Databricks and OpenAI have formalized a multi-year commercial agreement valued at $100 million. The goal is to make OpenAI’s models available directly within Databricks’ Data Intelligence platform and in the Agent Bricks product, for its roughly 20,000 customers.
The deal requires Databricks to pay this minimum amount to OpenAI, even if the actual usage of the models by customers falls short of expectations. This amount represents a commitment to purchase or consume services, such as model usage or API calls, rather than direct financing without consideration. Detailed terms on pricing, consumption, or any penalties were not disclosed.
The integration is designed to allow Databricks customers to access OpenAI models without moving their data off the platform, whether via SQL or API. The objective is to lower technical and contractual barriers for clients and thus foster broader adoption. GPT-5 is highlighted as the leading model.
A $100 Million Commercial Deal
Agent Bricks, Databricks’ AI agent-creation solution, will be used to develop, test, and deploy applications that rely on client data, combining OpenAI models with Databricks’ internal tools.
Unity Catalog, Databricks’ metadata and governance catalog, will ensure the integration of security, governance, and observability features. Finally, technical teams from both companies will collaborate to optimize AI models in contexts tailored to the needs of professional clients.
The exact duration of the agreement has not been made public, and no details are provided on the pricing structure or any potential revenue-sharing mechanisms.
For Databricks, the collaboration strengthens its positioning in AI integrated into its solutions, facing competitors such as Snowflake. OpenAI was already using its platform for certain data-processing tasks, notably to enhance ChatGPT.
