Defense AI: What to Do When a Model Shows "No Quota"
Last updated: November 5, 2026
When you deploy models in Defense AI, some may show No quota. Nothing is broken, and nothing is wrong with your setup. It's a limit Microsoft puts on new Azure subscriptions, given the demand for AI is so high.
Why this happens
Think of quota as a credit limit for AI. The newest AI models run on expensive hardware that Microsoft has a limited supply of, so Microsoft rations it. Every Azure subscription gets its own limit for each model.
A brand-new subscription has no history with Microsoft, so it starts with a low limit, and for the newest models that limit is often zero. Defense AI always sets up a new, dedicated subscription for you so its billing stays separate from the rest of your Azure account.
Defense AI can't raise the limit for you. Only Microsoft can, and a person at Microsoft reviews each request.
The fix, in short
Two things raise your limit:
Using what you already have. Microsoft raises limits automatically as a subscription is used and pays its bills. It also approves requests faster from subscriptions that are already active.
Asking Microsoft for more. You fill out a short form, one per model.
So: start with a model that has capacity today, get your team using it, and submit the form for the models you want. The steps below walk through each part. Once a subscription has some history, this usually stops coming up.
Step 1: Deploy a model that has quota
On the Deploy models step, pick any model that shows available quota (often gpt-4o) and deploy it. Your team can start using Defense AI right away, and you can change models later.
Do this before you request more quota. Microsoft gives priority to subscriptions that are already using the quota they have, and requests from unused subscriptions may be denied.
Step 2: Gather what you need
Subscription ID of your Defense AI pay-as-you-go subscription. Find it in the Azure portal under Subscriptions.
An admin email in the same tenant as that subscription. Microsoft sends its decision to this address, so use one someone actually checks.
Company name and address.
The models you want. We suggest one primary model, one reasoning model and one low-cost model.
Step 3: Submit the quota request form
Open Microsoft's form at https://aka.ms/oai/stuquotarequest. The form takes one model per submission. After you submit, click Submit another response for the next model. Your earlier answers usually carry over.
Field | What to enter |
|---|---|
Name, email, company, address | Your details (use the admin email from Step 2) |
Subscription Id | Your Defense AI subscription ID |
Quota Type | Model Deployment (PTU/RPM/TPM) |
Model Type | Azure OpenAI |
Quota Request Type | Data Zone Standard |
Data Zone Standard Model | The model you want |
Data Zone Standard Pool | US |
Data Zone Standard Quota | 100 |
The quota field is the total you want in thousands of tokens per minute, so 100 means 100,000 TPM. That's a reasonable starting request for a new subscription and is enough for a small team.
Why Data Zone Standard
Data Zone Standard (US) keeps processing inside the United States and is pay-per-token. This is the right choice for CUI and ITAR.
Global Standard may process requests outside the US. Don't choose it.
Standard (regional) doesn't offer most of the newest models.
Provisioned reserves capacity you pay for whether you use it or not.
Justification
Copy this and replace the bracketed parts. Change the last sentence for each model.
[Company] is a US defense contractor handling CUI. We are deploying a private AI gateway in our own Azure subscription for internal use in [proposal writing, project management, and engineering]. Due to DFARS and ITAR regulatory requirements, we can only use US compute, so Data Zone Standard (US) is required. We need [model] for [what you'll use it for], starting with our admin team and expanding company-wide.A specific use case and a sense of who will use it helps Microsoft approve the request.
Step 4: Wait for Microsoft, then deploy
Microsoft reviews requests in the order they arrive, and timing varies. Watch the inbox you entered on the form.
Once a request is approved, return to the Defense AI admin portal, go to Deploy models, and click Deploy on the newly available model. You can then make it your default.
If something doesn't look right
Approved, but the model still doesn't appear. Some models aren't offered in every Azure region. Contact your Atomus contact and we'll check whether your resources need to be in a different US region.
Request denied. Keep using the models you have for a few days, then resubmit with a more detailed justification.
You're on Azure Government. Azure Government uses a separate quota request form. Contact your Atomus contact and we'll send you the right one.