Databricks

Databricks is a platform that helps companies organize and analyze the data they've accumulated. It was founded in 2013 in Berkeley by the researchers who created Apache Spark.

In the AI era, what the company does is "let you safely run other people's good models on top of your own company's data." It brings together models from OpenAI, Anthropic, Google, and Meta in one place, and lets you connect them to your internal company data. The models belong to other companies; the layer that manages the data and controls access belongs to Databricks.

It used to build its own models too — Dolly and DBRX — but it has since stepped back from that path and now stands firmly on the side of being a unified platform. Saying "Databricks builds models too" would now be past tense.

In terms of scale, it's among the very top private companies. In August 2026, it confirmed a valuation of $190 billion and disclosed annual revenue of around $7 billion. That same month, it officially launched an "AI Gateway" that routes calls across multiple models through a single gateway to cut down on wasted tokens. Its CEO also sparked debate by saying "AGI is already here."

"Databricks supports Claude" — this doesn't mean Databricks built Claude, but that it made Claude usable on top of enterprise data.