This article highlights Microsoft's push for a unified, AI-powered data estate. Engineers gain access to new, integrated database solutions like SQL Server 2025 and Azure DocumentDB, simplifying data management and accelerating AI development across hybrid and multi-cloud environments.
As AI reshapes every industry, one truth remains constant: data is no longer just an asset—it’s your competitive edge. The pace of AI demands easy data access, faster insights, and the ability to iterate without friction. Yet many organizations are held back by fragmented data estates and legacy systems. Microsoft Fabric was designed to meet this moment—to unify your data, simplify your architecture, and accelerate your path to becoming an AI-led organization.
That mission is gaining traction at remarkable speed. Since Fabric launched two years ago, it has grown faster than any other data and analytics platform in the industry. More than 28,000 customers—including 80% of the Fortune 500—now rely on Fabric, and its ecosystem continues to expand as partners build solutions to solve the most complex data challenges.
As Fabric becomes the central connection point for data, we’re strengthening the database layer at the heart of your data estate—ensuring you have the scale and performance required for AI.
Microsoft already offers one of the industry’s most comprehensive database portfolios, and we’re expanding it even further—while deeply integrating these capabilities into Fabric. I’m excited to announce the general availability of SQL Server 2025, Azure DocumentDB, and SQL database and Cosmos DB in Fabric, along with the preview of our newest addition, Azure HorizonDB. With these new offerings, you have a world-class database option to build once and deploy at the edge, as platform as a service (PaaS), or even as software as a service (SaaS). And because our entire portfolio is either Fabric-connected or Fabric native, Fabric serves as a unified hub for your entire data estate. Below I’ll cover how these new databases are purpose-built to support your AI projects.
Microsoft has been shaping the SQL landscape for more than 35 years. Now, with the release of SQL Server 2025 into general availability, we’re introducing the next evolution—one that brings developer‑first AI capabilities at the edge, within the familiar T‑SQL experience. Smarter search combines advanced semantic intelligence with full‑text filtering to uncover richer insights from complex data. AI model management using model definitions in T-SQL allows seamless integration with popular AI services such as Microsoft Foundry.
Enterprise reliability and security remain best-in-class. Enhanced query performance, optimized locking, and improved failover help ensure higher concurrency and uptime for mission‑critical workloads. With strengthened credential management through Microsoft Entra ID via Azure Arc, SQL Server 2025 is secure by design. Your data is also instantly accessible for your AI and analytics in Microsoft OneLake with mirroring for SQL Server 2025 in Fabric, now also generally available.
SQL Server 2025 is the most significant release for SQL developers in a decade. And the response to our preview has been overwhelming, with 10,000 organizations participating, 100,000 databases already deployed, and download rate two times higher than SQL Server 2022. If you want to join all those who’ve already adopted SQL Server 2025, download it today.
We’re excited to announce Azure DocumentDB, a new service built on the open-source, MongoDB-compatible DocumentDB standard governed by the Linux Foundation. The first Azure managed service to support multi-cloud and hybrid NoSQL, Azure DocumentDB can run consistently across Azure, on-premises, and other clouds.
Azure DocumentDB gives you the freedom to embrace open source while achieving scale, security, and simplicity. It’s AI-ready, with capabilities like vector and hybrid search to deliver more relevant results. Instant autoscale meets demand, and independent compute and storage scaling keeps workloads efficient. Security and availability is standard, with Microsoft Entra ID integration, customer-managed encryption keys, 35-day backups included, and a 99.995% availability service-level agreement (SLA). And soon, enhanced full-text search will add features like fuzzy matching, proximity queries, and expanded language support, making it even easier to build intelligent, search-driven apps.
Azure DocumentDB is now generally available, so you can try it today. You can also learn more about Azure DocumentDB and all the Azure Database news by reading Shireesh Thota’s, Corporate Vice President of Azure Databases, announcement blog.
PostgreSQL has become the backbone of modern data solutions thanks to its rich ecosystem, extensibility, and open source foundation. Microsoft is proud to be the #1 PostgreSQL committer among hyperscalers, and we’re building on that leadership with Azure HorizonDB.
Now in early preview, Azure HorizonDB is a fully managed, PostgreSQL-compatible database service, built to handle the scale and performance required by the modern enterprise. It goes far beyond open source Postgres, with auto-scaling storage up to 128 TB, scale-out compute up to 3,072 vCores, <1 millisecond multi-zone commit latency, and enterprise security and compliance. Vector search is built-in, along with integrated AI model management and seamless connectivity to Microsoft Foundry so you can build modern AI apps. Combined with GitHub Copilot, Fabric, and Visual Studio Code integrations, it provides an intelligent and secure foundation for building and modernizing applications at any scale. To learn more about Azure HorizonDB, read our announcement blog.
We are also releasing a new class of SaaS databases, both SQL database and Cosmos DB in Fabric, into general availability. Data developers now have access to world-class database engines within the same unified platform that powers analytics, AI, and business intelligence.
Fabric Databases are designed to streamline your application development. You can provision them in seconds, and they don’t require the usual granular configuration or deep database expertise. They provide enterprise-grade performance, are secure by default with features like cloud authentication, customer-managed keys, and database encryption, and come natively integrated into the Fabric platform, even using the same Fabric capacity units for billing.
With Fabric databases, developers now have the flexibility to build applications grounded in operational, transactional, and analytical data. Together, these offerings make Fabric a developer-first data platform that is streamlined, scalable, and ready for modern data applications.
Learn more by reading Shireesh Thota’s, Corporate Vice President of Azure Databases, announcement blog.
We’re making it easier than ever to work with your entire Microsoft database portfolio in Fabric, giving you a single, unified place to manage and use all your data. Building on our existing mirroring support for Azure SQL Database and Azure SQL MI, we’re now announcing the general availability of mirroring for Azure Database for PostgreSQL, Azure Cosmos DB, and SQL Server versions 2016–2022 and 2025. With these databases mirrored directly into Fabric, you can eliminate traditional extract, transform, and load (ETL) pipelines and make your data instantly ready for analytics and AI.
Choosing the right database is essential, but it’s just the beginning. The major opportunity lies in driving frontier transformation, where data becomes the foundation for an AI-native enterprise. We recommend focusing on three core steps:
In this section, I’ll dive into the latest enhancements to Microsoft Fabric that help you achieve every step of your data journey. This includes expanded interoperability in OneLake with SAP, Salesforce, Azure Databricks, and Snowflake, the introduction of Fabric IQ—a new workload that adds semantic understanding—and enhanced agentic capabilities across Fabric to help you build richer, AI-powered data experiences.
This is the future of data, and it’s already within reach. With Fabric and our database innovations, Microsoft is helping organizations move seamlessly from insight to action—unlocking the full potential of your data and the AI built on top of it.
Microsoft OneLake unifies all your data—across clouds, on-premises, and beyond Microsoft—into a single data lake with zero-ETL capabilities like shortcuts and mirroring. Alongside the additional mirroring sources for Microsoft Databases, we’re also introducing the preview of shortcuts to SharePoint and OneDrive. This allows you to bring unstructured productivity data into OneLake without copying files or building ETL pipelines, making it easier to train agents and enrich your structured data.
Once connected to OneLake, your data becomes easily discoverable in the apps your teams use every day like Power BI, Teams, Excel, Copilot Studio, and Microsoft Foundry. Today, we are taking that a step further with native integration with Foundry IQ—the next generation of retrieval-augmented generation (RAG). Agents rely on context—Foundry IQ’s knowledge bases deliver high-value context to agents by simplifying access to multiple data sources and making connections across information. You can use the OneLake knowledge source in Foundry IQ to connect agents to multi-cloud sources like AWS S3, on-premises sources, and structured and unstructured data.
See how shortcuts and mirroring unify your data in OneLake and fuel the next generation of intelligent agents in Microsoft Foundry:
We are also seeing great momentum with dozens of partners outside of Microsoft deeply integrating with OneLake, including ClickHouse, Dremio, Confluent, EON, and many more. And now, we are thrilled to add new, deeper interoperability with SAP, Salesforce, Azure Databricks, and Snowflake.
First, we’re deepening interoperability with the systems organizations rely on most, SAP and Salesforce. With the launch of SAP Business Data Cloud Connect for Microsoft Fabric, customers can allow bidirectional, zero-copy data sharing between SAP Business Data Cloud (BDC) and Fabric. At the same time, we are working with Salesforce to integrate their data into Fabric using the same zero-copy approach, unlocking advanced analytics and AI capabilities without the overhead of traditional ETL.
We’re also strengthening interoperability with Azure Databricks and Snowflake so you can use a single copy of data across platforms. By the end of 2025, Azure Databricks will release, in preview, the ability to natively read data from OneLake through Unity Catalog, enabling seamless access without duplication or complex data movement. Looking ahead, Databricks will also add support for writing to and storing data directly in OneLake, allowing full two-way interoperability. Read more about this interoperability.
Our collaboration with Snowflake on bidirectional data access continues as well. We are introducing a new item in OneLake called a Snowflake Database and a new UI in Snowflake—both designed to allow OneLake to be the native storage solution for your Snowflake data. We’re also bringing Snowflake mirroring to general availability, allowing you to virtualize your external Snowflake-managed Iceberg tables in OneLake with shortcuts created and handled automatically. Together, these innovations let you run any Fabric workload—whether analytics, AI, or visualization—directly on your Snowflake-managed Iceberg tables.
Learn more about our Snowflake collaboration by reading our latest joint blog or by watching the following demo:
Finally, in close collaboration with dbt Labs, we are also excited to announce built-in support for their industry leading data transformation capability. Now in preview, dbt jobs in Microsoft Fabric let you build, test, and orchestrate dbt workflows in your Fabric workspaces. Learn more in this blog.
As Frontier Firms train agents on their enterprise data, it’s become clear that quality and context matter more than data volume. Agents need business context across relationships, hierarchies, and meaning to turn raw data into actionable insight. That’s why we’re introducing Fabric IQ—a new workload designed to map your datasets to the real-world entities they represent, creating a shared semantic structure on top of your data.
The power of IQ lies in how it unifies disparate data types under a single, coherent framework. Built upon Power BI’s industry-leading, rich semantic model technology, IQ brings together analytical data, time-series telemetry, and geospatial information, all organized under a semantic framework of business entities and their relationships, properties, rules, and actions. You can then create operations agents, a new type of agent in Fabric, which can use this model to act as virtual team members, monitoring real-time data sources, identifying patterns, and taking proactive action. Instead of forcing your teams and even agents to think in terms of tables and schemas, IQ allows you to align data with how your organization operates.
In short, Fabric IQ is designed to model reality with data, so that every insight, prediction, and action is grounded in how your organization actually operates. You can learn more about IQ in Yitzhak Kesselman’s, Corporate Vice President of Messaging and Real-Time Intelligence, announcement blog.
As organizations scale their AI initiatives, the ability to connect intelligent agents with enterprise-grade data is becoming a critical differentiator. Fabric is making this possible with a set of integrated AI experiences: Copilot in Power BI helps you ask questions of your data, Fabric data agents allow deeper analysis, and the new Fabric operations agents let you monitor your data estate and take action in real time. These experiences can be used across Fabric or as foundational knowledge sources in industry-leading AI tools like Microsoft Foundry, Copilot Studio or even Microsoft 365 Copilot to power smarter, more data-rich AI experiences.
Beyond introducing operations agents as part of Fabric IQ, we’re also expanding what data agents and Copilot can do. Along with existing integration with Microsoft Foundry and Copilot Studio, Fabric data agents can now be embedded directly in Microsoft 365 Copilot. This lets business users (with the right permissions) access trusted knowledge from OneLake and transforms Microsoft 365 from a productivity suite into an intelligent insights platform.
They can also act as hosted Model Context Protocol (MCP) servers, making it easy to integrate with other applications and agents across the AI ecosystem. Finally, data agents can now reason across both structured and unstructured data. Thanks to an integration with Azure AI Search, data teams can add their existing unstructured data search endpoints as a source in data agents. Learn more the Fabric data agent enhancements by reading the Fabric AI blog.
We’re also enhancing the standalone experience for Copilot in Power BI with a new search experience. Simply describe what you need, and Copilot will locate the relevant report, semantic model, or data agent and surface the right answers. This standalone experience is also coming to Power BI mobile so you can use it on the go.
Take a look at how you can apply all of these AI experiences together seamlessly:
In short, we’re redefining what it means to have an AI-powered data estate. With data agents, Copilot in Power BI, and operations agents in Fabric IQ, AI is now woven across Fabric. And with native integration to Microsoft Foundry and Copilot Studio, you can easily add Fabric agents as building blocks to create more intelligent, informed custom agents.
You also can see more innovation coming to the Fabric platform by reading Kim Manis’, Corporate Vice President of the Fabric Platform, Fabric blog or by checking out the more technical Fabric November 2025 Feature summary blog.
We are excited to announce SQLCon 2026, which will happen at the same time and the same location as the Microsoft Fabric Community Conference (FabCon), happening March 16–20, 2026 in Atlanta, Georgia. By uniting the powerhouse SQL and Fabric communities, we’re giving data professionals everywhere a unique opportunity to master the latest innovations, share practical knowledge, and accelerate what’s possible with data and AI, all in one powerful week. Register for either conference and enjoy full access to both, with the flexibility to mix and match sessions, keynotes, and community events to fit your interests.
If you’re interested in seeing these announcements live, I encourage you to join my Ignite session, “Innovation Session: Microsoft Fabric and Azure Databases – the data estate for AI” either in person or online at no cost. I’ll not only cover these major announcements but show you how they come together to help you create a unified, intelligent data foundation for AI.
You can also dive deeper into these announcements and so much more by watching the rest of the breakout sessions across Azure Data:
The post Microsoft Databases and Microsoft Fabric: Your unified and AI-powered data estate appeared first on Microsoft Azure Blog.
Continue reading on the original blog to support the author
Read full articleClaude Sonnet 4.6 brings frontier-level reasoning and a 1M token context window to Microsoft Foundry. For engineers, this enables more efficient large-scale code analysis, sophisticated browser automation, and better cost-performance control for agentic workflows in enterprise environments.
Pantone's approach provides a blueprint for scaling niche domain expertise via agentic AI. It demonstrates how a multi-agent architecture supported by a robust NoSQL database like Azure Cosmos DB can transform static data into interactive, high-value creative tools.
This integration brings Anthropic's most advanced reasoning to Azure, enabling engineers to build secure, agentic workflows with a 1M token context window. It simplifies the path to production by combining frontier intelligence with enterprise-grade governance and data connectivity.
PostgreSQL is evolving into a central hub for AI development. By integrating vector search, LLM orchestration, and seamless IDE workflows directly into the managed database service, Microsoft reduces the friction of building and scaling intelligent, data-driven applications.
