Announcing New OpenSharing and Marketplace capabilities for the AI era

Today, we're shipping three waves of innovation that fundamentally expand what it means to share and collaborate on the Databricks platform. Whether you're a data provider looking to monetize intelligence, a consumer deploying partner apps without moving a byte, or a platform team sharing data & AI assets across clouds and data formats, the announcements below are for you. Let's dive in.

Introducing OpenSharing: Next Evolution of Delta Sharing for the Agentic Era

Five years ago, we pioneered open sharing and collaboration with the Delta Sharing protocol — it is now the most widely adopted open protocol for secure zero-copy data sharing used by thousands of enterprises. Building on this momentum, last week, we announced OpenSharing, the next evolution of Delta Sharing. OpenSharing, now a project of the Linux Foundation, introduces the first open, vendor-neutral protocol for securely sharing AI assets, including Agent Skills, AI models, and unstructured data. 

OpenSharing on Databricks ships with several new capabilities organized into four pillars:

Open client interoperability

OpenSharing speaks every format.

• Share to any Iceberg client (GA): Databricks users can now share data to any external Iceberg-compatible client — including Snowflake and Trino — with full transactional consistency.

• High-performance open client sharing (GA): Achieve best-in-class query performance out-of-the-box with OpenSharing-vended cloud storage credentials when sharing to open ecosystem clients, including SAP Databricks and Snowflake.

• Sharing Lakebase tables and Lakebase CDF (Public Preview): Share live operational data in Lakebase — and its change data feed — to any clients. No ETL, no pipelines required.

• Share foreign Iceberg tables (GA coming soon): Share tables directly from AWS Glue, Snowflake Open Catalog, or any catalog that implements the Iceberg REST spec.

Agentic sharing

Share more than tables — share intelligence.

• Genie Agent Sharing (Beta): Share natural-language AI chat experiences with external partners, giving them curated, conversational access to your data to accelerate time-to-insights.

Secure, governed multi-cloud sharing

Connect anywhere, replicate everywhere, pay less.

• SecureConnect (Public Preview): Instantly connect storage across different clouds without complex firewall rules or DNS configurations. One-click secure connectivity between AWS, Azure, and GCP storage. Read more in the blog.

• Global Distribution (Private Preview): Minimize data egress fees and improve query performance with automatic cross-region and cross-cloud replication.

• Cross-regulatory domain sharing (Public Preview): Share data between Databricks environments across regulatory boundaries, including between AWS GovCloud/GovCloud DoD and US commercial clouds, between AWS GovCloud and AWS GovCloud DoD, and between Azure China and commercial clouds.

Friction-free sharing

Sharing that’s easy to set up and efficient to consume.

• Email-based sharing (Coming soon): Invite any collaborator by simply entering their email address. Databricks sends the invitation and handles the rest.

• CDF support for shared view (Private Preview): Consume Change Data Feed — inserts, updates, and deletes — on shared views to efficiently detect and analyze changes.

Identity Resolution in Clean Rooms

Identity resolution with leading partners is now available in Databricks Clean Rooms. Rather than shipping sensitive data to a third party, customers can stand up a clean room with identity partners such as LiveRamp, Acxiom, TransUnion, and others to resolve their first-party data in their lakehouse against partner identity graphs – all while their underlying 1st party data stays protected. Powered by OpenSharing and governed by Unity Catalog, the process involves no data movement, and partners never see the raw customer records. The result: teams can connect fragmented touchpoints to a durable identifier and build a richer customer profile.

Apps on Databricks Marketplace

Third-party applications are now available on Databricks Marketplace. For the first time, customers can discover, deploy, and run partner-built apps directly within their Databricks environment, with no data movement, no separate infrastructure, and no integration headaches.

Apps on Marketplace collapse the distance between "find a solution" and "run it in production." Customers stay inside their governed Databricks workspace while accessing best-of-breed functionality built by domain experts. Providers get instant distribution to the entire Databricks customer base without building custom integrations.

More than 20 launch partners span financial services, healthcare, advertising, marketing, and analytics, including Meta, Capital One, The Trade Desk, Slalom, Stagwell, Integral, Acxiom, Amperity, Bright Data, Datavant, Adstra, Aimpoint Digital, Celebal Tech, Cotality, LakeFusion, Opsera, Prometheux, Reflex, Spindle Health, Tier One Analytics, and Zeb.

Read more details in this announcement blog

Marketplace Commit Drawdown

Marketplace Commit Drawdown lets customers acquire third-party data and AI solutions directly from Databricks Marketplace using their existing Databricks universal commits — speeding up procurement and removing budget friction. 

See these innovations live at Data + AI Summit (June 15–18, San Francisco). From product deep dives to customer stories from Pfizer, Amadeus, Chevron, McKesson, and more, we have 20+ sessions in the Data Sharing & Collaboration track. Explore the full agenda.