In VMware Private AI Foundation (PAIF) 9.1, the default NVIDIA GPU Operator version used in the VCFA AI Kubernetes cluster blueprints is v25.3.1. This version may no longer be supported by NVIDIA.
This KB article guides users on how to update the PAIF 9.1 VCFA blueprints to use the updated NVIDIA GPU Operator v25.10.1.
VMware Private AI Foundation 9.1
Private AI Foundation Quickstart
NVIDIA GPU DirectPath Mode: vGPU mode or Fixed/Dynamic DirectPath mode (Passthrough)
Environment Connectivity
These two selections produce four possible combinations, each generating a different AI Kubernetes cluster blueprint variant.
Log in to VCF Automation.
Go to Build & Deploy > Blueprint Design > Blueprints tab.
Locate the AI Kubernetes cluster blueprint to be updated.
Identify Scenario & Open Attachment: Identify the specific deployment scenario among the four available options and open the corresponding README file in the attachment.
The README file contains a series of numbered changes for the scenario, each with a Find code block and a Replace with code block.
Navigate to the Code Tab: In the blueprint editor, click on the "Code" tab at the top right to view the underlying YAML configuration.
Find and Replace Code: Click the "SEARCH" button on the top right side of the code panel.
For each change listed in the README file, locate the exact Find code block within the blueprint's code.
Replace it entirely with the corresponding Replace with code block from the README file.
Repeat: Repeat the search and replace process until all changes specified in the README file for the selected scenario have been successfully applied. Save the changes when finished.
Click the "VERSION" button located at the bottom left of the main screen to open the versioning dialog.
In the "Creating Version" pop-up window, check the box next to "Publish this version to the catalog".
Click the "CREATE" button at the bottom right of the pop-up window to finalize and publish.