Linux Foundation Launches OpenSharing, an Open Protocol for Cross-Organization AI Asset Exchange Built on Databricks' Delta Sharing
The Linux Foundation and Databricks unveiled OpenSharing on June 10, a vendor-neutral protocol that extends Delta Sharing's zero-copy model to agent skills, AI models, and unstructured data.
Overview
The Linux Foundation on June 10 announced the launch of the OpenSharing Project, which it describes as “an open, vendor-neutral protocol designed to standardize how organizations share AI assets and data,” according to the Linux Foundation. Hosted by the Linux Foundation and contributed by Databricks, the project evolves the company’s existing Delta Sharing protocol to cover not just tables of data but the broader set of assets that enterprises now move between systems to build AI applications.
The foundation says OpenSharing provides “the first unified framework for exchanging agent skills, AI models, and unstructured data volumes across disparate platforms,” per the Linux Foundation. The announcement coincided with Databricks unveiling the protocol; InfoWorld reported that “Databricks on Wednesday unveiled OpenSharing, a new open protocol designed to let enterprises share AI models, agent skills, dashboards, and unstructured data across platforms without having to copy or move those assets.”
What We Know
OpenSharing builds directly on Delta Sharing, the open data-sharing protocol Databricks created. According to the Linux Foundation, the new project “evolves the widely adopted Delta Sharing protocol to meet the requirements of the agentic era,” providing “the first unified framework for exchanging agent skills, AI models, and unstructured data volumes across disparate platforms.”
The motivating problem, as the foundation frames it, is fragmentation. “As enterprises accelerate the deployment of agentic AI, the lack of a standardized exchange protocol has forced organizations to rely on point-to-point integrations or proprietary marketplaces,” the Linux Foundation said, adding that OpenSharing “eliminates these silos by enabling secure, cross-organizational sharing through a single, open protocol.” By abstracting underlying storage complexities, the Linux Foundation said, the project “allows enterprises to publish AI assets and data that can be consumed by anyone, regardless of their specific cloud environment or platform.”
A central design element is that assets are shared without being copied. InfoWorld described the mechanism as a “zero-copy credential vending model that allows recipients to securely access shared assets directly from a provider’s cloud storage using temporary, scoped credentials rather than requiring the assets themselves to be copied, moved, or replicated.” The scope of shareable content extends beyond conventional tables: InfoWorld reported the protocol lets enterprises “share AI models, agent skills, dashboards, and unstructured data across platforms without having to copy or move those assets.”
On interoperability, the foundation says OpenSharing reaches beyond the original Delta Sharing client base. “Building on Delta Sharing’s open connectors that support a wide range of platforms, OpenSharing expands this cross-platform interoperability with support for Iceberg IRC clients, expanding the universe of reachable recipients,” according to the Linux Foundation.
Linux Foundation chief executive Jim Zemlin tied the move to the foundation’s open-governance model. “OpenSharing addresses a critical need for a common, vendor-neutral framework that enables organizations to exchange AI assets securely and interoperably across platforms and ecosystems,” said Jim Zemlin, CEO, Linux Foundation, in the announcement. “By bringing this technology to the Linux Foundation, we can foster open collaboration, broad industry participation, and the shared governance needed to accelerate AI innovation at scale.”
Matei Zaharia, Co-founder and CTO of Databricks, cast OpenSharing as a continuation of the strategy behind Delta Sharing. “Delta Sharing proved the industry would choose open over locked-in,” he said in the press release. “OpenSharing extends that principle to the full AI stack, while expanding the cross-platform ecosystem to Iceberg recipients and on-premises providers. The agentic era deserves an open foundation, and OpenSharing delivers it.”
Several organizations endorsed the launch with supporting statements in the press release, including Cotality, Kythera Labs, LSEG, MinIO, and Stripe. Storage vendor MinIO emphasized the on-premises angle: “Native open source OpenSharing in AIStor opens up access to data that cannot move,” said AB Periasamy, Co-Founder and Co-CEO, MinIO. Stripe pointed to its own pipeline product, with Emily Sands, Head of Data and AI, Stripe, saying that “Leveraging OpenSharing natively within Stripe Data Pipeline ensures that our users can securely and effortlessly unlock advanced analytics and AI capabilities on their customer, billing, and transaction data.”
What We Don’t Know
The foundation says further details on contributing to or integrating the protocol “will be available soon” at the project’s GitHub repository and website, per the Linux Foundation, leaving the maturity of reference implementations and the breadth of client support open questions at launch. The announcements did not detail a timeline for the project advancing within the Linux Foundation’s governance process, nor the full roster of client and platform vendors committed to implementing it.
Analysis
Industry analysts framed OpenSharing as an attempt to address a cost that has grown alongside enterprise AI adoption. “Every organization building AI, such as multi-agentic systems, is hitting the same wall, i.e., the model, the skill, and the consumer reside on three different platforms. The integration tax is enormous, and it grows exponentially with every new partner, customer, or internal team,” Ashish Chaturvedi, Leader of Executive Research at HFS Research, told InfoWorld.
Dion Hinchcliffe, Lead of CIO Practice at The Futurum Group, pointed to the operational overhead that accumulates around AI assets. “Today, hidden costs include more than just model development. It is the endless packaging, translation, sync, and governance effort required to operationalize AI assets across organizational boundaries,” he told InfoWorld.
Stephanie Walter, Practice Lead of AI Stack at HyperFRAME Research, argued the value enterprises want to share has shifted beyond raw data. “Enterprises are quickly realizing that the value is no longer just in the dataset. It is in the governed context, logic, and intelligence built around the dataset. Existing approaches can share datasets well, but they often do not address the broader AI package,” she told InfoWorld.
OpenSharing joins a Linux Foundation portfolio that already spans widely used infrastructure projects, including Linux, Kubernetes, the Model Context Protocol, PyTorch, and RISC-V, according to the press release. Whether OpenSharing achieves similar reach will depend on how broadly client and platform vendors adopt it beyond the founding contributors.