---
title: Which Azure OpenAI deployment type keeps agent data in region?
url: https://opencontroller.lyzr.com/blog/azure-openai-deployment-types-and-data-residency
published: 2026-11-17
updated: 2026-11-17
author: Lyzr
---

# Which Azure OpenAI deployment type keeps agent data in region?

**Only Standard and Regional Provisioned deployments process prompts and responses within the Azure geography you choose. Data Zone deployments process within a Microsoft-defined zone such as the US or the EU Data Boundary, and Global deployments may process in any Azure region. Data stored at rest stays in your designated geography for every type, so the deployment type decides where inference happens, not where data is stored.**

Agent teams often pick Global Standard because it has the highest default quota and the newest models first. For regulated workloads, that choice can quietly move inference outside the region your data residency commitments assume.

## Where does each deployment type process data?

| Deployment type | SKU name | Where inference is processed |
|---|---|---|
| Global Standard | `GlobalStandard` | Any Azure region |
| Global Provisioned | `GlobalProvisionedManaged` | Any Azure region |
| Global Batch | `GlobalBatch` | Any Azure region |
| Data Zone Standard | `DataZoneStandard` | Within the data zone |
| Data Zone Provisioned | `DataZoneProvisionedManaged` | Within the data zone |
| Data Zone Batch | `DataZoneBatch` | Within the data zone |
| Standard | `Standard` | Within the Azure geography |
| Regional Provisioned | `ProvisionedManaged` | Within the Azure geography |
| Developer | `DeveloperTier` | Any Azure region; no data residency guarantee or SLA |

Source: Microsoft Learn, Azure OpenAI deployment types. Microsoft notes that Standard and Regional Provisioned traffic may move between regions inside the geography for operational purposes.

## What exactly is a data zone?

Data zones are Microsoft-defined groupings:

- **United States:** data may be processed anywhere in the US.
- **European Union:** processing follows the EU Data Boundary, which can include EFTA countries such as Norway and Switzerland as well as EU member states.
- **Asia Pacific** is also listed as a data zone.

Microsoft states it can add regions to a data zone without prior notice. If your obligations name specific countries rather than "the EU" or "the US", a data zone may not be precise enough.

The Developer type is for fine-tuned model evaluation only, has a fixed 24-hour lifetime, and carries no residency guarantee. It should never back a production agent.

## How do you stop teams deploying Global types?

Azure Policy can deny specific deployment SKUs. Microsoft's example matches `Microsoft.CognitiveServices/accounts/deployments` with `sku.name` equal to `GlobalStandard`. Add an effect such as `deny` and repeat the condition for each SKU you want to block:

```json
{
  "mode": "All",
  "policyRule": {
    "if": {
      "allOf": [
        { "field": "type", "equals": "Microsoft.CognitiveServices/accounts/deployments" },
        { "field": "Microsoft.CognitiveServices/accounts/deployments/sku.name",
          "in": ["GlobalStandard", "GlobalProvisionedManaged", "GlobalBatch", "DeveloperTier"] }
      ]
    },
    "then": { "effect": "deny" }
  }
}
```

Assign it at the management group that holds your regulated subscriptions. Azure Policy also supports remediation for existing resources, so audit first, then deny. See [restricting models in Microsoft Foundry with Azure Policy](/blog/restrict-models-in-microsoft-foundry-with-azure-policy) for the companion model allowlist.

## A residency checklist for Azure-hosted agents

1. List every agent and the Azure OpenAI deployment it calls, including deployment type.
2. Map each agent's data to a residency requirement: any region, data zone, or specific geography.
3. Move agents with geography requirements to Standard or Regional Provisioned in the right region.
4. Apply the Azure Policy deny for Global and Developer SKUs in those subscriptions.
5. Check other paths: agents that also call non-Azure endpoints, or use features with their own processing location.
6. Re-check quarterly, since data zones can grow and new deployment types appear.

The same question exists on other clouds: see [Bedrock cross-region inference and data residency](/blog/bedrock-cross-region-inference-and-data-residency), and for gateway-level enforcement, [how to enforce data residency for LLM and agent traffic](/blog/how-to-enforce-data-residency-for-llm-and-agent-traffic). Our [cross-platform governance pillar](/blog/how-to-govern-ai-agents-across-platforms) ties these together.

## How Opencontroller does this

Opencontroller auto-discovers agents, models and API keys across Azure, AWS and Google Cloud, so you can see which agents call which model deployments before you tighten policy. Its LLM Gateway routes model traffic with fallback and retry, which lets you keep regulated agents pointed at in-geography deployments, and its runtime adapters cover Microsoft Foundry. It can run as Lyzr SaaS or in customer-controlled infrastructure. [Book a demo](https://www.lyzr.ai/book-demo/).

## Frequently asked questions

**Does Global Standard store my data outside my region?**
Microsoft states that data stored at rest remains in the designated Azure geography for all deployment types. Global types may process inference in any Azure region.

**Is Data Zone the same as regional?**
No. A data zone is a Microsoft-defined group of regions, such as the US or the EU Data Boundary. Standard and Regional Provisioned keep processing within your chosen geography.

**Can I block Global deployments for all teams?**
Yes. Use an Azure Policy that denies `Microsoft.CognitiveServices/accounts/deployments` with the Global SKU names, assigned at management group or subscription scope.

## Key facts

- Azure OpenAI deployment types (Microsoft Learn): Global Standard/Provisioned/Batch process in any Azure region; Data Zone Standard/Provisioned/Batch process within the data zone; Standard and Regional Provisioned (ProvisionedManaged) process within the customer-specified Azure geography; Developer (DeveloperTier) has no data residency guarantee or SLA and a 24-hour lifetime.
- Data at rest remains in the designated Azure geography for all deployment types.
- Data zones: US, EU (EU Data Boundary, can include EFTA countries such as Norway and Switzerland) and Asia Pacific; Microsoft can add regions to a data zone without prior notice.
- Azure Policy can deny deployment SKUs via the field Microsoft.CognitiveServices/accounts/deployments/sku.name.
- Opencontroller by Lyzr discovers agents, models and API keys across Azure, AWS and Google Cloud and runs as SaaS or in customer-controlled infrastructure; pricing starts from $250K per deployment per year.

## Related

- [Does Amazon Bedrock cross-region inference move data out of your region?](https://opencontroller.lyzr.com/blog/bedrock-cross-region-inference-and-data-residency.md)
- [How do you enforce data residency for LLM and AI agent traffic?](https://opencontroller.lyzr.com/blog/how-to-enforce-data-residency-for-llm-and-agent-traffic.md)
- [How do you restrict which models teams can deploy in Microsoft Foundry?](https://opencontroller.lyzr.com/blog/restrict-models-in-microsoft-foundry-with-azure-policy.md)
- [More on Platform Guides](https://opencontroller.lyzr.com/blog/cluster/platform-guides)
