Databricks Investment Thesis

By Frank Downing, Jozef Soja

Director of Research, Next Generation Internet

As Director of Research, Frank guides the research of the Next Generation Internet team. Frank's research focuses on cloud computing and cryptoassets.

Research Analyst

Jozef Soja joined ARK in January 2023. As an Analyst, he focuses on artificial intelligence, cloud computing, and software-as-a-service in support of ARK’s Next Generation Internet strategy.

Enterprise technology tends to create a new platform layer that becomes essential to enterprise operations. Mainframes had operating systems. The internet era had databases and middleware. The artificial intelligence (AI) era demands something different: a unified data and intelligence platform capable of governing, processing, and learning from an organization's entire information estate. ARK believes Databricks is building that platform, and the economic evidence suggests that it is scaling faster than consensus appreciates.

The company’s timing also aligns with a structural inflection in enterprise AI adoption. Analysts estimate that global IT spending will reach $5.43 trillion in 2025, with AI infrastructure demand driving much of that growth.¹ That said, spending on AI models alone does not generate enterprise value. Organizations must prepare, govern, and operationalize data before any model can produce reliable output. Databricks addresses that structural bottleneck. Its "lakehouse" architecture collapses what historically were separate systems—data warehouses for analytics, data lakes for engineering, and bespoke environments for machine learning—into a single platform built on open data formats and consumption-based pricing. ARK's research indicates both that AI training costs have declined ~70% annually and that we are still in the early stages of AI use cases reaching production. As training and inference become less expensive, more enterprises can justify deploying AI workloads, driving greater consumption on the data platforms that prepare and govern the underlying data. As enterprises shift from AI prototyping to governed, production-grade deployment, the platform controlling data governance and model operationalization should capture disproportionate value.

Databricks’ revenue trajectory reflects that conviction, reporting $2.6 billion in fiscal year revenue ending January 2025 and expected annualized revenue to reach $3.7 billion by mid-2025—roughly 50% year-over-year growth.² By early 2026, what we believe is credible reporting placed the company’s annual revenue run-rate above $5.4 billion, with year-over-year growth exceeding 65%.³ For context, Snowflake reported $3.6 billion in fiscal year 2025 revenue, and net revenue retention (NRR) of 126%. Databricks has disclosed NRR exceeding 140% across multiple periods,⁴ meaning existing customers expand consumption far more aggressively than its public peer. This is the hallmark of a consumption flywheel: once an organization centralizes data engineering workloads on Databricks, it naturally expands into  Structured Query Language (SQL) analytics, then to machine learning, then to generative AI, with each new workload consuming additional compute units within the same governed environment.

The composition of that expansion reveals Databricks' multi-vector growth strategy. Databricks' SQL business alone was targeting a $1 billion run-rate by fiscal year-end January 2026, up from $600 million a year prior.⁵ Simultaneously, its AI product portfolio reached a $1 billion run-rate. ⁶ That dual-engine growth, displacing legacy data warehouses while attaching new AI workloads, creates nonlinear scaling dynamics. Databricks acquired MosaicML in 2023 as enterprise demand for custom large language model (LLM) training accelerated, securing AI training and inference expertise that positions it to serve enterprises seeking tailored model development. Unity Catalog, the company's centralized governance system for data and AI assets, serves as a single surface for access control, auditing, lineage, and discovery across every workload.

The customer cohort data reinforces enterprise commitment. Databricks reported ~15,000 customers, with 800 spending more than $1 million annually and 70 exceeding $10 million.⁷ Those are infrastructure-grade commitments. Moreover, its strategic acquisitions—MosaicML for LLM training, Tabular for open table format leadership, and Neon for Postgres-based transactional capabilities—signal Databricks' deliberate expansion from analytics into operational database territory, positioning the company to serve emerging agentic AI application patterns that require both analytical and transactional data access.⁸ We believes that such execution velocity reflects the leadership of CEO Ali Ghodsi, Ph.D., and a founding team that created Apache Spark, Delta Lake, and MLflow open-source projects that anchor developer workflows across the industry.

The asymmetric case rests on what happens when a consumption-based platform with 140%+ NRR continues compounding across a customer base still early in its AI adoption curve. Enterprise AI workloads are becoming core operational infrastructure. ARK's research suggests that AI tools already improve knowledge-worker productivity by roughly two times in domains like AI-assisted coding, and we believe those productivity gains will expand as models improve and cost declines continue. The platform governing and processing the data beneath those workloads benefits from every incremental model deployment, every new agent, every additional governed dataset. Open-source ecosystem gravity around Delta Lake and MLflow anchors developer workflows in Databricks-originated standards, extending the company's relevance beyond any single product cycle.

Several structural risks warrant attention. Databricks remains private, limiting visibility into its audited margins and detailed unit economics. Microsoft Fabric's rapid adoption of 21,000 organizations within 18 months represents genuine bundling pressure.⁹ Consumption models can decelerate during enterprise cost optimization, and execution in the transactional database adjacency remains unproven at scale.

ARK’s research indicates that Databricks represents a defining platform for the AI era, one whose economic flywheel is accelerating as enterprises move from AI experimentation to production-scale deployment. In our view, this convergence of data governance, analytical processing, and AI operationalization on a single open platform represents a structural shift that the broader market has neither recognized nor priced fully.

Important information 

We offer our view of Databricks because it is part of the total ARK Venture Fund portfolio. To see the most updated portfolio, please click here. Holdings subject to change. Not a recommendation to buy, sell, or hold any specific security. 

BEFORE INVESTING YOU SHOULD CAREFULLY CONSIDER THE FUND’S INVESTMENT OBJECTIVES, RISKS, CHARGES AND EXPENSES. THIS AND OTHER INFORMATION IS IN THE PROSPECTUS, A COPY OF WHICH MAY BE OBTAINED HERE. PLEASE READ THE PROSPECTUS CAREFULLY BEFORE YOU INVEST. 

There is no assurance that the Fund will meet its investment objective. The value of your investment in the Fund, as well as the amount of return you receive on your investment in the Fund, may fluctuate significantly. You may lose part or all of your investment in the Fund or your investment may not perform as well as other similar investments. Therefore, you should consider carefully the risks at the bottom of this page before investing in the Fund. 

ARK Investment Management LLC is the investment adviser to the ARK Venture Fund. 

Foreside Fund Services, LLC, distributor