Databricks Unveils OpenSharing for AI Assets, Extending 2021 Delta Sharing Under Linux Foundation
Updated
Updated · Databricks · Jun 10
Databricks Unveils OpenSharing for AI Assets, Extending 2021 Delta Sharing Under Linux Foundation
3 articles · Updated · Databricks · Jun 10
Summary
OpenSharing gives enterprises a vendor-neutral way to share AI models, agent skills and unstructured data across organizations, with the project now hosted by the Linux Foundation and available on GitHub.
The protocol expands Databricks' 2021 Delta Sharing standard, adding Apache Iceberg client support and standard APIs for discovery, authorization and access so providers can publish once and reach more platforms.
Three current storage partners—Everpure, MinIO and Qumulo—let customers connect on-premises or private-cloud assets directly to cloud AI and analytics tools without moving the underlying data.
Databricks said Delta Sharing is already used by thousands of enterprises, including Amadeus, Atlassian, LSEG, SAP, Stripe and TheTradeDesk, while more on-premises partners are slated to join.
Will OpenSharing prevent AI vendor lock-in, or just crown Databricks the ruler of a new AI economy?
Are 'Agent Skills' the start of an open AI market, or just the next front in the tech platform wars?
As AI 'skills' become tradable assets, who is liable when a shared skill causes a catastrophic system failure?
Breaking Data Silos: How OpenSharing Sets a New Benchmark for Secure, Interoperable AI Asset Exchange
Overview
OpenSharing, unveiled by Databricks and the Linux Foundation in June 2026, is an open and vendor-neutral protocol that transforms how AI assets are exchanged. Before OpenSharing, organizations struggled with costly custom integrations and proprietary marketplaces due to the lack of a universal standard, leading to silos and slow innovation. OpenSharing solves this by introducing unified APIs for discovery, authorization, and access, ensuring seamless interoperability across platforms. This fosters an ecosystem where AI models and agent skills can be easily shared and integrated, accelerating development and breaking down barriers to collaborative innovation.