Compare A100 prices across 10 cloud providers. On-demand from $0.93/hr.
The NVIDIA A100 is the workhorse GPU of modern ML infrastructure. Based on the Ampere architecture, it introduced Multi-Instance GPU (MIG) partitioning, allowing a single 80 GB card to be sliced into up to seven isolated instances. While no longer the fastest for frontier training, it remains the most widely available datacenter GPU and offers the best price-to-performance ratio for most fine-tuning and medium-scale training jobs.
| Architecture | Ampere (GA100) |
| GPU Memory | 80 GB HBM2e (also 40 GB) |
| Memory Bandwidth | 2.0 TB/s (80 GB) |
| FP16 Tensor Core | 312 TFLOPS |
| TDP | 400W (SXM4) / 300W (PCIe) |
| Interconnect | NVLink 3.0 (600 GB/s) |
| Release Year | 2020 |
A100 80 GB instances start around $1.39/hr on specialized providers and can reach $5/hr on GCP and AWS. Spot pricing often dips below $1/hr. As the most broadly deployed datacenter GPU, the A100 has deep availability across all major and niche clouds, with prices dropping significantly since 2024.
| Provider | Instance Type | GPUs | On-Demand | Spot |
|---|---|---|---|---|
| Vast | 1x-A100_SXM4-32-65536 | 1 | $0.93 | N/A |
| Shadeform | denvr_A100-sxm4 | 1 | $1.25 | N/A |
| Lambda | gpu_1x_a100 | 1 | $1.29 | N/A |
| PrimeIntellect | lambdalabs__1xA100_40GB__30__200 | 1 | $1.29 | N/A |
| Shadeform | lambdalabs_A100-sxm4 | 1 | $1.29 | N/A |
| Vast | 2x-A100_PCIE-32-65536 | 2 | $1.33 | N/A |
| Cudo | intel-ice-lake-a100-pci_1x2v4gb | 1 | $1.37 | N/A |
| Cudo | intel-ice-lake-a100-pci_1x4v8gb | 1 | $1.40 | N/A |
| Lambda | gpu_1x_a100_sxm4 | 1 | $1.48 | N/A |
| Cudo | intel-ice-lake-a100-pci_1x12v24gb | 1 | $1.49 | N/A |
| Cudo | epyc-genoa-a100-pcie_1x2v4gb | 1 | $1.53 | N/A |
| Cudo | epyc-genoa-a100-pcie_1x4v8gb | 1 | $1.56 | N/A |
| Cudo | intel-ice-lake-a100-pci_1x24v48gb | 1 | $1.63 | N/A |
| Cudo | epyc-genoa-a100-pcie_1x12v24gb | 1 | $1.68 | N/A |
| Cudo | epyc-genoa-a100-pcie_1x24v48gb | 1 | $1.85 | N/A |
| Vast | 2x-A100_SXM4-32-65536 | 2 | $1.87 | N/A |
| Seeweb | ECS1GPU3 | 1 | $1.89 | N/A |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $2.45 | $1.46 |
| Lambda | gpu_2x_a100 | 2 | $2.58 | N/A |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $2.64 | $1.70 |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $2.68 | $1.49 |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $2.68 | $1.53 |
| Cudo | intel-ice-lake-a100-pci_2x8v16gb | 2 | $2.79 | N/A |
| Cudo | intel-ice-lake-a100-pci_2x16v32gb | 2 | $2.89 | N/A |
| Cudo | intel-ice-lake-a100-pci_2x24v48gb | 2 | $2.98 | N/A |
| Cudo | epyc-genoa-a100-pcie_2x8v16gb | 2 | $3.12 | N/A |
| Shadeform | paperspace_A100 | 1 | $3.19 | N/A |
| Cudo | epyc-genoa-a100-pcie_2x16v32gb | 2 | $3.24 | N/A |
| Cudo | intel-ice-lake-a100-pci_2x48v96gb | 2 | $3.26 | N/A |
| Cudo | epyc-genoa-a100-pcie_2x24v48gb | 2 | $3.35 | N/A |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $3.69 | $1.68 |
| Cudo | epyc-genoa-a100-pcie_2x48v96gb | 2 | $3.71 | N/A |
| Seeweb | ECS2GPU3 | 2 | $3.78 | N/A |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $3.96 | $1.49 |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $4.06 | $1.43 |
| Google Cloud | n1-standard-16 + 1xA100 | 1 | $4.35 | $1.98 |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $4.90 | $2.91 |
| Lambda | gpu_4x_a100 | 4 | $5.16 | N/A |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $5.29 | $3.41 |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $5.36 | $2.99 |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $5.36 | $3.05 |
| PrimeIntellect | datacrunch__8xA100_40GB__176__960 | 8 | $5.77 | N/A |
| Cudo | intel-ice-lake-a100-pci_4x32v64gb | 4 | $5.77 | N/A |
| Cudo | intel-ice-lake-a100-pci_4x48v96gb | 4 | $5.96 | N/A |
| Cudo | epyc-genoa-a100-pcie_4x32v64gb | 4 | $6.47 | N/A |
| Cudo | intel-ice-lake-a100-pci_4x96v192gb | 4 | $6.52 | N/A |
| Cudo | epyc-genoa-a100-pcie_4x48v96gb | 4 | $6.71 | N/A |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $7.39 | $3.35 |
| Cudo | epyc-genoa-a100-pcie_4x96v192gb | 4 | $7.41 | N/A |
| Seeweb | ECS3GPU3 | 4 | $7.56 | N/A |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $7.92 | $2.97 |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $8.13 | $2.85 |
| Google Cloud | n1-standard-32 + 2xA100 | 2 | $8.70 | $3.96 |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $9.80 | $5.82 |
| Shadeform | denvr_A100-sxm4x8 | 8 | $10.00 | N/A |
| Shadeform | lambdalabs_A100x8 | 8 | $10.32 | N/A |
| PrimeIntellect | lambdalabs__8xA100_40GB__124__1800 | 8 | $10.32 | N/A |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $10.57 | $6.82 |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $10.72 | $6.11 |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $10.72 | $5.98 |
| Cudo | intel-ice-lake-a100-pci_8x64v128gb | 8 | $11.55 | N/A |
| Lambda | gpu_8x_a100 | 8 | $11.84 | N/A |
| Cudo | intel-ice-lake-a100-pci_8x96v192gb | 8 | $11.92 | N/A |
| Cudo | epyc-genoa-a100-pcie_8x64v128gb | 8 | $12.94 | N/A |
| Cudo | epyc-genoa-a100-pcie_8x96v192gb | 8 | $13.41 | N/A |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $14.78 | $6.70 |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $15.83 | $5.95 |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $16.25 | $5.70 |
| Google Cloud | n1-standard-64 + 4xA100 | 4 | $17.40 | $7.91 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $17.93 | $11.07 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $19.27 | $13.33 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $19.50 | $11.42 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $19.50 | $11.64 |
| AWS | p4d.24xlarge | 8 | $21.96 | $11.93 |
| AWS | p4d.24xlarge | 8 | $21.96 | $20.60 |
| AWS | p4d.24xlarge | 8 | $21.96 | $11.79 |
| AWS | p4d.24xlarge | 8 | $23.72 | $14.54 |
| AWS | p4d.24xlarge | 8 | $23.72 | $7.29 |
| OCI | BM.GPU4.8 | 8 | $24.40 | N/A |
| AWS | p4d.24xlarge | 8 | $25.25 | $5.16 |
| AWS | p4d.24xlarge | 8 | $26.34 | $26.34 |
| Azure | Standard_ND96asr_v4 | 8 | $27.20 | $2.99 |
| AWS | p4d.24xlarge | 8 | $27.43 | $26.40 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $28.03 | $12.92 |
| AWS | p4d.24xlarge | 8 | $28.54 | $15.19 |
| AWS | p4d.24xlarge | 8 | $28.54 | $14.55 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $29.95 | $11.40 |
| AWS | p4d.24xlarge | 8 | $30.10 | $10.33 |
| AWS | p4d.24xlarge | 8 | $30.41 | $11.00 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $30.83 | $11.00 |
| Azure | Standard_ND96asr_v4 | 8 | $32.63 | $3.59 |
| Google Cloud | n1-standard-96 + 8xA100 | 8 | $32.97 | $15.29 |