Databricks, Inc.

Databricks sells the plumbing of the AI era: the storage, governance and compute layer on which enterprise models are trained and run. Its 2025 numbers show that the plumbing business can grow like software even when it runs on someone else's cloud.

Industry Position

Databricks is the reference platform for the open lakehouse. Its founders wrote Apache Spark, and the company has spent a decade converting that heritage into a commercial platform that now houses the majority of enterprise data workloads in North America and a growing share of Europe and Asia. The acquisition of Mosaic ML in 2023 moved it from data infrastructure into model training and serving, putting it in front of the same buyers who purchase GPUs.

Its competitive position is defined by a rare combination: open formats that prevent lock-in, and a governance layer that customers are reluctant to rip out. Once Unity Catalog holds the metadata for a company's data estate, replacing it means rebuilding lineage, access control and audit trails across every downstream tool.

Competitive Advantages

The first advantage is technical legitimacy. Spark, Delta Lake and MLflow are standards, not marketing claims, and enterprise architects prefer a platform whose primitives they can inspect. The second is the land-and-expand motion: data engineers adopt the platform for ETL, analysts follow for SQL, and AI teams arrive later for training and serving, each step increasing switching costs.

The third advantage is multi-cloud neutrality. Databricks runs on AWS, Azure and Google Cloud without favouring one, which lets customers keep their cloud commitments while standardising the data layer. Hyperscaler-owned platforms cannot credibly make that promise.

Strategic Expansion

Growth in 2026 comes from three directions: warehouse migrations away from legacy and competing platforms, generative-AI applications built on governed enterprise data, and industry-specific solutions in financial services, healthcare and manufacturing. Each carries higher margin than raw compute because it bundles governance, tooling and support rather than selling storage by the terabyte.

Internationally, the company is expanding through cloud marketplaces and sovereign deployments in Europe and Japan, where data-residency rules make a neutral platform more attractive than a single hyperscaler's stack. It has also begun packaging vertical agents on top of Mosaic AI, a move that competes directly with application vendors.

Risks & Outlook

Consumption revenue cuts both ways: when customers optimise cloud spend, Databricks feels it immediately, and gross margins are structurally capped by the compute it resells. Competition is converging from two sides — Snowflake on data, the model labs and hyperscalers on AI — and the company must keep shipping platform features faster than either can bundle them.

The path to a public listing will be judged on durable growth and margin discipline rather than headline valuation. On both counts Databricks has more room than most private software companies. VerityRank Score of 93/100.