Tech news in 3 minutes
Architect Labs Raises $24M Seed to Democratize Custom Chip Design
On June 10, 2026, Databricks announced the Databricks Software-Defined Storage (SDS) Ecosystem, a new partner category designed to bring its intelligence platform to enterprise data located on-premises, in private clouds, or at the edge. This launch addresses a growing trend: many enterprises—including semiconductor manufacturers, global trading firms, tier-1 banks, and major pharmaceutical companies—cannot or will not move all their data to the cloud due to data sovereignty regulations (e.g., GDPR, HIPAA), high egress costs at petabyte scale, latency requirements for edge workloads, and the untapped AI value in massive backup and secondary datasets. The ecosystem uses OpenSharing, an open-source protocol, enabling storage partners to expose their data directly to Databricks Serverless Compute via a secure endpoint integrated with Unity Catalog. This allows customers to query and run AI workloads on data that never leaves the premises, eliminating data migration, duplication, and compliance risk. Key partners include MinIO (GA), offering native OpenSharing integration for live on-premises Apache Iceberg and Delta tables; Everpure (Private Preview), providing a connector for on-prem object storage; and Qumulo (Private Preview in July 2026), integrating NeuralSearch with OpenSharing. Databricks emphasizes that over 2 zettabytes of enterprise data are managed by these storage partners, and the launch is just the first step toward making all hybrid data instantly ready for generative AI without copying a byte.