Microsoft and Snowflake: Delivering on the promise of openness and interoperability
WRITTEN BY
/en-us/microsoft-fabric/blog/author/arun-ulagaratchagan
This blog post is co-authored by Christian Kleinerman, Executive VP of Product, Snowflake.
The typical organization’s data estate now includes hundreds of specialized and often disconnected applications—all of which generate data that you need to understand. To capture and analyze this data, each department and team has their own data and AI service that meets their specific needs, whether that be ease-of-use, familiarity, or specialized functionality. Interoperability between not only the applications, but the data platforms themselves, is no longer a technical aspiration but a necessity for most businesses.
Microsoft and Snowflake announced a shared vision one year ago with this in mind: simplify interoperability, reduce data movement, and accelerate insights by enabling seamless data access between Snowflake and Microsoft OneLake—our single, unified SaaS data lake—to enable mutual customers to more easily access all their data. This vision is anchored in open standards like Apache Iceberg and Parquet, allowing customers to use one copy of data across platforms and choose the right tool for the job at hand. This approach can help our customers do what makes the most sense for their business without creating data silos or duplicating data.
Today, we’re excited to share the progress we’ve made to expand interoperability including what’s available now, what’s in preview, and what’s coming next.
How interoperability works
Snowflake and Microsoft OneLake interoperate through open standards. Snowflake supports Apache Iceberg tables stored in OneLake, so you can create and manage Iceberg tables in OneLake and access them from both Snowflake and Fabric engines without moving or duplicating data. It also means you only need to load the data into the lake once, and all the engines can operate on the same copy of data. This single copy approach means teams can collaborate on a single source of truth rather than fragmenting information across data platforms.
OneLake uses open formats (Parquet, Delta, Iceberg) and exposes standard endpoints, which means your existing tools and SDKs work without custom connectors. This ensures that data remains in OneLake while being accessible by multiple engines, giving you one copy of data for analytics, AI, and BI. While most Fabric engines store their data using the delta lake format and Snowflake uses Iceberg, this is not a problem. Microsoft OneLake seamlessly provides both metadata formats automatically so that all data can work in any platform.
What’s available (GA and preview)
OneLake supports Iceberg, and Snowflake supports OneLake. With many items now going into general availability, you can access this bi-directional data support, enabling seamless interoperability without data duplication.
We are excited to announce the following features are moving into either preview or general availability in the next few weeks:
- General availability: Automatic translation of Iceberg metadata to Delta Lake metadata for use with all Fabric engines.
- General availability: Shortcut Snowflake Iceberg data stored in Azure, Amazon S3, or GCS into your existing OneLake.
- Preview: Automatic translation of Fabric data to Iceberg format for easy use within Snowflake.
- Preview: New OneLake table APIs which seamlessly integrate with Snowflake's catalog-linked database feature.
And finally, you can store Snowflake Iceberg data natively in OneLake, already available in preview.
Coming soon
- Deeper experience integrations designed to make setup and collaboration between the Fabric and Snowflake services even easier.
One copy across platforms—why this matters
The ability to use a single copy of your data across both Snowflake and Fabric enables several compelling use cases:
- Build an open lakehouse on your terms: Pick the right engine and tool across either platform depending on the task at hand, your skillset, or your needs.
- Turn Microsoft 365 into hubs for discovering and applying insights: With the OneLake catalog embedded in hundreds of the most widely used apps like Microsoft Teams, Microsoft Excel, and Microsoft Copilot Studio, you can help all your business users discover and use data. For example, your Teams users can infuse data into their everyday work with embedded channels, chat, and meeting experiences. And security defined in OneLake travels with the data so you can ensure business users only access what they should.
- More efficiently scale team resources and shift from data movement to drive innovation: With a single copy of data across Snowflake and Fabric, you can spend less time organizing and governing your data estate, reduce the costs of duplicating data across various locations, and dedicate more time to discovering insights.
- Deliver high fidelity AI and analytics data products by unifying your data: Whether you’re building an agent in Microsoft Copilot Studio or curating a Power BI dashboard, you can integrate your data in Snowflake with your Microsoft solutions to enrich your results without any data movement. Likewise, data stored in OneLake can be extended to Snowflake to build AI apps or more easily share data, among many other workloads.
- Build a single, connected and governed view of your open lakehouse: Integrate OneLake via Iceberg REST APIs to Snowflake using Catalog Linked Databases to securely centralize and access all your Iceberg tables from a single governed pane of glass.
Get started building with Snowflake and Fabric
- Request an exploration call or join the latest private preview of Iceberg Table Virtualization by signing up here.
- Try the Snowflake Quickstart for Iceberg in OneLake.
- Read the original partnership blog for architecture details.
- Explore our OneLake developer guidance for endpoints and SDKs.
- Learn how to build data-rich agents at FabCon Vienna 2025.
