Databricks just raised $7 billion in equity and debt at a $134 billion valuation. If your eyes glazed over, I don’t blame you. This is the company’s umpteenth mega-round, and at some point, another private company raising another pile of money stops being news.
But here’s why this one matters. As Databricks inches closer to an IPO, it’s starting to act more like a public company. And the financials it’s releasing tells us something more interesting than the size of the check.
Of the databases on Databricks’ platform, 80% are now being built by AI agents, not people. And these aren’t all tech companies. Databricks has over 20,000 customers.
It’s the clearest evidence yet of something we’ve been debating since 2026 started: AI agents aren’t just writing code, they’re building real software inside the world’s biggest companies. And that has major implications for the trade.
We sat down with Databricks CEO Ali Ghodsi. He sits at the intersection of all of it — the models, the data, the infrastructure and the 20,000 companies actually trying to put it all to work. He sees which AI models are winning, which ones enterprises are using and how fast agents are improving, and he has a front row seat to the question the entire software industry is grappling with right now: what happens when AI can build it for you?
