Databricks closes $5B at $190B, and the ROI it is priced on is still someone else's problem

Databricks has finalized a $5 billion financing round at a $190 billion valuation, closing above the $188 billion figure it disclosed in July when it signed a term sheet led by existing backer Coatue ^[1][2]. At roughly 27 times a revenue run rate that Forbes puts above $7 billion, the round prices in a level of enterprise AI platform spend that the enterprises themselves have not yet shown produces returns ^[3][1].

The increment from term sheet to close is small in relative terms: $2 billion, or about 1.1% ^[2][2]. Ghodsi told Forbes the difference reflects a larger raise and extra shares issued during the process ^[4]. Coatue led again, joined by Blackstone, MGX and T. Rowe Price, with Sixth Street Growth investing for the first time ^[5]. The $5 billion cheque buys roughly 2.6% of the company at the post-money mark ^[3].

The ladder matters more than the round. Databricks was worth $62 billion in late 2024, passed $100 billion by August 2025 and reached $134 billion in December 2025 ^[6][7][8]. That makes $190 billion a little over three times the late-2024 price and about 42% above the December mark ^[4][5]. The source describes the last step as an almost 40% jump in about five months, though it dates the term sheet only to July 16 without a year, so the interval is not checkable from the material ^[9]. Forbes puts year-over-year growth at a maximum of 80% ^[10].

The capital is earmarked for three products ^[11]. Unity AI Gateway routes AI workloads across models and lets customers set spending budgets ^[12]. Lakebase, a serverless Postgres database aimed at software written by AI agents, is past a $100 million run rate according to Forbes, which is about 1.4% of company revenue ^[13][6]. Genie is meant to give models the unstructured internal context, including emails and meeting recordings, that they need to act inside a business ^[14]. Ghodsi frames the shift as customers moving from "tokenmaxxing to valuemaxxing," and says more than one quadrillion tokens have passed through the gateway, which gives Databricks a read on how customers switch models when a cheaper or stronger one appears ^[15][16].

The bear case is in the same coverage. Owen Lau of Clear Street told Yahoo Finance that ARR growth above 50% with gross margin stabilizing above 70% should justify the valuation, then added that the ROI debate at the application layer is unsettled and that enterprises unable to monetize AI tools will likely cut these investments ^[17][18]. Three years of 50% growth from $7 billion would put revenue near $23.6 billion and, at an unchanged price, the multiple near 8 times ^[7]. That is the arithmetic the round requires, and none of it is inside Databricks' control; it depends on 20,000-odd customer organizations, including 70% of the Fortune 500, finding measurable output ^[19].

Ghodsi also argues that AGI, under the definition the industry used before 2022, has already arrived, while what people now mean by the term is closer to superintelligence and is not here ^[20][21]. His explanation for why so little inside companies looks autonomous is that a model cannot reason about a business without its records, rules and permissions ^[22]. His summary is the most honest line in the round: "The world remains largely unchanged, except that token spending is rising" ^[23]. Investors just paid $190 billion for the second half of that sentence.