SQL Server 2025

Organizations are using generative AI to stay ahead of the competition, but the real advantage lies in harnessing the power of your own data securely and at scale.

SQL Server 2025, now in public preview, empowers customers to develop modern AI applications securely using their data, complete with best-in-class security, performance, and availability. It provides built-in, extensible AI capabilities, enhanced developer productivity, and seamless integration with Microsoft Azure and Microsoft Fabric—all within the SQL Server engine using familiar T-SQL language. 

Your data, any model, anywhere 

One of the most exciting new capabilities of SQL Server 2025 is the integration of AI directly into the database engine, enabling more intelligent search. With built-in vector search capabilities, you can perform semantic searches over your own data to find matches based on similarity, alongside full text search and filtering you are already using in SQL Server. This built-in capability opens up a host of exciting new use cases such as discovering deeper connections within large datasets, providing a natural conversational experience across various enterprise systems.  

SQL Server 2025 introduces enhanced model management by building model definitions directly into T-SQL, enabling seamless integration with popular AI services such as Azure AI Foundry, Azure OpenAI Service, OpenAI, Ollama, and others. Models are all accessed through REST APIs allowing you to deploy any model securely isolated from the SQL Server engine, anywhere from ground to cloud. As developers test embedding models to find the best fit for their use cases, whether running open-source models on laptops or hosting purpose-trained models, SQL Server 2025 makes it convenient to switch models without needing to change the code. 

This release also provides other essential building blocks for AI development and operational retrieval-augmented generation (RAG) patterns powered by AI agents. It includes vector embedding generation and text chunking built into T-SQL, using Disk Approximate Nearest Neighbor (DiskANN) as a vector index for faster, resource-efficient, and accurate results. Additionally, SQL Server 2025 offers seamless integration with popular AI frameworks like LangChain, Semantic Kernel, and Entity Framework Core. 

“With the new semantic search and RAG capabilities in SQL Server 2025, we can empower existing GenAI solutions with data embeddings to create next-generation, more intelligent AI applications. By connecting systems, we deliver a seamless, natural conversational experience across enterprise environments.”

—Markus Angenendt, Data Platform Infrastructure Lead, Kramer & Crew  

Microsoft’s most significant release for SQL developers in the last decade 

We understand that developers need the right tools and interfaces for modern, data-intensive applications. SQL Server 2025 delivers a rich set of feature enhancements that significantly streamline development process, reduce code complexity, and improve developer productivity. Along with built-in AI capabilities, this release makes SQL Server 2025 the most significant release for SQL developers since the introduction of SQL Server 2016 a decade ago.  

Enhancing data enrichment is our first area of focus in this release. SQL Server 2025 offers native JSON support, empowering developers to process JSON documents natively. Combined with REST APIs and Regular Expressions (RegEx) enablement, developers can now enrich, validate, and manipulate their datasets with external data sources. This allows for building more dynamic, enterprise-grade applications with richer functionality and enhanced performance. 

Empowering developers to build real-time, event-driven applications with SQL Server is another scenario that this release unlocks. Change Event Streaming allows users to consume transaction log changes as events directly from SQL Server to Microsoft Azure Event Hubs. This provides a new method to mitigate some of the issues developers have seen with the Input/Output (I/O) overhead of Change Data Capture (CDC). It also opens new possibilities such as developing real-time, event-driven applications powered by AI agents. 

There’s also excitement on the language and tooling front. We’re thrilled to announce the preview of our new open-source Python driver for SQL Server.¹ Built from the ground up, this driver offers Python developers a robust, efficient, and fully open-source solution for connecting to SQL Server, as simple as pip install. In addition, we are bringing AI-powered assistance directly into your workflow with the integration of MSSQL Extension for Visual Studio Code with GitHub Copilot, now in preview. With GitHub Copilot aware of your SQL Server database connection, developers can generate SQL and object-relational mapping (ORM) migrations, explore schemas, optimize queries with intelligent suggestions, and streamline database interactions—all within in Visual Studio Code.  

Finally, we are introducing a new Standard Developer edition to complement our existing Developer edition, which has all Enterprise features. This will give developers a way to build and test against the Standard edition surface area without incurring license costs.  

“I am genuinely enthusiastic about the AI advancements in SQL Server 2025. These features, along with the enhancements in RegEx and JSON data support, promise to make AI functionalities accessible to a broader range of software applications, and significantly enhance our database operations.” 

—Jacob Saugmann, SQL Specialist, J.H. Schultz Information A/S

Best-in-class security, performance, and availability 

This release builds on SQL Server’s history as an industry leader in database security, performance, and availability. For our enterprise customers, security is non-negotiable. SQL Server remains as the most secure database in the last decade.² SQL Server 2025 continues the product’s legacy of top-notch security by incorporating modern identity and encryption practices. Support for Microsoft Entra managed identities improves credential management and reduces potential vulnerabilities.³

We’re bringing Optimized Locking to SQL Server, to reduce lock memory consumption and minimize blocking for concurrent transactions through Transaction ID (TID) Locking and Lock After Qualification (LAQ). This capability enables customers to increase uptime and enhance concurrency and scale for SQL Server applications. 

SQL Server 2025 has over 50 enhancements made to the database engine including key improvements for HADR all based on customer feedback. This new release will bring enhancements to performance for applications with no code changes required through Intelligent Query Processing (IQP) and columnstore improvements, enhancements for query processing, and enabling Query Store for readable secondaries.  

“Security Cache Improvement proved invaluable for high-demand environments like ours, reducing disruption when applying permissions on servers with 20,000–25,000 active connections. This enhancement ensures minimal performance impact, streamlining security management. The ordered non-clustered Columnstore index significantly improved query performance by over 63%, optimizing workloads reliant on analytical processing.”

—Madhab Paudel, Database Engineer, Entain

Cloud agility through Azure 

To build scalable analytics, data needs to be extracted, transformed, normalized, and made available in a central place. SQL Server 2025 will support database mirroring in Fabric, giving you near real-time analytics using a zero extract, transform, and load (ETL) experience and allowing you to offload your analytical workloads to Fabric.³  

Azure is a critical component of SQL Server. With Azure Arc, SQL Server 2025 will continue to offer cloud capabilities to enable customers to better manage, secure and govern your SQL estate at scale across on-premises and cloud. 

“Fabric mirroring for SQL Server 2025 helps MSC to build the bridge to bring our operational data into Microsoft Fabric.”

—Javier Villegas, IT Director of DBA and BI Services, Mediterranean Shipping Company

Get started with SQL Server 2025 today

With every AI-powered query for hybrid search, every millisecond saved in query execution, every change event streamed in real time, SQL Server 2025 is a critical building block for modern data-intensive applications in this AI era. Ready to try it out? Learn more about SQL Server 2025.

SQL Server Management Studio (SSMS) 21, now generally available, is based on Visual Studio 2022 and includes 64-bit support. This modernized version is available from the Visual Studio Installer, offers automatic updates, and introduces Git integration, query editor enhancements, and a new connection experience.

Microsoft Copilot in SSMS, now in preview, is available as an optional workload when installing SSMS 21, and assists customers in writing, editing, and fixing T-SQL queries using natural language. Leveraging database context, it also helps with database administration, maintenance, configuration and more.³

Explore solutions and capabilities

• Learn more from our Microsoft Build 2025 session: SQL Server 2025: The Database Developer Reimagined.
• Deep-dive blog on Tech Community.
• Explore solutions from Microsoft SQL Server.
• MSSQL Extension for Visual Studio Code with GitHub Copilot.

---

¹ Public preview of Python driver is June 1, 2025. The alpha version is available on GitHub today. 

² According to the National Institute of Standards and Technology Comprehensive Vulnerability Database, as of December 2024.

³ Although SQL Server 2025 in public preview is free to try, using some features such as Microsoft Entra, Fabric and Copilot in SQL Server Management Studio 21 could incur costs based on usage. Try Azure for free and explore Fabric trial capacity.