As part of the slew of product development updates out of Microsoft Ignite, Azure HorizonDB, a fully managed PostgreSQL database service from the tech giant that is designed to meet the security, resilience and scale requirements of the AI era, is now generally available.
Microsoft also unveiled Fabric HorizonDB, a version of the new database optimized to bring PostgreSQL into its Fabric platform for analytics and data management, in preview.
PostgreSQL is distinguished by its flexibility and versatility, handling traditional relational database workloads as well as geospatial, time series, JSON and the vector database workloads that have become critical over the past few years due to their role helping enterprises index and operationalize unstructured data and data spread across disparate systems. Because of its adaptability, PostgreSQL is now the most popular database format.
Other new Fabric capabilities include Database Hub, a single environment to monitor and manage databases across customers' Microsoft estates that is now generally available, and database agents, now in preview, to provide users with AI-assisted SQL and PostgreSQL capabilities for observing their database operations.
While beneficial for Microsoft users, the new PostgreSQL database capabilities are not unique, according to Sanjeev Mohan, founder and principal of analyst firm SanjMo. However, Microsoft does distinguish its PostgreSQL capabilities from those of rivals such as AWS and Google Cloud through integrations with the broad Microsoft data and AI ecosystem.
"Every major vendor is converging on the same pattern: Postgres for operational data, an open lakehouse for analytics, and a semantic layer that gives agents business context," Mohan told TechTarget. "Microsoft's edge is integration by tightly connecting the database, OneLake, Power BI semantic models, Fabric Apps and Copilot."
Michael Ni, an analyst at Constellation Research, similarly noted that while the individual capabilities don't differentiate Microsoft from its competitors, the way it integrates them does create separation.
"Microsoft is keeping pace with … AWS and Google to deliver on converged database and AI technologies," he told TechTarget. "All are investing in bringing operational data, analytics, and AI closer together. Where Microsoft may have an advantage is how it packages these capabilities for enterprise customers. ... That could reduce the amount of new infrastructure and integration work required."
However, whether Microsoft can deliver simplicity in practice has not yet been proven, Ni added.
Prioritizing PostgreSQL for AI
Microsoft previously provided PostgreSQL database capabilities with Azure Database for PostgreSQL. Now, with the launch of HorizonDB, Microsoft is providing PostgreSQL database capabilities purpose-built for the unique demands of AI.
"We are reimagining the data layer," Microsoft CEO Satya Nadella said during the keynote address at Ignite. "Agents need databases that can keep up with the speed and scale while giving them the context they need for all of the reasoning and actions. That's why we are building HorizonDB."
HorizonDB,which is built to handle 128 terabytes of data per cluster to serve agents the data volume in real time that they require to deliver optimal results, performed eight times faster than self-managed PostgreSQL in benchmark testing conducted by Microsoft -- independent, third-party test results were not provided.
"Microsoft's edge is integration by tightly connecting the database, OneLake, Power BI semantic models, Fabric Apps and Copilot."

Sanjeev Mohan
Founder and principal, SanjMo
By bringing HorizonDB into Fabric, Nadella noted that Microsoft is fully unifying previously separate transactional and analytical data for agents.
"With HorizonDB and Fabric coming together, agents now can reason over live transactions along with petabytes of historical records in real time," he said. "This vision that we have had for 20-plus years is now finally coming together for [customers] to build the next generation of systems."
Mohan likewise noted that HorizonDB's launch is a significant step for Microsoft as it builds the foundational capabilities enterprises need to develop and deploy agents at scale. Previously, the tech giant merely provided a managed version of open-source PostgreSQL.
With HorizonDB, Microsoft is adding a cloud-native database purpose-built for the modern era rather than one when databases served as sources for data assets such as reports and dashboards.
"PostgreSQL has become the default database for new applications and, increasingly, for AI agents," Mohan said. "[HorizonDB] gives Microsoft customers a Postgres option built for enterprise scale and agentic workloads, not just a hosted copy of the community version."
The new capabilities in Fabric are similarly significant, according to Ni.
Fabric HorizonDB integrates PostgreSQL database capabilities with Fabric to copy tables into Fabric without requiring users to build complex and expensive data pipelines; it also unifies disparate data types to provide a foundation for both AI and traditional BI. Database Hub is designed to help system administrators more easily oversee their swath of databases so they can identify potential problems, prioritize work and more easily apply governance. And database agents are built to enable data teams from reactive monitoring toward automating their database operations.
"The Fabric additions are significant because Microsoft is making it easier to build applications around the data and analytics they already have," Ni said. "Users can more simply create and manage applications that use trusted data, accept input and write changes back to operational systems. That's a practical step toward closing the gap between analyzing the business and acting on … analysis."
The test of their true efficacy will be how well the new features work with external operating systems enterprises use in addition to Microsoft capabilities, he added.
Context still a concern
Beyond launching HorizonDB and augmenting Fabric with new capabilities, Microsoft updated Azure DocumentDB with new deployment options to address data sovereignty, added tools that enable developers to build AI applications in Azure Managed Redis, introduced capabilities designed to help users create purpose-built applications in Power BI, and made the Rayfin service for application development generally available.
Collectively, the new features move Microsoft closer to enabling customers to build agents that reason over massive amounts of data in real time, according to Mohan. However, Microsoft still needs to augment features such as Microsoft IQ and Agent 365 to better serve agents with relevant context in addition to data volume.
"Microsoft has a long track record of handling extreme scale without much fanfare, but scale isn't the hard part -- context is," Mohan said. "Agents need to know what the data means, whether it's current and whether they're allowed to act on it."
Consequently, Microsoft IQ and Agent 365 need to ensure that semantic definitions and ontologies remain current as data changes, govern what agents do in addition existing capabilities that govern what they read, and clearly show customers what its costs to run agents running over massive amounts of data, he added.
Ni, meanwhile, noted that the new HorizonDB and Fabric capabilities complement Microsoft's existing foundation for real-time data processing through Fabric Real-Time Intelligence. But like Mohan, he suggested that what Microsoft needs to focus on is bringing context into the mix.
"Microsoft is assembling many of the pieces, [but] the remaining challenge is bringing real-time signals, current operational state, business context and execution policies together for a particular decision," Ni said. "That's where enterprises will determine whether these capabilities can support reliable agentic operations."
Eric Avidon is a senior reporter for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.