Content area
Billing is built around consumption, aligning costs with actual database activity rather than reserved infrastructure, which can be helpful for variable or seasonal applications. Fabric Databases support vector data and retrieval-augmented generation patterns out of the box, allowing applications to store embeddings, run semantic searches, and connect directly to Fabric's AI services. Because operational tables and vector indexes live within the same managed environment, organizations can build intelligent applications—such as recommendation systems, conversational interfaces, and anomaly detection tools—on top of their existing datasets while keeping everything synchronized through OneLake. Enterprise-grade security controls apply consistently across SQL and Cosmos DB–style databases, including role-based access, encryption options, and integration with Microsoft Entra ID for identity management. Because the databases are native to Fabric, they inherit centralized policy management, auditing, and data lineage capabilities, making it easier for organizations to demonstrate compliance for regulated workloads that involve sensitive data.
Microsoft has debuted Fabric Databases as a unified, SaaS- native database experience inside Microsoft Fabric that merges SQL Database and Azure Cosmos DB capabilities into a single operational data layer. Announced at Ignite 2025, Fabric Databases are designed to simplify how organizations manage transactional and operational data that feeds analytics and AI workloads. Instead of juggling separate database services and data pipelines, teams can spin up databases within Fabric that are serverless, autonomous, and tightly integrated with Fabric's OneLake data foundation.
In practice, Fabric Databases provide instant provisioning of SQL and NoSQL data stores that run under Fabric's control plane. Developers can choose between SQL-style relational schemas and Cosmos DB–style document and key–value models, but manage them as part of a single platform that handles infrastructure, scaling, and lifecycle tasks. The architecture is serverless, so capacity automatically adjusts to workload demands without manual sizing of compute instances. Billing is built around consumption, aligning costs with actual database activity rather than reserved infrastructure, which can be helpful for variable or seasonal applications.
A major focus of the release is native AI integration. Fabric Databases support vector data and retrieval-augmented generation patterns out of the box, allowing applications to store embeddings, run semantic searches, and connect directly to Fabric's AI services. Because operational tables and vector indexes live within the same managed environment, organizations can build intelligent applications—such as recommendation systems, conversational interfaces, and anomaly detection tools—on top of their existing datasets while keeping everything synchronized through OneLake. This reduces data movement and helps maintain a single, consistent view of important datasets across operational and analytical systems.
Fabric Databases also emphasize unified security and governance. Enterprise-grade security controls apply consistently across SQL and Cosmos DB–style databases, including role-based access, encryption options, and integration with Microsoft Entra ID for identity management. Because the databases are native to Fabric, they inherit centralized policy management, auditing, and data lineage capabilities, making it easier for organizations to demonstrate compliance for regulated workloads that involve sensitive data. Customer-managed keys and granular configuration options help align database deployments with internal security standards and external regulatory requirements.
For developers, Fabric Databases are meant to streamline the path from prototype to production. Teams can build applications that use SQL database or Cosmos DB for operational storage and then query live operational data directly from analytics and reporting tools in Fabric, without maintaining separate ETL jobs. This architecture is particularly useful for scenarios such as real-time dashboards, embedded analytics, and AI agents that need up- to-date transactional context. By centralizing operational and analytical access around OneLake, Fabric aims to reduce the proliferation of duplicated datasets and stale data copies that often accumulate across traditional data warehouses and data lakes.
Early commentary from Microsoft and partner ecosystems positions Fabric Databases as a foundational piece of an AI- first data estate. Sessions and customer stories highlight use cases where business applications, telemetry pipelines, and external data feeds converge into Fabric, with Fabric Databases serving as the operational hub that exposes consistent APIs to both line-of-business apps and AI agents. For organizations already investing in Fabric for analytics, bringing operational databases into the same platform may simplify architecture and reduce the number of separate systems to operate, patch, and secure.
About Microsoft
Microsoft Fabric is a unified data and analytics platform from Microsoft that integrates data engineering, data warehousing, real-time analytics, data science, and business intelligence, now including Fabric Databases as a managed operational data layer for modern applications.
For more information, visit www.microsoft.com/fabric.
Copyright Worldwide Videotex Jan 1, 2026