Snowflake (NYSE: SNOW) has announced significant advancements to its AI Data Cloud, including the forthcoming general availability of Snowflake Postgres, designed to ensure data is AI-ready by integrating transactional, analytical, and AI capabilities on a unified platform, alongside new governance and resilience features for enterprise-scale AI systems.
The introduction of Snowflake Postgres aims to eliminate traditional data silos and complex pipelines that separate transactional and analytical databases. By enabling the world’s most popular database to run natively within the AI Data Cloud, enterprises can consolidate their diverse data workloads onto a single, secure platform. This approach supports full compatibility with open source Postgres, allowing organizations to migrate existing applications without requiring code modifications.
Snowflake Postgres, powered by pg_lake—a set of PostgreSQL extensions—facilitates direct querying, management, and writing to Apache Iceberg tables using standard SQL within a familiar Postgres environment. This integration is intended to remove costly data movement between transactional and analytical systems. Jake Hannan, Head of Data at Sigma Computing, stated, “With Snowflake Postgres, we can work directly on fresh transactional data inside Snowflake without relying on complex pipelines or external systems. That gives our teams and customers a simpler, more reliable foundation to build governed analytics and AI-powered experiences that respond in real time.”
Rob Sandberg, SVP and Head of Advisory Consulting at BlueCloud, further commented on the impact of this development. “For BlueCloud, Snowflake Postgres represents a major opportunity to help our customers eliminate data pipelines, without compromising performance. Its enterprise-grade Postgres foundation brings real credibility, particularly for the financial services organizations we support. With Snowflake Postgres, we can deliver low-latency transactional workloads alongside analytics and AI on a single platform, reducing overhead and helping our customers be more agile in meeting their business goals.”
As AI deployments transition from experimentation to production, Snowflake is also enhancing its platform with additional features for data governance, openness, and resilience to meet enterprise demands. These advancements aim to ensure data remains open, governed, and robust across various engines, formats, and environments.
Snowflake Horizon Catalog now provides context and governance for AI across all data, enabling consistent enforcement of governance policies even when Snowflake data is queried from other engines. This feature supports secure access to Apache Iceberg tables (now generally available) and will soon allow creation, updating, or management of data stored in Iceberg tables (public preview soon). Science and technology company Merck and connected vehicle intelligence provider Motorq are leveraging this capability.
To promote seamless data collaboration across open formats, Open Format Data Sharing extends Snowflake’s zero-ETL sharing model to include formats like Apache Iceberg and Delta Lake. This enables secure data sharing across teams, clouds, and regions without data duplication or fragile pipelines. A new integration with Microsoft OneLake (now generally available) provides mutual customers with secured bidirectional read access for Iceberg data managed by Snowflake or Microsoft Fabric.
Furthermore, Snowflake Backups (now generally available) bolster data resilience by protecting business-critical data. This feature aims to enable quicker recovery from ransomware or disruptions, ensuring data remains unaltered or undeleted once created, thereby providing greater confidence in data preservation during unexpected events or security incidents.
Christian Kleinerman, EVP of Product at Snowflake, emphasized the strategic importance of these developments. “As businesses move from AI experimentation to production, the real challenge is ensuring AI systems can consistently access data that is connected, governed, and discoverable across the enterprise. That means eliminating data silos, fragile pipelines, and closed systems that slow down AI deployment and increase risk. By bringing unified operational and analytical data, as well as open interoperability together in one platform, we’re empowering customers to develop enterprise-ready AI systems that work with real business data, securely and at scale.”
