Reference · Chapter 41

Databricks by the Numbers

Every figure on this page is traced to the entity that produced it, dated, and linked. Most are Databricks' own press releases. The Snowflake figures come from its filings with the SEC, because Snowflake is public and Databricks is not, which is also why the two are measured differently and this page says so. Nothing here is copied from another roundup, and a number we couldn't trace is left off rather than guessed.

Key numbers #

• $190 billion valuation, on a $5 billion strategic round closed in August 2026.
• More than $7 billion annual revenue run-rate, growing over 80% year on year, as of August 2026.
• More than 20,000 customer organisations, including 70% of the Fortune 500.
• More than 1,000 customers each spending over $1 million a year, and more than 100 spending over $10 million.
• $1.4 billion revenue run-rate from AI products alone, as of February 2026.
• Valuation tripled in twenty months, from $62 billion in December 2024 to $190 billion in August 2026.
• Snowflake's revenue for the fiscal year to 31 January 2026 was $4.68 billion, up 29% on the year before.
• MosaicML was bought for $1.3 billion in June 2023, the largest disclosed acquisition price.
• Free Edition, the permanently free tier, launched on 11 June 2025 and replaced Community Edition.
• Founded in 2013 in San Francisco by the creators of Apache Spark.

What is Databricks worth? #

Databricks is private, so its valuation is whatever its last funding round said. The company announces each round itself, and the last five read as a straight line up.

The July term sheet said $188 billion and the round closed a month later at $190 billion, led by Coatue with Blackstone, MGX, T. Rowe Price and Sixth Street Growth. The Series K was announced as a term sheet on 19 August 2025 and described as closing in the 8 September 2025 release, co-led by Andreessen Horowitz, Insight Partners, MGX, Thrive Capital and WCM.

What is Databricks' revenue? #

Databricks doesn't publish revenue. It publishes an annual revenue run-rate, which is the most recent period's revenue scaled to a year, and it publishes it at every funding announcement.

Growth accelerated through 2025 and 2026 rather than slowing, which is unusual at this size and is the number investors are paying for.

Two product lines are broken out. AI products passed a $1 billion run-rate in September 2025 and reached $1.4 billion by February 2026. The data warehousing product passed $1 billion within four years of general availability, per the December 2025 release. The Series J release also said the company expected to be free cash flow positive in the quarter to 31 January 2025; that was a forecast at the time, and Databricks hasn't published cash flow since.

How many customers does Databricks have? #

Databricks' company page, read on 12 September 2026, says more than 20,000 organisations worldwide use the platform, including 70% of the Fortune 500, and names Block, Comcast, Condé Nast, Rivian and Shell. The Series K release of 19 August 2025 said more than 15,000, so the customer count grew by about a third in a year.

The more telling count is the number of large customers, which Databricks reports at each announcement.

The million-dollar cohort doubled in twenty months. Sources are the same five releases as the revenue table.

How does Databricks compare with Snowflake on revenue? #

Carefully, because the two companies publish different things. Snowflake is listed and reports audited revenue under GAAP. Databricks is private and reports a run-rate, which scales the latest period to a year and will always read higher than trailing revenue for a company that's growing.

Snowflake's fiscal 2025 revenue was $3.626 billion and fiscal 2024 was $2.806 billion, from the same filing. Its latest quarter annualises to about $6.2 billion, which is the fairest number to set beside Databricks' run-rate, and even then the two aren't the same measure. What can be said plainly is that Databricks is reporting faster growth from a larger base. Why the bill differs between the two platforms is its own chapter.

What has Databricks acquired? #

Databricks announces acquisitions but rarely prices them. Two carry enough detail to state.

Press reports put Neon at about $1 billion, but Databricks didn't say so and this page doesn't either. Lakebase is what came of it, and What Everything Used to Be Called tracks what happened to the MosaicML name.

When did Free Edition launch, and what did it replace? #

On 11 June 2025, at the Data + AI Summit, per the announcement. It replaced Community Edition with a permanently free workspace that includes Unity Catalog, serverless compute, notebooks, dashboards, Genie and the Academy's self-paced training. It isn't licensed for commercial use. Getting Started Free is the guided first hour on it.

What we read into the numbers #

The tables above are facts anyone can find. What follows is what we think they mean, and it's the part that changes.

The growth is coming from inside the customer base. Run-rate rose about 75% between September 2025 and August 2026 while the customer count rose about a third. New logos can't close that gap; existing customers consuming more can, and the million-dollar cohort growing from 650 to over 1,000 says the same. If you're a customer, part of that 80% is your invoice. The growth is your invoice is the longer argument, and Cost and Performance is what to do about it.

A quarter of the revenue is AI products. $1.4 billion of a $5.4 billion run-rate in February 2026. Two years earlier the AI line didn't exist as a reported number. That's the platform's centre of gravity moving from where data is kept to where it's used, which is also why the three products named in the August 2026 release, Lakebase, Genie and Unity AI Gateway, all sit close to an end user. Databricks bought a Postgres follows that thread.

The valuation cadence is an IPO cadence. Five rounds in twenty months, each at a higher price, is a company pricing itself for a listing. For a buyer the consequence is on margins: a seller whose investors expect margins to hold has less room to discount consumption than one buying share. Negotiate committed spend knowing that.

Snowflake at 29% is not standing still. It's growing faster than most software companies its size. The gap to Databricks is about the AI line and about consumption from existing accounts, and the comparison chapter argues that for most estates the choice between them is decided by the bill rather than by features.

New readings go in Field notes as the numbers change, dated.

Who wrote this, in numbers #

TechFabric is a Databricks partner with 80 Databricks-certified engineers inside a team of 115+, building on the platform since the company was founded in 2017. This guide is the manual those engineers work from. The commercial side is at techfabric.com/databricks.

How to read these numbers #

A run-rate is a forward-looking scaling of one period, and a private company chooses when to disclose it, so the Databricks series above is a set of high-water marks announced at fundraising moments rather than a quarterly record. Treat growth rates between them as indicative. Customer counts are rounded and self-reported. Valuations are what the last investor paid, which is a price, not a measurement.

Changelog #

• 12 September 2026. First published. Figures current to the 13 August 2026 Databricks release and Snowflake's 10-Q filed 4 September 2026.

Databricks has announced figures every three to six months; this page is checked after each announcement and after each Snowflake quarterly filing.

Table of Contents