OpenSharing targets AI asset exchange

The Linux Foundation announced on June 10, 2026 the launch of OpenSharing, an open and vendor-neutral protocol meant to standardize how organizations exchange AI assets and data across organizations, platforms and clouds. The project is hosted by the foundation and contributed by Databricks. Its core idea is narrow but useful: give companies a common framework for sharing agent skills, AI models and unstructured data volumes, instead of relying on custom integrations or proprietary marketplaces for each partnership.

The word "protocol" is important. OpenSharing is not another AI model. It is a technical set of rules for publishing, discovering and consuming resources. It builds on Delta Sharing, a mechanism already used to share structured data across platforms. The change is the move into generative and agentic AI: the items being exchanged are no longer only tables, but also models, software-agent skills and less orderly data, such as documents or media collections. In plain terms, it tries to make AI resources portable in the same way earlier data-sharing standards tried to make datasets portable.

The announcement lands as enterprises are connecting software agents to knowledge bases, business tools and specialized models. Without a common standard, each participant often has to build its own technical bridge, with separate security rules, formats and operational limits. OpenSharing’s promise is to reduce that fragmentation. An organization could publish an asset once and make it consumable by several environments, including stacks that rely on Delta Sharing or Apache Iceberg. Iceberg is an open table format used to make data lakes more interoperable across engines and vendors.

OpenSharing does not automatically solve trust, licensing or data quality. Those still require policies, audits and contracts. What it offers is a neutral governance venue for a layer of infrastructure that could otherwise remain locked inside individual vendors. That is the Linux Foundation’s usual role: moving common plumbing into a broader contribution model. For AI teams, the practical question is whether enough platforms adopt OpenSharing to make it worth using as a default exchange layer, rather than another connector to maintain.