Databricks Puts Its Own Operations on Azure in Expanded Microsoft Deal

Databricks’ Genie One interface, showing options to draft documents, create skills, schedule tasks and analyze data. Source: Databricks.

The product pitch is governed AI in Microsoft workflows

The expansion is also a distribution and integration deal. Databricks says its Genie conversational-analytics product and Unity AI Gateway will be more deeply integrated with Microsoft Entra, Azure Data Lake Storage, OneLake, Power BI, Purview, Foundry, Power Platform, Microsoft 365, Teams and Copilot.

The company describes Genie as an AI co-worker that grounds agents in enterprise data, and Unity AI Gateway as a layer to govern models, agents and their cost. The stated premise is that customers can bring governed, real-time data and AI into an existing Microsoft environment rather than assemble separate identity, data and workplace connections.

That is an attractive proposition for organizations that already use Microsoft's stack, but it remains a product claim. Databricks and Microsoft say enterprises often struggle to connect AI to trusted business knowledge, govern models and agents consistently, and control costs. Neither disclosed adoption results, implementation times or like-for-like operating costs for the newly expanded integrations, so the announcement does not show whether the combined approach outperforms other ways of deploying enterprise AI.

Cobalt adds an infrastructure layer

Databricks also says it will expand use of Azure Cobalt, Microsoft’s Arm-based infrastructure, for data-intensive and agentic-AI workloads. It currently uses Cobalt 100 and plans to adopt Cobalt 200.

Databricks cites Microsoft’s claim that Cobalt 200 can deliver up to 50% better performance than Cobalt 100 and has memory encryption enabled by default. “Up to” is a vendor performance ceiling, not a Databricks benchmark: the companies did not disclose the workloads tested, instance counts, timing of the move or measured savings. The planned chip adoption therefore makes the partnership more concrete at the compute layer without quantifying its economic impact.

What would turn the promise into evidence?

The next evidence to watch is operational rather than rhetorical: the share and type of Databricks workloads that run on Azure, the timing and scale of the Cobalt migration, and customer deployments that report implementation time, performance and cost on comparable workloads. Until then, the clearest confirmed change is a deeper strategic and operational tie between a major data-and-AI platform provider and the cloud service with which it has partnered for years—not a demonstrated new advantage for every enterprise customer.