News
Commvault Taps Databricks and NVIDIA to Put Backup Data to Work for AI
Commvault is expanding the role of protected enterprise data in AI, announcing work with Databricks and NVIDIA to make that information usable while retaining governance and business context.
The two collaborations, announced Nov. 10 at SHIFT 2026, build on Data Activate. Commvault's capability overview describes a workspace for finding, preparing and exporting data from backups, with classification, redaction and policy enforcement built into the process.
The platform supports structured, semi-structured and unstructured sources. Its preparation controls include sensitivity tagging and approval before export, with audit trails recording actions. Commvault lists Parquet and Iceberg among the open formats available for curated datasets.
Governed Tables for Databricks
The planned Data Activate for Databricks solution will prepare enterprise data managed by Commvault and share it through OpenSharing as governed tables. Customers will manage those tables with Unity Catalog for analytics, AI applications and agents.
Commvault says the approach will let customers select protected data without building a separate connection to every original source. Source information, ownership, access permissions and lineage will accompany the data, while the solution will reflect new, changed and deleted content over time.
"Customers tell us their data is most valuable in Databricks when that data's context travels with it," said Stephen Orban, senior vice president of product partnerships and ecosystem at Databricks, in the release.
Early access is planned for December 2026 in select Azure regions. Commvault also provides protection for the Databricks Data + AI Platform and is a certified Databricks partner.
NVIDIA Retrieval Technologies
The separate NVIDIA collaboration combines Data Activate with NVIDIA NIM microservices and NemotronRetriever. The announcement identifies NVIDIA NIM Nemotron Retriever and NVIDIA cuVS with Apache Solr as technologies for extracting and organizing protected content for retrieval by applications and agents.
Data Activate will update the information available to AI as source content changes or is deleted, retention periods expire, or permissions change. Ownership, classification and lineage will help preserve a record of where data originated and how it can be used.
"Enterprise agents need current, trusted data to turn business knowledge into action," said Pat Lee, vice president of strategic partnerships at NVIDIA, in the announcement.
The NVIDIA reference design for SaaS and software is targeted for early release in December 2026 and general availability in 2027. The Databricks announcement specifies early access but does not give a general availability date.
About the Author
David Ramel is an editor and writer at Converge 360.