TFLOPS
INT8
624 TOPS
Bandwidth
1935 GB/s
Common Use Cases
Training and fine-tuning of models that need up to 80 GB per GPU, large-batch inference, HPC workloads using FP64, and MIG partitioning into up to seven instances
Full Specifications
Hardware
- Manufacturer
- NVIDIA
- Architecture
- Ampere
- CUDA Cores
- 6,912
- Tensor Cores
- 432
- RT Cores
- 0
- Process Node
- 7nm
- TDP
- 300W
Memory & Performance
- VRAM
- 80GB
- Memory Interface
- 5120-bit
- Memory Bandwidth
- 1935 GB/s
- FP32
- 19.5 TFLOPS
- FP16
- 312 TFLOPS
- BF16
- 312 TFLOPS
- FP64
- 9.7 TFLOPS
- INT8
- 624 TOPS
- Release
- 2021