Databricks Raises $5bn at $190bn and Keeps Dodging the IPO Question
A $56bn jump in six months says more about how scarce profitable AI infrastructure companies are than about Databricks itself.
Databricks has raised $5bn at a $190bn valuation, up from roughly $134bn six months ago. Coatue led the round, with Blackstone, MGX and accounts advised by T. Rowe Price among the participants and Sixth Street Growth joining as a new investor. The company says it passed a $7bn annualised revenue run-rate, grew more than 80 per cent year on year in the second quarter, and has been cash-flow positive on an adjusted basis over the past twelve months.
What the numbers say
At $190bn on a $7bn run-rate, Databricks is priced at about 27 times run-rate revenue. That is rich for an enterprise software company and cheap next to the frontier labs, which is roughly where it wants to sit. The pitch is that Databricks is the place where companies' data already lives, so it is the natural place to build and govern AI applications on top of that data.
The product detail supports the story more than the headline does. The Lakebase database has passed a $100m run-rate and the core Lakehouse warehousing business is above $1.5bn. The new money is earmarked for Lakebase, the Genie assistant and the Unity AI Gateway. In other words: a transactional database to compete with the operational stores customers already run, a natural-language front end, and a control plane for routing and governing model calls. Each of those moves Databricks closer to the application layer and further from being just the place analysts run Spark jobs.
Why it keeps raising privately
Databricks has been described as an IPO candidate for years, and this round does nothing to settle when. The private market has been willing to pay up repeatedly, and at this scale a private round looks a lot like a pre-IPO crossover round without the disclosure obligations. The window for large listings is open, and Databricks is choosing not to walk through it yet.
There are reasonable explanations. Staying private lets the company keep investing heavily without quarterly scrutiny of margins while it builds out database and AI products. It also avoids being benchmarked line by line against Snowflake, its closest listed rival, before the new products have matured. The less charitable reading is that a private mark at $190bn is easier to defend than a public one.
What to watch
- Lakebase adoption. If it keeps compounding, Databricks has a credible second engine beyond analytics. If it stalls, the valuation leans entirely on AI enthusiasm.
- Gateway lock-in. A governance layer that sits between every model call and every dataset is powerful, and engineers should weigh how portable their setup remains once policies live there.
- The listing. Positive adjusted cash flow removes the usual excuse. At some point employees and early investors will want liquidity that tender offers cannot fully supply.
For data teams, the practical consequence is that Databricks now has a very large war chest to bundle features you might otherwise buy separately. That is good for prices in the short term. It is worth thinking about the long term before every workload ends up behind one vendor's gateway.
Sources