Azure Databricks: Storage-Optimized Vector Search (AI Search) endpoint blocked - "not supported for this workspace"

Question

Azure Databricks: Storage-Optimized Vector Search (AI Search) endpoint blocked - "not supported for this workspace"

Andy McGurty 0

Attempting to create a Storage-Optimized Vector Search (AI Search) endpoint fails with:

STORAGE_OPTIMIZED endpoint type is not supported for this workspace at the moment. Please contact Databricks support for more information.

Confirmed identically across multiple independent paths:

Databricks CLI / REST API (POST /api/2.0/vector-search/endpoints, endpoint_type=STORAGE_OPTIMIZED)
Workspace UI (Compute → AI Search → Create endpoint, Storage-optimized selected)
Two separate workspaces, in two separate Azure regions: australiasoutheast and australiaeast (Serverless compute mode)

STANDARD endpoint type creates successfully in both workspaces, via both the CLI and the UI. Since the failure reproduces identically in australiaeast - a region Microsoft's own documentation lists as fully supporting AI Search - this looks like an account-level entitlement/feature flag rather than a regional rollout gap.

Question: What's required to enable the Storage-Optimized endpoint type for this account? Is this a manual enablement Databricks/Microsoft needs to action, or is there a self-service setting I'm missing?

Thanks

Andy

SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-08T08:28:36.6433333+00:00
Hi @Andy McGurty ,

The error message:

"STORAGE_OPTIMIZED endpoint type is not supported for this workspace at the moment"

indicates that the workspace is currently not enabled to create Storage-Optimized AI Search (Vector Search) endpoints. Since the issue reproduces consistently across multiple workspaces, deployment methods (UI, CLI, and REST API), and Azure regions, while STANDARD endpoints can be created successfully, this suggests that the limitation is likely related to workspace-level feature availability or entitlement rather than a configuration issue in the workspace itself.

As documented by Databricks, Storage-Optimized endpoints are a specialized AI Search offering and may have availability requirements that differ from Standard endpoints. The behavior observed aligns with the service rejecting the endpoint type before endpoint creation begins.

To verify availability, review the AI Search endpoint documentation and confirm whether Storage-Optimized endpoints are supported for your workspace and subscription. If the feature is not available for the workspace, Databricks support may need to confirm feature eligibility or enablement status.

References

Create AI Search endpoints and indexes - Azure Databricks

Announcing Storage-Optimized Endpoints for AI Search | Databricks Blog

Debugging guide for Model Serving - Azure Databricks

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.
SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-18T08:20:23.9033333+00:00

Hi @Andy McGurty ,
I hope you had a chance to review the information shared earlier, and I hope this information has been helpful! If you still have questions, please let us know what is needed in the comments so the question can be answered.

1 answer

Your answer

SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-08T08:28:36.6433333+00:00

Hi @Andy McGurty ,

The error message:

"STORAGE_OPTIMIZED endpoint type is not supported for this workspace at the moment"

indicates that the workspace is currently not enabled to create Storage-Optimized AI Search (Vector Search) endpoints. Since the issue reproduces consistently across multiple workspaces, deployment methods (UI, CLI, and REST API), and Azure regions, while STANDARD endpoints can be created successfully, this suggests that the limitation is likely related to workspace-level feature availability or entitlement rather than a configuration issue in the workspace itself.

As documented by Databricks, Storage-Optimized endpoints are a specialized AI Search offering and may have availability requirements that differ from Standard endpoints. The behavior observed aligns with the service rejecting the endpoint type before endpoint creation begins.

To verify availability, review the AI Search endpoint documentation and confirm whether Storage-Optimized endpoints are supported for your workspace and subscription. If the feature is not available for the workspace, Databricks support may need to confirm feature eligibility or enablement status.

References

Create AI Search endpoints and indexes - Azure Databricks

Announcing Storage-Optimized Endpoints for AI Search | Databricks Blog

Debugging guide for Model Serving - Azure Databricks

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.
SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-18T08:20:23.9033333+00:00

Hi @Andy McGurty ,
I hope you had a chance to review the information shared earlier, and I hope this information has been helpful! If you still have questions, please let us know what is needed in the comments so the question can be answered.

Answer 1

There is no self-service switch that enables storage-optimized AI Search endpoints. The documented workspace prerequisites are Unity Catalog and serverless compute, but storage-optimized endpoints also exclude workspaces using customer-managed keys and FedRAMP-compliant workspaces.

Because Standard endpoint creation succeeds, your general AI Search permissions and API syntax are working. The rejection occurs before provisioning and is therefore a workspace eligibility decision, not a CLI or REST issue.

Check both workspaces for customer-managed keys and any applicable compliance profile. If neither limitation applies, the next step is Databricks support or your Databricks account team. Ask them to verify storage-optimized endpoint entitlement and control-plane availability for each workspace ID. Include both workspace IDs, Azure regions, the complete API response, timestamp, and request ID.

Region support for Azure Databricks or AI Search generally does not guarantee that every endpoint type is enabled on every workspace/control-plane shard. Recreating the endpoint, changing the SDK, or granting additional workspace permissions will not bypass this check. Continue with Standard endpoints until Databricks confirms or enables eligibility.