Linux Foundation OpenSharing Sets AI Data Standards

The Linux Foundation's OpenSharing project establishes the first vendor-neutral open source AI standard for sharing agent skills, models, and datasets across enterprise platforms.

OpenSharing, contributed by Databricks, is the first open protocol for exchanging AI assets — agent skills, models, and unstructured data — across platforms and organizations.
The project evolves Delta Sharing, adopted by thousands of enterprises including SAP, Stripe, LSEG, Atlassian, and OpenAI since its 2021 launch.
Storage integrations from Everpure, MinIO, and Qumulo are live at launch; HPE, NetApp, Nutanix, and others follow.

Lead

The Linux Foundation on June 10, 2026, announced the OpenSharing project, a vendor-neutral open protocol for the secure exchange of AI assets and data across organizations and platforms. Contributed by Databricks, the initiative evolves the company's existing Delta Sharing standard — used by thousands of enterprises since 2021 — into a unified framework designed for the agentic era, covering agent skills, AI models, and large-scale unstructured data volumes.

What Happened

The OpenSharing project addresses a structural gap in enterprise AI: the absence of a common standard for moving AI assets between organizations and across cloud environments. Companies relying on AI models and agent capabilities sourced from external providers have had to maintain custom point-to-point integrations or depend on proprietary marketplace platforms, both of which impose lock-in risk and operational overhead.

OpenSharing resolves this by extending the technical foundation of Delta Sharing — whose user base already includes Amadeus, Atlassian, LSEG, SAP, Stripe, TheTradeDesk, and OpenAI — to encompass not just structured datasets but AI-native assets. The protocol introduces standard APIs for asset discovery, authorization, and access, enabling any organization to publish or consume AI resources without being bound to a specific cloud or vendor stack.

Technical Architecture

The protocol supports interoperability across major open table formats, adding compatibility with Apache Iceberg IRC clients and broadening the universe of potential data recipients. It also enables on-premises and private-cloud deployments to exchange assets with public cloud platforms without requiring data movement, a feature with direct relevance for regulated industries managing data residency and sovereignty requirements.

Storage service integrations with Everpure, MinIO, and Qumulo are live at launch. A second wave of partners — including Cohesity, Commvault, HPE, NetApp, Nutanix, Rubrik, and VAST Data — is set to follow in the coming months.

Strategic Context

The announcement arrives as enterprises accelerate agentic AI deployments, pushing demand for interoperability infrastructure higher. Databricks, which serves more than 20,000 organizations globally including 70 percent of the Fortune 500, positions OpenSharing as the continuation of its open-ecosystem strategy. By donating the protocol to the Linux Foundation AI governance umbrella, the company removes itself as sole steward and invites broader industry participation — a move that mirrors the model behind earlier open-source data standards such as Apache Arrow and Apache Parquet.

The Linux Foundation has become the neutral home for industry-level AI standards, housing the LF AI & Data Foundation and the Agentic AI Foundation announced earlier in 2026. OpenSharing extends that mandate into asset exchange, an area that touches both the infrastructure and commercial layers of the AI supply chain.

AI and Technology Angle

Open source AI interoperability is emerging as a structural requirement for enterprise adoption. Organizations that depend on third-party models or proprietary agent platforms face compounding switching costs as AI stacks grow more complex. A standardized exchange layer reduces those costs and lowers the barrier to participation in shared AI ecosystems, whether for data monetization, model distribution, or collaborative development. LSEG framed OpenSharing's value in terms of reach: enabling customers to access financial datasets using any tool or any cloud. OpenAI described it as a standard way to discover and authorize access to AI assets, language that signals broad perceived utility across the industry well beyond Databricks' own platform.

Outlook

OpenSharing is available on GitHub and at opensharing.io. The open governance structure under the Linux Foundation gives technology vendors, enterprises, and cloud providers a shared forum for contributing to AI data standards without ceding influence to a single commercial actor. Wider adoption will hinge on whether cloud hyperscalers and AI platform vendors not yet listed as launch partners move to support the protocol natively — a signal the market will watch closely through the remainder of 2026.

Technology