SAN FRANCISCO, November 22, 2026 — Databricks is in early talks to raise several billion dollars at a $225 billion valuation, nearly 20 percent above the $190 billion price tag it commanded in August. The shift is clear: investors are betting the data lakehouse becomes the control plane for enterprise AI, capturing the compute workloads that would otherwise flow straight to hyperscalers.
Valuation Surge Reflects Bet on AI Infrastructure Layer
The Wall Street Journal reported the discussions on Friday, citing people familiar with the matter. Terms could still change. The August round, a $5 billion raise led by Coatue Management with participation from Andreessen Horowitz, NEA, Insight Partners, Fidelity, and J.P. Morgan Asset Management, valued the company at $190 billion. Blackstone and T. Rowe Price also joined that round, according to Yahoo Finance.
Databricks passed a $7 billion revenue run rate in August and is growing 70 to 80 percent annually. CEO Ali Ghodsi told the All-In podcast he is targeting growth “way over 100” percent. The company is staying private for now. Ghodsi said the volatility of public markets during the AI transition creates a distraction; when a stock falls, staff effectively take a pay cut and some leave before the price recovers. He acknowledged an IPO will happen eventually but prefers to arrange tender offers for employees annually while the company remains private.
The valuation ask stands in sharp contrast to its closest public rival. Snowflake trades at about $90 billion, 34 percent below its 52-week high. The Nasdaq Composite fell 5.28 percent on the day the Databricks talks surfaced, with the PHLX Semiconductor Index dropping 10.73 percent.
Unity Gateway Aggregates Model Access Across Providers
What’s changing underneath the valuation is the product architecture. Unity Gateway gives customers governed access to models from OpenAI, Anthropic, and Google through a single control plane. Since May 2026, a new flagship proprietary or open-weight model has launched roughly every 3.7 days, according to the Databricks AI Index. That pace is driving enterprise demand for a layer that can switch providers, enforce policy, and track spend without rewriting applications.

The AI Index, launched November 16, draws on aggregated, anonymized data from more than 10,000 companies using Unity Gateway, including Fortune 500 and Global 2000 enterprises. It combines that usage data with internal model evaluations across 12,000-plus Databricks employees. The early finding: enterprise adoption has fragmented. At mid-year, most organizations relied on a handful of Anthropic models. Today, workloads are spread across model versions and providers alike.
Genie Automates SQL for Non-Technical Users
Genie One answers questions from a company’s business data using natural language, effectively automating SQL generation. A Slack integration launched this month brings that capability inside the chat app where teams already work. The goal is to keep the query workload on the Databricks runtime rather than pushing raw data out to external AI services.
Lakehouse Architecture Locks Compute Onto Databricks Runtime
Databricks developed the data lakehouse architecture, combining data warehouse and data lake capabilities for structured and unstructured data. The platform runs on Amazon Web Services, Microsoft Azure, and Google Cloud Platform, positioning it above raw hyperscaler services. Delta Lake, the open-source project adding ACID transaction support to data lakes, underpins the architecture. The company was founded in 2013 by the original creators of Apache Spark, commercializing the open-source distributed computing framework.
Databricks SQL Warehouse offers Serverless, Pro, and Classic tiers with different VM cost inclusion and auto-stop behaviors. Serverless includes VM costs in the DBU unit price with a default 10-minute auto-stop and 2-to-6-second startup. Pro and Classic charge VM costs separately, take about four minutes to start, and default to a 45-minute auto-stop. The tier choice changes the economics of sporadic versus continuous workloads.
Token Spending Shifts Toward Open-Weight Models
The AI Index reveals enterprise token spending is shifting toward open-weight models as new options arrive weekly. The Index tracks model adoption through token and spend shares across proprietary and open-weight models, plus weekly AI spending patterns for individual users. It also publishes continuously updated coding-agent evaluations drawn from real workflows captured through Unity Gateway, measuring task success rates and cost per completed task across model-harness combinations.
Ghodsi told All-In he believes AI has already reached artificial general intelligence by the definition his Berkeley lab used around 2010. He puts the risk of AI wiping out humanity close to zero and names cyberattacks as the real danger. “I think the cyber risk is real and really bad things can happen with that,” he said.
Sources
- Databricks valuation talks hit $225bn, nearly 20% above August
- Databricks
- Introducing the Databricks AI Index: The state of enterprise AI tokenomics
- Databricks Wants a $225 Billion Valuation Even as AI Stocks Slide
- Databricks SQL Warehouse Pricing: Steps to Compare and Estimate Serverless, Pro, and Classic|たっく|人生プチリタイヤ中ww
- Databricks discusses funding at $225 billion valuation
Photo: DKP415 (CC BY-SA 4.0)





