Databricks and the Linux Foundation have launched OpenSharing, an open standard for sharing data and AI assets across platforms and organisations. The project expands Databricks' Delta Sharing protocol into a broader framework for AI-related assets.
The standard is designed to let organisations share not only structured data, but also AI models, agent skills and unstructured data through a common protocol. It is now hosted by the Linux Foundation, which Databricks said positions it as a vendor-neutral effort rather than a tool tied to a single platform.
OpenSharing launches as companies face a more fragmented market for data and AI exchange. Many businesses use a mix of cloud providers, on-premises systems and software tools, while AI deployments have introduced new asset types that often sit behind platform-specific interfaces or custom integrations.
It builds on Delta Sharing, the open data-sharing protocol Databricks introduced in 2021 through the Delta Lake open source project. Databricks said Delta Sharing is used by thousands of customers and partners for data collaboration, sharing and monetisation.
Broader scope
The main change is that OpenSharing covers more than datasets. It is described as the first open protocol designed to support the exchange of agent skills and AI models between organisations, alongside traditional data-sharing functions.
That could matter for businesses that want to distribute AI components to customers or partners without sending files directly or building a separate integration for each recipient. Databricks said the protocol includes standard application programming interfaces for discovery, authorisation and access.
The standard also adds support for Apache Iceberg clients, widening the range of systems that can receive shared assets. Delta Sharing is already used across tools including Databricks, Apache Spark, Oracle, Power BI, Tableau and Snowflake, and OpenSharing is intended to expand that reach further.
Another focus is on on-premises data. Organisations that keep information in private environments will be able to connect those assets to cloud-based AI and analytics systems without moving the underlying data, according to Databricks.
Managed OpenSharing services are already available from Everpure, MinIO and Qumulo, Databricks said. The company also named Cohesity, Commvault, HPE, NetApp, Nutanix, Rubrik and VAST Data as partners.
Matei Zaharia, Co-Founder and Chief Technology Officer of Databricks, outlined the company's case for the move.
"Delta Sharing proved the industry would choose open over locked-in," said Zaharia. "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."
Industry backing
Several technology and data companies backed the launch, saying they were using or supporting the standard in their own products and services.
OpenAI framed its support around common rules for access to AI assets. "We believe in open AI ecosystems and are excited to collaborate with Databricks on providing a standard, secure way to discover and authorise access to AI assets," said Alexander Embiricos, Head of Enterprise Product, OpenAI.
Amadeus linked the development to the travel industry's efforts to introduce AI into operational systems. "AI has the potential to transform how the travel industry operates and serves travellers, but only if it is trusted and applied in a meaningful way," said Vvivi Hu, Chief Strategy Officer, Amadeus. "At Amadeus, we act as a system of record and the embedded execution layer for the industry, helping orchestrate AI in an integrated and scalable way across the travel ecosystem. We're excited to be working with Databricks to help enable the open and secure exchange of data and AI assets needed to deliver this value."
Atlassian said it had already launched data shares using the protocol. "Atlassian Analytics launched data shares, leveraging OpenSharing from Databricks, to unlock access to critical cloud data for our customers at scale," said Ben Jackson, Senior Group Product Manager, Data & Analytics, Atlassian. "It provides flexibility and accelerates customers' time to insight. OpenSharing's ecosystem of connectors enables customers to easily power their environments with data directly from the Atlassian Data Lake."
LSEG also tied the standard to customer demand for portability. "Our customers rely on our AI-ready data and insights to deliver critical value to their clients," said Ron Lefferts, Divisional Chief Executive Officer of Data & Analytics, LSEG. "We chose OpenSharing because it's consistent with our LSEG Everywhere strategy. It allows any customer to use our data in any tool or any cloud with any model."
For storage provider MinIO, the emphasis was on access to data that stays in place. "Enterprises should not have to choose between keeping sensitive data on-premises and using modern AI and analytics platforms to extract value from it," said AB Periasamy, Co-Founder and Co-Chief Executive Officer, MinIO. "Native open source OpenSharing in AIStor opens up access to data that cannot move. This creates a foundation for unlocking the vast untapped AI value hidden across enterprise data environments."
SAP, Stripe and Acxiom also endorsed the approach as a way to distribute data and AI assets across different customer environments.
"We chose OpenSharing for SAP Business Data Cloud because it is the open protocol for sharing data and AI assets, and allows us to reach customers wherever they are," said Senthil Krishnapillai, Vice President and Head of Cloud Services, SAP. "We're excited to continue partnering with Databricks to lay the open foundation for AI-first collaboration."
"Financial data is the lifeblood of our customers' operations, and they require the flexibility to analyse it in their preferred environments," said Emily Sands, Head of Data and AI, Stripe. "Our partnership with Databricks is built on a shared vision of openness. 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."
"With Acxiom's Real ID available natively via the OpenSharing protocol, enterprise marketers can connect data directly where it already lives," said Greg Morton, Head of Ecosystem Growth & Industry Partner, Acxiom. "They can make their data interoperable across identifiers, platforms, partners, and clouds, unlocking insights from otherwise inaccessible partner data sources for collaborative and truly connected audience engagement, analytics, and measurement."