NVIDIA P100 end of support

NVIDIA P100 GPUs on Google Cloud will reach End of Support (EOS) on September 15, 2026.

What do I need to know

After September 15, 2026, you cannot create or access any Google Cloud resources that run NVIDIA P100 GPUs. This deprecation affects resources for the following services:

  • Compute Engine: VM instances
  • Google Kubernetes Engine (GKE): nodes
  • Gemini Enterprise Agent Platform: models, jobs, and endpoints
  • Cloud Workstations: workstations
  • Dataflow: pipeline jobs
  • Managed Service for Apache Spark: clusters and serverless batches
  • Deep Learning VM and Container-Optimized OS instances

How will this EOS impact my existing resources

Until September 15, 2026, your resources that are running NVIDIA P100 GPUs are not affected. However, your projects might be at risk because you are running a GPU model that is approaching its EOS. EOS products or features are not supported by Google Cloud.

What do I need to do

To transition your workloads to supported GPUs, you must choose a new GPU model, review its availability in your regions, and then migrate your workloads.

Choose a GPU model

We recommend moving to either the NVIDIA T4 or L4 GPUs. To determine the best GPU model for your workload, compare the following options:

Feature NVIDIA T4 NVIDIA L4
Best for Cost-efficient inference, small machine learning (ML) model training, data analytics, and legacy workloads. High-performance AI inference, generative AI, media streaming, and graphics. L4 GPUs provide up to four times the performance of T4 GPUs.
Architecture and memory Turing architecture with 16 GB GDDR6 GPU memory. Ada Lovelace architecture with 24 GB GDDR6 GPU memory.
Machine series N1 general-purpose G2 accelerator-optimized
Migration path In-place upgrade: Modify your existing VM to switch to a T4 in-place without migrating data. New VM migration: Move to a new VM to transition to the G2 machine series.

For a comprehensive list and comparison of all GPU models, see GPU machine types.

Review available locations

To verify that your selected GPU model is available, review the location details for your service:

Migrate your workloads

Based on the service that you use, select one of the following transition paths:

Compute Engine workloads

Depending on the machine type and GPU model that you want to switch to, choose one of the following methods:

Move to a new VM

You must move to a new VM if the new GPU model is not supported in the zone of the existing VM, or if you are switching to a GPU that runs on an accelerator-optimized machine type.

To move to a new VM, complete the following steps:

  1. If your existing VM uses Local SSD disks that contain data you want to keep, move the contents of those disks to a Persistent Disk volume.

  2. Create a new VM. For example, to create a VM that runs on the G2 (NVIDIA L4) machine type, see Create a G2 instance.

  3. Move Persistent Disk volumes from the old VM to the new VM. You can do this by detaching the Persistent Disk volume from the old VM and adding it to the new VM. You can alternatively transfer files from one VM to the other.

  4. Optional: Move the saved data from Persistent Disk volumes back to a Local SSD disk.

  5. Reassign any static IP addresses that were associated with the original VM to the new VM.

  6. Install GPU drivers on the VM.

  7. Install your applications on the new VM.

  8. Delete the old VM.

Modify your existing VM

If the new GPU model that you want to switch to is supported in the same zone, and runs on the N1 general-purpose machine types, then you can modify your existing VM to switch from an NVIDIA P100 to the new GPU model.

Console

To modify your GPU type in the console, complete the following steps:

  1. Verify that you have enough quota for the new GPU type.
  2. Verify that all of your critical applications are stopped on the VM.

  3. In the Google Cloud console, go to the VM instances page.

    Go to VM instances

  4. Click the name of the VM that is running the NVIDIA P100 GPU. The VM instance details page opens.

  5. Click Stop. If there is no Stop option, then click More actions > Stop. Check the notification panel to see when the VM is stopped.

  6. Click Edit and then complete the following steps:

    1. In the Machine configuration section, select the GPU machine family.
    2. From the GPU type list, change the GPU model from NVIDIA P100 to your new GPU model.

  7. To apply your changes, click Save.

  8. To restart the VM, click Start/Resume.

gcloud

To modify your GPU type using the Google Cloud CLI, complete the following steps:

  1. Verify that you have enough quota for the new GPU type.
  2. Verify that all of your critical applications are stopped on the VM.

  3. Export the VM configuration to a local YAML file:

    gcloud compute instances export VM_NAME \
   --destination=config.yaml \
   --zone=ZONE

  4. Open the exported config.yaml file and update the guestAccelerators block:

    guestAccelerators:
- acceleratorCount: ACCELERATOR_COUNT
 acceleratorType: https://www.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/acceleratorTypes/ACCELERATOR_TYPE

    Ensure that scheduling.onHostMaintenance is set to TERMINATE under the scheduling block:

    scheduling:
 automaticRestart: true
 onHostMaintenance: TERMINATE

  5. Stop the VM.

    gcloud compute instances stop VM_NAME \
   --zone=ZONE

  6. Apply the modified configuration using the update-from-file command.

    gcloud compute instances update-from-file VM_NAME \
   --source=config.yaml \
   --most-disruptive-allowed-action=RESTART \
   --zone=ZONE

  7. Start the VM.

    gcloud compute instances start VM_NAME \
   --zone=ZONE

Replace the following:

  • PROJECT_ID: your project ID.
  • VM_NAME: the name of the VM instance.
  • ZONE: the zone where the VM is located.
  • ACCELERATOR_COUNT: the number of GPUs that you want attached.
  • ACCELERATOR_TYPE: the GPU model (for example,nvidia-tesla-t4).

REST

To modify your GPU type using the Compute Engine API, complete the following steps:

  1. Verify that you have enough quota for the new GPU type.
  2. Verify that all of your critical applications are stopped on the VM.

  3. Stop the VM.

    POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/instances/VM_NAME/stop

  4. Update the scheduling options on the stopped VM to terminate during host maintenance.

    POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/instances/VM_NAME/setScheduling

{
 "onHostMaintenance": "TERMINATE",
 "automaticRestart": true
}

  5. Add or modify the attached GPUs by calling setMachineResources.

    POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/instances/VM_NAME/setMachineResources

{
 "guestAccelerators": [
   {
     "acceleratorCount": ACCELERATOR_COUNT,
     "acceleratorType": "https://www.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/acceleratorTypes/ACCELERATOR_TYPE"
   }
 ]
}

  6. Start the VM.

    POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/instances/VM_NAME/start

    Replace the following:

    • PROJECT_ID: your project ID.
    • VM_NAME: the name of the VM instance.
    • ZONE: the zone where the VM is located.
    • ACCELERATOR_COUNT: the number of GPUs that you want attached.
    • ACCELERATOR_TYPE: the GPU model (for example, nvidia-tesla-t4).

Other Google Cloud workloads

If your workloads run on services other than Compute Engine such as GKE, Gemini Enterprise Agent Platform, Cloud Workstations, Dataflow, or Managed Service for Apache Spark, then you must do the following:

  1. Update your configurations to reference a supported GPU model.

    • For GKE, Gemini Enterprise Agent Platform, or Cloud Workstations, update your configuration templates.
    • For Dataflow, update your pipeline specifications.
    • For Managed Service for Apache Spark, update your cluster definitions.

    For instructions about configuring GPUs for your service, see the product documentation for that service.

  2. Restart or recreate your resources.

What's next