Databricks highlights a new partnership with Nvidia that enables users to accelerate AI workloads through the Data Intelligence Platform.

No AI without data and no AI without Nvidia: At its Data + AI Summit, Databricks unsurprisingly announced a closer collaboration with Nvidia. Just like its major competitor Snowflake a week earlier, Databricks wants to make it clear to users that the necessary hardware for AI workloads is available directly on the platform.

Specifically, Databricks brings Nvidia CUDA acceleration into play Data intelligence platformThis will make numerous AI-related workloads on corporate data more efficient. Databricks itself will incorporate Nvidia acceleration for its Photon query engine. For the customer, the collaboration will bring more performance for the same price.

NIM

Databricks has previously introduced (just like Snowflake) its own LLM: DBRX. In analogy to the competitor’s LLM, DBRX is one Mix of experts (MoE) model. It is available as an Nvidia NIM microservice.

Jensen Huang, CEO of Nvidia, believes the partnership is worthwhile: “By bringing Nvidia CUDA acceleration to the Databricks stack, we are laying the foundation for customers everywhere to leverage their data to drive enterprise-level generative AI.” Last week, Huang was also present (virtually) at the Moscone Center in San Francisco to tell a similar story about the collaboration with Snowflake.