Field noteNov 16, 20266 min read1 source

Databricks launched Publish to Power BI for Unity Catalog metric views, currently in Private Preview.

In one minute

Databricks has introduced Publish to Power BI, a Private Preview feature that turns Unity Catalog metric views into Power BI semantic models. It gives teams a route to keep business definitions in Databricks while continuing to build and consume reports in Power BI. The important design choice is one-way ownership: metric logic originates in Unity Catalog rather than being maintained separately in each reporting tool.

What changed: Databricks adds private-preview publishing of Unity Catalog metric views to Power BI semantic models, keeping metrics governed and synchronized across reporting and AI experiences.

What happened

Databricks says Publish to Power BI connects Unity Catalog metric views to a Power BI semantic model. Each metric view becomes a table in that model, while dimensions become columns and measures become DAX measures. The company presents this as a way to retain one definition of business terms such as revenue, margin, or active customers across reporting, notebooks, AI applications, AI/BI dashboards, and Genie.

The metric view remains the source of truth. Power BI queries the metric view itself as though it were a table, according to Databricks, while joins defined in the metric view run in Databricks. Joined dimensions, such as campaign channel or region, then appear as ordinary columns in Power BI. A team therefore does not need to recreate the table relationships there. In the company’s campaign-performance example, a view contains channel and region dimensions alongside measures for total spend, total revenue, and return on ad spend. A published Power BI model exposes those dimensions and translates the measures into DAX.

Databricks lists support for standard aggregations including SUM, COUNT, AVG, MIN, MAX, MEDIAN, STDDEV, and VARIANCE. It also lists distinct counts, ratios, filtered measures, measures composed from other measures, and current, offset, and trailing window measures. Descriptions and display formats, including currency, decimal, and percentage settings, carry into Power BI field properties. DirectQuery is the default storage mode for live queries against Lakehouse/RT or Databricks SQL; Import mode loads data into Power BI instead.

Teams can publish on demand from Catalog Explorer, or automate the action with a Power BI task in Lakeflow Jobs. Databricks says a job can run on a schedule or after upstream tables refresh. When a metric view changes, updated measures, new measures, and new dimensions can consequently flow into the target semantic model without manual rebuilding. The documented setup sequence is to prepare named fields and measure inputs, select Publish to Power BI in Catalog Explorer, authenticate with Power BI credentials, choose a target workspace, and validate the resulting table, columns, and DAX measures in Power BI.

Governance differs by storage mode. With DirectQuery, Databricks says each query honors Unity Catalog access controls and security policies, while lineage remains connected to metrics used elsewhere in the Databricks environment. Lakehouse/RT is described as enforcing Unity Catalog row filters, column masks, and ABAC policies on managed Delta tables. In Import mode, data is loaded into Power BI, and permissions on that copy follow Power BI’s controls. The release is one-way: Power BI edits do not write back to Unity Catalog. Teams must alter the metric view in Databricks and republish it, either manually or through a Lakeflow Job. Databricks says interested customers can test the Private Preview through their account team.

What the archive adds

New Runtime's earlier coverage of Google Cloud Knowledge Catalog's GA ingestion of semantic models described another attempt to make semantic definitions portable across data systems. This is a new integration event: Databricks is publishing its own governed metric views outward into Power BI rather than ingesting models into a catalog.

Trend signal

This modestly supports agent-ready software exposes capabilities. A governed semantic layer makes business definitions reusable by multiple consumers, including AI tools, instead of leaving definitions locked inside a single reporting interface. But this is principally a BI interoperability release, not a new permissioned capability designed for external agents, so it is supporting evidence rather than a decisive example.

Why it matters

Teams often operate Power BI as the reporting surface while using a separate warehouse, lakehouse, or agent environment as the place where data is modeled. That split creates a familiar failure mode: a KPI is reimplemented in several places, then changes in one system fail to reach the others. Databricks is offering a workflow that preserves Power BI for self-service reports while assigning ownership of metric logic to Unity Catalog.

The distinction between DirectQuery and Import is operationally important. DirectQuery keeps Unity Catalog policy checks in the path of each query, according to Databricks. Import creates a Power BI copy whose access is controlled there. Teams that adopt the feature should decide which boundary matches their reporting needs rather than treating the modes as equivalent performance settings. They should also establish a clear review process for metric-view edits, because Power BI is a downstream consumer and cannot become the authoritative place to amend a published definition.

What happens next

  • Databricks will broaden the preview or state a general-availability path for Publish to Power BI. (6 months, 60%). Watch for: documented eligibility, regional coverage, or public pricing.
  • Early users will test whether scheduled Lakeflow publishing avoids semantic drift in production reporting. (3 months, 70%). Watch for: customer implementation details on refresh failures, permissions, and schema changes.

Limits and uncertainty

Publish to Power BI is in Private Preview, so broad availability and production readiness are not established by Databricks’s announcement. The company does not quantify query latency, operating cost, or the behavior of synchronization when a job fails, credentials expire, or a schema changes unexpectedly. It says metric-view materialization will soon benefit Power BI, but gives no release date. The announcement is also a vendor account rather than an independent evaluation, and it does not report customer outcomes. Organizations should test supported expressions, workspace permissions, DirectQuery behavior, and refresh handling before making the integration the basis for critical reporting.

Skill to build now

Inventory one KPI that currently exists in both a reporting model and an AI or data workflow. Write down its authoritative inputs, joins, calculation, owner, and access rules, then identify which copy should be retired if a shared semantic definition becomes the source of truth. If you use Power BI and Databricks, test that definition first with DirectQuery and a restricted user role before automating publication through a Lakeflow Job.

Reader takeaway

Databricks is making Unity Catalog metric views publishable into Power BI while keeping the definition owned by Databricks. For teams already using both platforms, the useful question is not simply whether publishing works: it is whether DirectQuery or Import gives the correct security and operating boundary for the reports that depend on those metrics.

Evidence / sources

  1. 1databricks.com/blog/unity-catalog-metric-views-now-power-bi-one-unified-governed-semantic-layer-all-your-reporting
All Field Notes

Next reads

Read next

Open the graph
  1. 01topicModelsExplore the models topic hub.
  2. 02archiveField NotesOpen the latest editorial analysis.
  3. 03sourcesSourcesSee the original sources this archive cites.
Who read this page?Machine requests, hidden until opened

Loading the privacy-safe route aggregate...

Open the JSON contract