GPU regions and zones

This document describes GPU availability for Compute Engine, Google Kubernetes Engine (GKE), and AI Hypercomputer. If you use GPUs with other Google Cloud services, GPU availability might differ. For those services, read about GPU availability in the following documentation:

• Vertex AI: accelerator locations
• Cloud Run: regions that support GPUs
• Confidential Computing: zones that support GPUs

Multiple locations worldwide host GPU resources. These locations consist of regions and zones, which you can read about on the Regions and zones page. Each zone offers a variety of GPU models. When you create resources that use GPUs, such as Compute Engine instances or GKE node pools, you can only use the GPU models available in your selected zone. For example, in the asia-east1-a zone, you can create an N1 general-purpose VM. You can attach T4, V100, or P100 GPUs to this VM, but not a P4 GPU. In the same zone, you can also create G2 accelerator-optimized VMs that automatically have L4 GPUs attached. However, you cannot create an A2 accelerator-optimized VM with A100 80GB GPUs attached.

GPU availability by regions and zones

The following sortable table lets you select different options to see where GPU resources are available. You can search either by location or GPU machine type, or a combination of both. In the Select a GPU machine type list, the machine types that have NVIDIA GPUs automatically attached show the GPU type in brackets.

APAC North America South America Europe Middle East Africa A4X (NVIDIA GB200 Superchips) A4 (NVIDIA B200) A3 Ultra (NVIDIA H200) A3 Mega (NVIDIA H100) A3 High (NVIDIA H100) A3 Edge (NVIDIA H100) A2 Ultra (NVIDIA A100 80GB) A2 Standard (NVIDIA A100 40GB) G4 (NVIDIA RTX PRO 6000) G2 (NVIDIA L4) N1+T4 N1+V100 N1+P100 N1+P4 Clear all

Limitations and restrictions

• For A3 Mega machine types, the following regions and zones are available with limited capacity. To access these regions and zones, contact your Technical Account Manager (TAM).

  • Sydney, Australia, APAC: australia-southeast1-c
  • St. Ghislain, Belgium, Europe: europe-west1-b,c
  • Council Bluffs, Iowa, North America: us-central1-b

• For A3 High machine types, the following regions and zones are available with limited capacity. To access these regions and zones, contact your Technical Account Manager (TAM).

  • St. Ghislain, Belgium, Europe: europe-west1-b

• For A2 Standard machine types, the a2-megagpu-16g machine is only available in the following regions and zones:

  • Iowa, North America: us-central1-a,b,c,f
  • Netherlands, Europe: europe-west4-a,b
  • Singapore, APAC: asia-southeast1-c

Regions and zones that support NVIDIA RTX Virtual Workstations

NVIDIA RTX Virtual Workstations (vWS) are supported on the following machine series.

• G4: you can enable NVIDIA RTX PRO 6000 vWS on these machine types.
• G2: you can enable NVIDIA L4 vWS on these machine types.
• N1: you can enable NVIDIA T4, V100, or P100 vWS GPUs on these machine types.

For these machine types, vWS is supported in any zone where the machine type is available.

Regions and zones that support Local SSD

For all GPU machine types, Local SSD disks are available in the same zones as the machine type with the following exceptions:

• N1+V100: Local SSD disks aren't available in the us-east1-c zone.
• N1+P4: Local SSD disks are only supported in zones us-central1-c and northamerica-northeast1-b.

What's next?

• Learn more about geography and regions.
• If you encounter resource availability errors when requesting a GPU, see how to troubleshoot resource availability errors.
• Add Local SSD disks to your VMs. Local SSD disks pair well with GPUs when your apps require high-performance storage.
• Learn how to create VMs with attached GPUs.
• Learn about GPU pricing.