Databricks customers on Google Cloud running in Classic compute environments can already leverage Google’s C4A VMs powered by its custom Arm-based Axion processor to power their data warehousing, AI, and ETL workloads. This generally available capability delivers meaningful performance and efficiency improvements for organisations running modern, data-intensive applications today.
With the combination of Axion, Titanium SSDs, and Hyperdisk balanced storage, customers can achieve ever stronger efficiency gains compared to the previous generation of general-purpose VMs, all while benefiting from the openness, governance, and scalability of the Databricks lakehouse architecture.
Proven Performance for the Data + AI Era
C4A VMs are purpose-built for cloud workloads, delivering up to 65% better price-performance and 60% better energy efficiency than comparable x86-based instances. In Databricks workloads, these gains translate into:
- Faster analytics: Reduced query latency and improved concurrency for the Databricks SQL data warehouse
- Accelerated AI/ML: Shorter training and inference cycles for large-scale models
- Optimized ETL: Improved shuffle performance and reduced processing times for complex pipelines
Customers can adopt C4A instances without changing their workflows or rewriting code.
Part of an Expanding Databricks + Google Cloud Partnership
Support for C4A VMs builds on a series of recent innovations designed to help customers do more with their data on Google Cloud, including:
Together, these capabilities give customers a unified foundation for AI and analytics with flexibility, performance, and enterprise-grade governance.
A Shared Vision for Efficiency and Scale
Integrating Google’s Axion processors into the Databricks Data Intelligence Platform is a significant step in helping organizations accelerate their data and AI initiatives. The combination of Axion’s performance and efficiency with Databricks’ unified architecture provides the ideal foundation for customers to innovate, scale, and unlock new value from their data. — Salil Suri, Sr. Director of Product Management, Google Cloud
Epsilon’s goal for achieving unified consumer engagement is built upon a foundation of high-volume data processing. Leveraging Google Cloud’s Axion-based C4A virtual machines for our Databricks workloads provides a significant TCO benefit, achieving a 20-25% reduction in runtime and a 10-15% cost efficiency over previous-generation VMs for core machine learning pipelines. We plan to scale our Databricks implementation in GCP, utilizing Axion C4As as its infrastructural foundation. — Gairik Chakraborty, Senior Vice President, Database, Epsilon
As enterprises scale their data and AI workloads on Google Cloud, performance and efficiency become key to unlocking faster insights. By running Databricks on Axion-based C4A VMs with Titanium SSDs, customers can achieve ever-stronger gains in both price-performance and efficiency, while supporting sustainability goals. This collaboration with Google Cloud underscores our shared commitment to helping customers innovate faster and operate more sustainably. — Abhishek Rai, Sr. Director of Engineering, Databricks
Generally Available and Ready for Your Workloads
C4A VMs with Axion and Titanium SSDs are available today in multiple Google Cloud regions for Classic Databricks environments. The energy-efficient Arm-based architecture supports both performance goals and sustainability commitments, making them an attractive choice for enterprises scaling in the AI era.
If you’re preparing for enterprise agreements (EA) or major platform expansion, integrating C4A VMs into your Databricks deployment can deliver immediate performance gains while reducing long-term infrastructure costs.
To start running Databricks on Google Cloud C4A VMs, contact your Databricks account team or get Started with a Free 14-day Trial of Databricks on Google Cloud. Connect with your Databricks or Google Cloud representative to learn how you can bring unified data and AI to your organization today.
Get Started with a Free 14-day Trial of Databricks on Google Cloud.