Databricks sells a cloud platform where companies store, govern, and analyze their data, then build machine learning models and AI agents on top of it. Based in San Francisco, it grew out of UC Berkeley's AMPLab and the Apache Spark project and later coined the 'lakehouse' idea of running warehouse-style analytics directly on low-cost cloud object storage. Customers such as adidas, AT&T, Bayer, Block, Mastercard, Rivian, and Unilever use it for workloads ranging from ETL pipelines and BI dashboards to fraud models and generative AI applications. Its main rival is Snowflake, with the cloud providers' own data stacks as the other big competitors. The company is headquartered in San Francisco, California.
Databricks makes money by charging for compute consumed on its platform, measured in Databricks Units (DBUs). Customers usually sign annual or multi-year committed-spend contracts and then draw them down as their engineers run pipelines, SQL queries, model training, model serving, and AI agents. There is no flat seat fee for most workloads, so revenue rises as data volumes and AI usage grow. Databricks runs on AWS, Microsoft Azure (where it is sold as the first-party Azure Databricks service), and Google Cloud, and customers can often pay for it through existing cloud commitments. Newer products such as Lakebase (serverless Postgres), Genie, and Agent Bricks extend the same consumption model to operational databases and AI agents.