Microsoft is fundamentally reshaping its database portfolio to address the demands of artificial intelligence, unveiling a new generation of databases – SQL Server 2025, Azure Document DB, Azure Horizon DB, and Fabric Databases – each redesigned with AI workloads in mind, according to an announcement detailed in a recent podcast interview.
Shireesh Thota, Corporate Vice President of Azure Databases at Microsoft, explained that data is no longer simply an asset but a critical competitive advantage, particularly in the age of AI. “The speed at which AI is evolving requires not only easy data access and faster insights but likewise the ability to iterate quickly without being hampered by fragmented data estates and legacy systems,” Thota said in the interview on The Stack Overflow Podcast.
Microsoft’s Fabric platform is positioned as a central solution to these challenges, unifying data and simplifying architecture to accelerate AI adoption. The platform has seen rapid growth, boasting over 28,000 customers, including 80% of the Fortune 500, and is becoming a central connection point for data, prompting Microsoft to strengthen the database layer underpinning the entire data estate.
Currently available within Fabric are SQL Server 2025, Azure DocumentDB, and SQL database and Cosmos DB. A preview of Azure HorizonDB has also been released. Thota leads product management, engineering, and cloud operations for Azure Databases, overseeing a comprehensive portfolio that includes SQL Server, CosmosDB, and Postgres.
Microsoft’s commitment to the PostgreSQL open source project extends beyond consumption, with a dedicated team contributing to its development. This was highlighted in a recent episode of the “Talking Postgres” podcast featuring Thota, where the importance of contributing to the project was discussed.
The company is also focused on making data more accessible and readily usable for AI applications, a shift signaled by the new database offerings. This evolution is driven by the need for faster insights and the ability to iterate quickly, avoiding the limitations of fragmented data and legacy systems.
Shireesh Thota is based in Redmond, Washington, and has over 500 connections on LinkedIn.
