AI infrastructure & chips · AWS vs Google Cloud
Amazon EC2 accelerated computing (NVIDIA GPU) vs Compute Engine GPU machines
EC2 GPU instances map to Compute Engine accelerator-optimized and N1 machines with NVIDIA GPUs.
Side-by-side comparison
| Aspect | Amazon Web Services | Google Cloud |
|---|---|---|
| Service | Amazon EC2 accelerated computing (NVIDIA GPU) | Compute Engine GPU machines |
| Lifecycle | Generally available | Generally available |
| What it is | GPU instances. Training families include P5/P5e/P5en (H100/H200), P6-B200, P6-B300, and P6e-GB200 UltraServers. Inference and graphics families include G6/G6e/G7/G7e. | Accelerator-optimized and N1 VMs with NVIDIA GPUs, also used by GKE and Agent Platform. |
| Pricing model | On-demand, Savings Plans, Spot, and Capacity Blocks for ML (reserved GPU time slots). | Billed per second for the VM and attached GPUs, with Spot, committed-use, and reservation options. Prices vary by region. |
| Docs | Amazon EC2 accelerated computing (NVIDIA GPU) docs | Compute Engine GPU machines docs |
| Pricing page | Amazon EC2 accelerated computing (NVIDIA GPU) pricing | Compute Engine GPU machines pricing |
Key differences
- GPU machine types and regions differ; check zone availability before you plan capacity.
- Capacity can be reserved, bought as committed use, or run as Spot VMs, similar to EC2 options.
Pricing summary
Pricing models only. No prices are quoted because they change by region and over time.
- AWS
- On-demand, Savings Plans, Spot, and Capacity Blocks for ML (reserved GPU time slots). Amazon EC2 accelerated computing (NVIDIA GPU) pricing →
- GCP
- Billed per second for the VM and attached GPUs, with Spot, committed-use, and reservation options. Prices vary by region. Compute Engine GPU machines pricing →
And on Azure?
Direct match Azure's equivalent is Azure ND-series GPU VMs. AWS vs Azure comparison →
FAQ
What is the Google Cloud equivalent of Amazon EC2 accelerated computing (NVIDIA GPU)?
EC2 GPU instances map to Compute Engine accelerator-optimized and N1 machines with NVIDIA GPUs.
How does pricing compare between Amazon EC2 accelerated computing (NVIDIA GPU) and Compute Engine GPU machines?
AWS: On-demand, Savings Plans, Spot, and Capacity Blocks for ML (reserved GPU time slots). Google Cloud: Billed per second for the VM and attached GPUs, with Spot, committed-use, and reservation options. Prices vary by region. Check the official pricing pages for current rates.
What is the Azure equivalent?
Azure ND-series GPU VMs (direct match). Amazon EC2 P-family instances (P5/P5e/P5en, P6-B200, P6-B300, P6e-GB200) map to Azure ND-series VMs (ND H100 v5, ND H200 v5, ND GB200 v6, ND GB300 v6).