The Quiet Land Grab in Enterprise Data
Snowflake has spent the last two years doing something that gets little attention outside enterprise IT circles: systematically building a data platform that makes switching away feel less like a business decision and more like a surgery. Its Data Cloud strategy, which links storage, compute, governance, and increasingly AI workloads under one billing relationship, has started pulling customers away from the modular, best-of-breed stacks that Databricks built its reputation on. The shift is not dramatic. There are no splashy announcements or hostile takeovers. It is happening deal by deal, renewal conversation by renewal conversation, inside procurement cycles that never make the news.
Databricks, for its part, is not standing still. The company has pushed hard into the enterprise with its Unity Catalog governance layer and a series of acquisitions designed to close the gap on Snowflake’s native capabilities. But closing a gap is harder when the competitor is simultaneously widening it. Snowflake’s recent product releases around Cortex AI, document processing, and native application frameworks have added surface area that Databricks now has to match on two fronts – the technical and the commercial.
What the Data Cloud Actually Sells
The phrase “Data Cloud” is marketing language, but the architecture behind it is real. Snowflake’s core proposition to enterprise buyers is consolidation: instead of managing separate tools for ingestion, transformation, warehousing, machine learning, and data sharing, customers can run all of it inside Snowflake’s environment. That is appealing to CFOs and CIOs who are under pressure to reduce vendor sprawl and unify security posture. Databricks pitches a similar vision around the Lakehouse, but Snowflake has historically been stronger in the finance and operations departments that write the largest checks.
Snowflake’s go-to-market motion leans heavily on its Marketplace, where third-party data providers and application builders publish offerings that run natively inside a customer’s Snowflake account. This creates a network effect that has nothing to do with technical performance and everything to do with procurement convenience. When a healthcare company can buy claims data, run analytics on it, and share outputs with a partner – all inside one platform without moving data – the value of leaving drops considerably.
This is where Databricks faces a structural problem. Its strength is among data engineers and ML practitioners, the people who care deeply about open formats like Delta Lake and the flexibility of notebook-driven development. Those are real strengths. But enterprise purchasing decisions are not made solely by practitioners. When the conversation moves to the boardroom, Snowflake’s pitch – one platform, one contract, one security perimeter – often sounds cleaner than Databricks’ more technically nuanced story about interoperability and open standards.
Where the Pipeline Competition Gets Specific
The enterprise data pipeline market is where this competition becomes most concrete. Companies building modern data stacks need to move data reliably from source systems into wherever analytics and AI workloads run. Databricks has historically benefited from pipelines feeding its Lakehouse, with tools like Delta Live Tables designed to keep data processing inside its ecosystem. Snowflake has responded by building out its own native pipeline capabilities and partnering aggressively with ETL vendors to ensure their tools write to Snowflake first.
That partnering strategy matters more than it might seem. When Fivetran, dbt, and similar tools are tightly integrated and co-marketed with Snowflake, the practical effect is that a new enterprise data team setting up infrastructure defaults toward Snowflake as the destination. Defaults are sticky. Changing a default in a production data environment requires a business case, engineering resources, and executive sign-off – exactly the kind of friction that keeps renewal rates high.
Snowflake’s Cortex AI layer adds another dimension to this. By offering large language model access and document AI features directly inside the platform, Snowflake is making a bet that enterprises would rather run AI workloads where their data already lives than move data to a separate AI environment. Databricks has its own answer in the form of MLflow and its Model Serving infrastructure, but Snowflake’s version requires less configuration for a buyer who is not a machine learning specialist. That audience – the data analyst, the business intelligence developer, the finance team running forecasting models – is exactly where Snowflake has always been strongest.
The deeper tension is about what “enterprise data platform” will mean in three years. If Databricks successfully positions itself as the open-standard alternative – the platform that doesn’t lock you in, that plays well with everything, that gives practitioners control – it has a real story to tell. But openness is a harder sell when the person across the table from a Snowflake account executive is looking at a renewal with bundled discounts, native AI features, and a Marketplace full of data products their team is already using. Snowflake is not winning because it is technically superior in every dimension. It is winning, in many cases, because it has made the cost of switching feel higher than the benefit of leaving.
