Tech and AI

Databricks closes a 5 billion dollar round at a 190 billion dollar valuation after asking for one fifth of it

By Staff Writer  |  14 August 2026

The company set out to raise 1 billion dollars and found 15 billion dollars of interest waiting. Its chief executive says finance directors alarmed by AI running costs are now driving the demand.

Databricks closed a 5 billion dollar funding round on 13 August at a valuation of 190 billion dollars. Coatue led, with Blackstone, MGX, accounts advised by T. Rowe Price and, as a new name on the register, Sixth Street Growth. Roughly two dozen investors were named as participants.

The company had intended to raise a fifth of that. Ali Ghodsi, its co-founder and chief executive, says a press report during the company's own conference in June turned a quiet plan into a queue. "As soon as that article went out, there was a long line of investors that started calling. My phone blew up," he said, adding that among the select group it looked at there was 15 billion dollars of interest. Turning long-standing backers away at that point tends to sour relationships, so the company issued more stock instead. It announced a close at 188 billion dollars in July without saying how much it had raised; the figures published this week put the final number at 190 billion dollars.

Six months ago the same company raised 5 billion dollars and took on 2 billion dollars of new debt capacity at a valuation of 134 billion dollars. It has raised about 20 billion dollars in 20 months.

What the numbers behind it look like

Annualised revenue run rate has passed 7 billion dollars and is growing at more than 80 per cent year on year, and the business is cash-flow positive. The cloud data warehouse at the centre of the product accounts for 1.5 billion dollars of that and is still doubling. Lakebase, a database aimed at software written by AI agents and launched in June 2025, has reached a 100 million dollar run rate.

Ghodsi's account of why the money is needed is unglamorous and worth repeating: multibillion dollar cloud commitments with all three of the large hosting providers, an AI research team of a hundred people in the most competitive hiring market there is, and acquisitions. The company bought an AI security business in June, two smaller firms in March, and this week added the maker of a lightweight Postgres database that lets agents spin up their own storage.

The cost argument that is selling the product

The more interesting part of the day was Ghodsi's reading of his own market. Buyers who a year ago insisted on frontier proprietary models are now looking hard at what those models cost to run at volume, and are reconsidering open and Chinese alternatives they previously ruled out. The company sells a gateway that routes work between models and controls the spend, which is a product that only exists because the bills became frightening.

The attitude a year or two ago was we just need frontier proprietary, and we can just ignore Chinese models. What has happened is that this token maxing has freaked out the CFOs.

Ali Ghodsi, Co-founder and Chief Executive, Databricks

On going public, Ghodsi says the intention remains and the timing does not. He wants to spend on AI products rather than explain quarterly volatility to a public register, and with that many investors already on the books he is unlikely to be able to say never.

For anyone buying AI tooling under a long contract, the sentence about finance directors is the one to keep. A supplier whose fastest growing product is a meter has just told the market that its customers were surprised by the bill.