Cosmos DB in Microsoft Fabric (Preview)

Cosmos DB in Fabric makes it easy to build agentic AI apps, offering an AI-optimized database that is automatically configured to meet your application’s needs. It’s built on the high availability, dynamic scaling, and AI-ready capabilities of Azure Cosmos DB. 

Building on the momentum from the launch of SQL database in Fabric, we are expanding the database's workload in Fabric with this new addition. You can now store semi-structured NoSQL data in Cosmos DB in Fabric, alongside your relational data in SQL databases enabling a unified data platform for your applications. 

This further positions Fabric as a complete data platform to handle all your organizational needs from operational to analytics and BI. 

Your existing or new applications can instantly benefit from deep integration with Fabric OneLake, bringing databases, analytics, data science, real-time intelligence, and Copilot-powered BI all in one place, rather than assembling them individually. 

We built Cosmos DB on three key themes, to make it the database of choice for AI apps, without overhead: simplified, autonomous, and optimized for AI. 

Simplified and Autonomous 

We are removing the database management overhead with the right defaults, so you can focus on building your applications. 

Simple to get started 

You can get started with a Cosmos DB database with just a name. Complex settings such as networking, backup, or scale are not required. The database starts within seconds with optimal defaults for your apps. 

You can also easily upload data with a couple of clicks in the familiar data explorer. Exploring with sample data and connecting your apps is just a click away. 

If you want to import custom data, you can do so easily through JSON files or bulk inserts via SDKs. We’re also working on adding pipeline-based data ingestion in the future. 

You can directly query transactional data from the database using familiar SQL queries. 

Automatic scaling with unified CU usage reporting 

Cosmos DB in Fabric automatically configures the right scale settings for most application needs. Currently, Cosmos DB usage is not billed in Fabric. In the future, scale usage will be reported as Consumption Units (CUs) for unified billing. 

Secure by default 

Cosmos DB in Fabric is configured with the right security defaults from the start. 

• Primary authentication uses Microsoft Entra IDs. 

• Workspace permissions are automatically inherited and enforced on data plane requests. 

• Tenant-level Private Links are supported for network security (workspace-level coming soon). 

  
  

Simple application connectivity 

Apps or clients, whether in Azure or Fabric, can connect to the Fabric database using the database endpoint and existing SDKs. 

Supported authentication methods include: 

• Microsoft Entra IDs of users logged into Fabric with workspace permissions. 

• Service Principals for programmatic access. 

  
  

Databases created in Fabric can be managed programmatically via REST APIs, with Application Lifecycle Management (ALM) coming soon. 

Cosmos DB in Fabric currently supports the NoSQL API. 

Automatically available in OneLake 

Cosmos DB data is automatically available in OneLake in Delta Parquet format. This allows you to: 

• Seamlessly join Cosmos DB data with other data in OneLake (SQL, Lakehouses, mirrored data, shortcuts, etc.). 

• Perform cross-container and cross-database queries across Cosmos DB databases. 

Your data is instantly available for analytics and BI in Fabric. You can also write back analyzed data from Spark notebooks into Cosmos DB for low-latency millisecond serving also known as reverse ETL. 

Deep integrations for your applications 

We are adding support for user data functions, which, combined with real-time intelligence, allow you to build powerful, bespoke solutions without managing multiple services individually. 

We’re also working on adding GraphQL API support — enabling queries across SQL and Cosmos DB in Fabric. This gives applications unmatched flexibility to access data seamlessly across databases. 

Optimized for AI 

Cosmos DB in Fabric inherits Azure Cosmos DB’s AI capabilities, including: 

• Vector indexing 

• Full-text search 

• Hybrid search 

These capabilities make Cosmos DB in Fabric ideal for AI and agentic applications, especially those leveraging retrieval-augmented generation (RAG) and real-time intelligence. 

With its simple and autonomous design, Cosmos DB in Fabric makes it easier than ever to start building AI-driven applications. Combined with data science and BI in Fabric, your applications can go further on AI scenarios and use cases with ease.