Azure AI Foundry: Private IP Compute Creation Confusion

Question

Using Azure AI Foundry with the following settings:

• Public network access: Disabled
• Workspace managed outbound access: Disabled

The goal is to create compute resources using private IPs from a virtual network (subnet) to connect with enterprise resources. However, there's an inability to select the virtual network or opt for no public IP in the Azure AI Foundry portal, unlike in Azure Machine Learning.

A limitation noted in the documentation (see below) is causing confusion, as compute resources can be created via the CLI without a public IP and using a virtual network.

Documentation Limitation:

Clarification is needed on the following:

1. Is it permitted to create compute resources in a virtual network without a public IP in Azure AI Foundry when managed network outbound mode is set to "Disabled"?
2. Is this behavior indicative of a bug considering the documented limitations?

Examples or insights would be greatly appreciated to better understand this limitation.

Thank you!