Databricks and Microsoft have extended their partnership into the next decade, deepening integration across infrastructure, data and AI. The expanded deal sees Databricks run more of its own core operations on Azure Databricks, adopt next-generation Cobalt 200 chips, and push tighter native ties between Databricks Genie and the Microsoft 365 stack.
A Decade-Long Bet Gets Deeper
Databricks is growing its own use of Azure Databricks to run core business operations and analytics, while both companies advance native integration across the Microsoft stack, including Databricks Genie and Microsoft 365 tools already familiar to enterprise IT teams.
Judson Althoff, CEO of Microsoft Commercial Business, frames the move around organisational knowledge, arguing that Databricks running its own workloads on the platform signals confidence to customers evaluating the same infrastructure at enterprise scale.
"The next generation of AI will be defined by how effectively organisations turn their unique knowledge into intelligence."— JUDSON ALTHOFF, CEO OF MICROSOFT COMMERCIAL BUSINESS
Cobalt 200: Faster Chips, Built-In Encryption
Databricks currently runs on Azure Cobalt 100 chips and plans to move to Cobalt 200, which Microsoft says could deliver up to 50% better performance than the prior generation.
Cobalt 200 also ships with memory encryption enabled by default, a detail that matters to security and compliance teams assessing cloud infrastructure at scale. Neither company treats this as optional given the size of the workloads involved.
Democratising Data Access
Rich Radley, Vice President of Field Engineering for EMEA at Databricks, said democratising data isn't just about making information easier to access, but about giving more people the ability to turn ideas into action, rather than relying on technical teams to answer questions or build reports.
He argues Genie One addresses that gap directly by putting AI into the hands of more employees, enabling them to explore information, automate routine work and make better-informed decisions without needing deep technical expertise.
