Microsoft has affirmed support for Apache Ossie, a vendor-neutral semantic metadata specification now incubating at The Apache Software Foundation, after the project entered incubation on 22 June 2026. The move matters for Fabric and Power BI teams because Microsoft and Snowflake are working on conversion between an Ossie document, a Snowflake Semantic View and a Power BI semantic model.
Apache Ossie was previously known as Open Semantic Interchange. The project was renamed during its move into the Apache Incubator to avoid confusion with other projects that use the OSI acronym. The specification and the YAML-based format for defining metrics, dimensions and relationships did not change.
What Apache Ossie defines
Ossie is an open specification for semantic layers and ontology. It defines a vendor-neutral format for business metrics, dimensions, relationships, broader business concepts and rules. The intended users are BI platforms, query engines and AI agents that need to consume or produce semantic definitions without losing their business meaning.
The problem is familiar in larger data estates. A metric such as Monthly Active Users may be defined differently in customer relationship management systems, data warehouses and BI tools. Ossie gives those definitions a shared, machine-readable form, so a human analyst or an AI agent is not left to infer which version is authoritative.
For Microsoft, Power BI semantic models are the relevant asset. Microsoft says Power BI semantic models have more than 38 million monthly active users and are used by 94% of Fortune 500 companies. The company is bringing that modelling experience to the Apache Ossie work with Snowflake and other contributors.
Fabric and Snowflake conversion
Microsoft says the current collaboration with Snowflake focuses on cross-platform semantic-layer conversion without moving or duplicating the underlying data. The integration scenario highlighted by Microsoft converts an Ossie document into both a Snowflake Semantic View and a Power BI semantic model. The aim is to define shared semantic context once, then use it across Snowflake and Fabric while keeping the capabilities of each platform.
This work sits alongside existing Microsoft and Snowflake integration points, including Iceberg tables with OneLake, Mirrored Databases from Snowflake, Catalog Integration for OneLake and the Power Query Snowflake connector. The semantic layer work is different from data movement. It concerns the business definitions that sit above the data and are used by BI tools and AI agents.
Microsoft also plans to help establish DAX as an Ossie-recognised query language. It says it will contribute to ontology support, so semantic models can carry business logic beyond analysis by AI agents and into platform-agnostic operational intelligence.
Governance and maturity
Apache Ossie is still an incubating Apache project. The Apache Incubator page lists 17 committers and seven Podling Project Management Committee members, with incubation started on 19 June 2026 and the project recorded as entering incubation on 22 June 2026. Incubation means the project is working towards the governance, infrastructure and community practices expected of established Apache projects.
The community around Ossie is broader than Microsoft and Snowflake. Since the repository opened in November 2025, more than 100 commits and 35 merged pull requests have landed from contributors at Snowflake, Dremio, Salesforce, Databricks, dbt Labs, RelationalAI, GoodData and Honeydew. The participating coalition has grown from 17 launch partners to more than 50 organisations.
Three working groups cover Metric Language, Catalog and Ontology. Potential future work includes a more expressive specification for enterprise models, converters for additional platforms and frameworks, a standard semantic query specification and integration with Apache Polaris. None of those items is predetermined, because changes go through the project’s public discussion and vote process.
For Fabric and Power BI teams, the practical question is not whether every platform will support Ossie at once. It is whether semantic definitions can be managed with less manual recreation between platforms that already coexist in many estates. The first useful test will be conversion quality between real semantic models, especially where measures, relationships and governance rules carry business assumptions that are not obvious from table structure alone.
Teams trying to keep Power BI semantic models, Snowflake Semantic Views and AI agent context aligned will find the same issues in ECS's Data topic, alongside Microsoft Fabric, Power BI, SQL Server and Dataverse. The conference gives attendees sessions, workshops, product group contact and practitioner experience on semantic models, governance and platform choices. Bring the model definitions, conversion assumptions and governance questions that are hard to settle from documentation alone.
Photo: Coolcaesar, CC BY-SA 4.0, via Wikimedia Commons, cropped
This article was written by the ECS editorial team with the assistance of Amal Agent, run.events' AI assistant, which supported the research, drafting and proofreading.