Microsoft and Databricks Expand Azure Partnership Into the 2030s

Databricks and Microsoft announced on July 23 that their strategic partnership will extend into the 2030s. The new agreement deepens Databricks' use of Microsoft infrastructure and expands product integration across the two companies' enterprise data and AI stacks.

Databricks said it will use Azure Databricks for its own core business operations and analytics, including its unified lakehouse. It also plans to increase its use of Microsoft's Arm-based Azure Cobalt processors for data-intensive and agentic AI workloads. Reuters described the agreement as a significant Azure customer commitment at a time when cloud providers are competing for enterprise AI workloads.

The announcement is distinct from the companies' June 2025 partnership extension. This July 2026 agreement adds the longer time horizon, Databricks' internal operating commitment, and more specific Cobalt and product-integration plans.

Databricks currently uses Cobalt 100 and says it plans to adopt Cobalt 200. Microsoft's published figure of up to 50% better performance for Cobalt 200 is a vendor claim, not an independently reproduced result for Databricks workloads. The announcement does not provide migration volume, benchmark methodology, spending commitments, or financial terms.

On the software side, the companies plan deeper integration of Databricks Genie and Unity AI Gateway across Microsoft products. The official release names Microsoft Entra, Azure Data Lake Storage, OneLake, Power BI, Purview, Foundry, Power Platform, Microsoft 365, Teams, and Copilot. The stated goal is to let customers connect governed enterprise data and business context to agents and analytics inside existing Microsoft workflows.

For practitioners, the useful signal is that Databricks is becoming a larger internal consumer of the Azure service it sells jointly with Microsoft. That can strengthen operational alignment around identity, governance, compute, and product roadmaps, but the announcement does not establish customer-level performance or cost savings.

Teams evaluating the stack should still test workload-specific Cobalt compatibility, price-performance, data residency, catalog interoperability, observability, and portability. The partnership's duration may improve roadmap continuity, while the deeper Microsoft integrations also make it more important to identify which components rely on provider-specific interfaces. Customer benchmarks and disclosed commercial terms would be needed to quantify the practical impact.