Databricks closed a $5 billion round at a $190 billion valuation — $2 billion above the number attached to the same round when it opened in July

[Databricks](#/company/databricks) closed a $5 billion funding round on Aug. 13 at a $190 billion valuation — up from the $134 billion the company was valued at in February, and roughly $2 billion above the $188 billion figure attached to the same round when it was first announced in mid-July. Coatue led the round; Blackstone, MGX, and accounts advised by T. Rowe Price joined alongside returning backers Andreessen Horowitz, Thrive Capital, Goldman Sachs Alternatives, and Temasek, plus new investor Sixth Street Growth.

It's also Databricks' second $5 billion raise in six months, and by outside accounts an oversubscribed one: a venture source told TechCrunch when the round first opened in July that "the deal is solid, with so many firms wanting in." Sixth Street Growth joining as a first-time investor in this round, alongside five returning backers from February, is consistent with that framing — nobody with a prior stake sat this one out, and at least one new large fund bought in at the higher price.

Four valuations in twenty months, and what each one actually was

Databricks' valuation has now moved four times since December 2024: $62 billion on a $10 billion raise that month, $100 billion on a $1 billion raise the following September, $134 billion on a Series L round closed in February 2026 (split between $3 billion in equity and $2 billion in debt), and now $190 billion. That's roughly a threefold increase in the valuation in under two years, on a company that has never gone public and reports no earnings a public investor can check independently.

The move from $188 billion to $190 billion isn't two competing numbers — it's the same transaction at two different moments: TechCrunch reported the $188 billion figure in mid-July when Coatue led an opening round that hadn't yet closed, and Databricks' Aug. 13 announcement confirms the final $190 billion print once the rest of the syndicate joined. The number that matters for comparing against the February round is $190 billion — the one investors actually paid.

One more wrinkle worth flagging: some coverage of the Aug. 13 close, including material drawing on Databricks' own framing, describes the new valuation as "up from $134 billion in December 2024." That's off by more than a year — the $134 billion figure was set in December 2025 and closed in February 2026, not December 2024, which is when Databricks was valued at $62 billion on an entirely different raise. CNBC's own Feb. 9, 2026 reporting and Tech Funding News's roundup both agree on the later date; this piece uses that timeline rather than the conflated one.

$7 billion run-rate, and what's driving it

Databricks says quarterly revenue run-rate crossed $7 billion in the most recent quarter, up more than 80% year-over-year, Databricks says. The company points to three specific products: Lakebase, a serverless Postgres database built for AI agents, now past a $100 million run-rate; Genie, an AI assistant that answers from a company's own business data; and Unity AI Gateway, which manages and controls costs across multiple AI models. More than 1,000 customers now spend over $1 million a year on its platform, and more than 100 spend over $10 million, Databricks says.

That multiple moves with either input — a higher run-rate at the next disclosure would make today's valuation look more ordinary in hindsight; a stalled one would make it look increasingly like a bet on future growth rather than current revenue. Databricks hasn't said whether it expects run-rate to keep compounding at above 80% annually, only that it has for the trailing year.

Databricks isn't the only company repricing itself upward this fast this summer. Coding-agent startup Cognition has reportedly been in talks to raise at a $40 billion valuation, up from $26 billion three months earlier, and OpenAI-backed roll-up Thrive Holdings raised over $2 billion at a $12 billion valuation earlier in August. None of those numbers moves in lockstep with Databricks', but the shape is the same across all three: private valuations resetting on a timescale of months rather than years, ahead of any public filing that would let an outside investor check the growth story against audited numbers. In each case, the number that gets the headline is a private mark agreed between a company and a small group of investors, not a price discovered by a broad public market.

CEO Ali Ghodsi has been explicit about why none of these valuation jumps has come with an IPO. In June, he told Bloomberg Television that Databricks is IPO-ready — governance, financial reporting, and compliance frameworks already built out — but called 2026 "a terrible year to go public," pointing to a possible listing as early as 2027. That timeline means at least three of Databricks' four valuation jumps, including this one, will have happened entirely inside private markets, priced by a rotating group of institutional investors rather than tested against a public order book.

Why the bull case isn't the only one that adds up