Clarification on GPU VM Eligibility for Sponsorship Subscription

Question

My previous request for NCasT4v3 was denied because my subscription is a Sponsorship subscription.

Could you please tell me which GPU VM families, if any, are eligible for quota approval under my specific subscription?

I do not need a specific GPU model. I only need access to a GPU-enabled VM capable of running inference for a fine-tuned Llama 3.1 8B model.

If no GPU VM families are eligible for this subscription, please confirm that so I can evaluate alternative deployment options.

Answer 1

Hi @Meghana Kadari

Thanks for the details. Based on Microsoft documentation, there is not a single published list of GPU VM families that are always quota-approvable for all Sponsorship subscriptions.

What can be confirmed:

For NC-family GPU sizes (including NCasT4_v3), Azure documents that quota starts at 0 and that "your subscription level might not support selecting or deploying these SKUs." Source: https://learn.microsoft.com/azure/virtual-machines/sizes/gpu-accelerated/nc-family#series-in-family

Quotas can be adjustable or non-adjustable, and non-adjustable limits can be tied to subscription scope. If increase is unavailable, Azure directs you to open a support request or use other series/regions. Source: https://learn.microsoft.com/azure/quotas/per-vm-quota-requests

Sponsorship is a distinct Azure offer type (MS-AZR-0036P), so entitlement behavior can differ from PAYG. Source: https://learn.microsoft.com/azure/cost-management-billing/costs/understand-cost-mgt-data#supported-microsoft-azure-offers

Given your denied NCasT4_v3 request, the safest conclusion is:

• NCasT4_v3 is currently not approvable for your current Sponsorship subscription/region combination.
• Other GPU families may or may not be allowed, but must be validated directly in your subscription.

Recommended next check (authoritative for your tenant):

1. Azure portal -> Quotas -> Compute -> filter for GPU families (NC/ND/NV) in your target region.
2. Check whether each family is marked adjustable vs non-adjustable.
3. Try "Request quota" for each visible GPU family.
4. If unavailable, open a quota support request from the same pane to get an explicit allow/deny by Azure Support.

If all GPU families are non-adjustable/denied in your Sponsorship subscription, then yes, you should plan an alternative deployment route (for example, a PAYG subscription for GPU inference workloads).

Please 'Upvote' (Thumbs-up) and 'Accept' as answer if the response was helpful. This will be helpful to other community members who face similar issues.

---

Best regards,

Andrew S Taylor

Answer 2

Hey Meghana — I see why you’re asking. However, the documentation you provided doesn’t include any matrix that maps Azure Sponsorship subscription types (including “Sponsorship” / “Azure Pass sponsorship” / other sponsored offers) to specific eligible GPU VM families for quota approval.

What we can confirm from the provided docs is general eligibility/entitlement behavior for sponsorships and quotas are controlled through entitlement/approval, but the docs don’t state which VM families (NC/ND/NV, etc.) are eligible under a sponsorship subscription.

• Microsoft Azure Sponsorship is email invitation only and selected customers get an offer/activation flow. (So the approval/eligibility for compute—including quota requests—depends on the specific entitlement your account received.)
• For Azure Pass Sponsorship, the docs focus on redemption/activation and do not provide any GPU-VM-family quota eligibility list.
• For Azure Sponsorship/EA Sponsorship, the docs mention general quota/resource availability constraints, but again no specific GPU VM families eligible for approval are listed.

Because there’s no GPU-family eligibility list in the provided material, we can’t accurately tell you “which GPU VM families are eligible for quota approval” under your sponsorship type based solely on this documentation.

The most reliable way is to check against the service quota/limits behavior in your subscription and/or request the quota for the specific GPU family you care about (since your earlier NCasT4v3 request was denied, it likely indicates your sponsorship entitlement doesn’t support that GPU family—but the docs don’t let us generalize to other GPU families).

Reference list (from the provided documentation)

• Microsoft Azure Sponsorship / redeem & activate (Azure Pass + Sponsorship activation overview): https://azure.microsoft.com/offers/ms-azr-0036p/
• “sponsorship request” (portal entitlement details; includes activation/balance/redemption info): https://supportabilityhub.microsoft.com/solutions/apollosolutions/42bb56f3-9f6d-d60f-00bc-cd8bdae2b6bc/apollo-benefitsofferssponsorshiprequest

Answer 3

Hey, GPU quota availability depends on the type of Azure subscription and the restrictions associated with it. Unfortunately, support forums and moderators do not have visibility into the specific quota policies applied to an individual Sponsorship subscription.

Though Azure don't provide gpu's vm env for the free trails .... there are specific CPU that can use check this link for more to find out...if any helps.
https://learn.microsoft.com/en-us/azure/virtual-machines/quotas

https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/azure-subscription-service-limits

https://azure.microsoft.com/en-us/pricing/offers/ms-azr-0036p/

https://portal.azure.com/#view/Microsoft_Azure_Capacity/QuotaMenuBlade

If your request for the NCasT4v3 family was denied because of the subscription type, it may indicate that GPU-enabled VM families are restricted for your sponsorship offer. However, we cannot confirm whether all GPU VM families are blocked or whether certain families remain eligible without checking the subscription's quota and offer details.

To get a definitive answer, please open a support request under Quota or Subscription Management and ask Microsoft to confirm which GPU VM families (if any) are eligible for quota approval on your Sponsorship subscription....

If no GPU families are supported under the sponsorship offer, Microsoft support can confirm that directly, allowing you to evaluate alternatives such as:

• Upgrading to a Pay-As-You-Go subscription
• Using managed AI services instead of self-hosted GPU VMs
• Deploying your model through Azure AI offerings that may not require dedicated GPU VM quota

Hope this helps 😊