VAST Targets Databricks and Snowflake
VAST Data
VAST is trying to turn storage into the control point for the whole data stack. Instead of landing data in one system, copying it into a warehouse, then spinning up separate compute to transform and query it, VAST keeps files, metadata, SQL access, and serverless execution in one fabric. That makes it a real alternative for AI teams that want warehouse style analytics and data engineering without sending data into a cloud native platform first.
-
Databricks and Snowflake still start from cloud software economics. Snowflake separates storage and compute into cloud services and virtual warehouses. Databricks sells SQL warehouses and serverless compute on top of its lakehouse. VAST is different because the same on premises system stores the data, catalogs it, and runs processing against it.
-
The practical wedge is unstructured AI data. A team can keep raw video, model checkpoints, and training files in VAST, search them through a catalog, run SQL to pick subsets, and launch preprocessing jobs without ETL into another platform. That is closer to Databricks workloads than to a traditional storage array.
-
This also changes who can buy VAST. Classic storage vendors mostly sell infrastructure budgets. Databricks and Snowflake sell data platform budgets. By bundling storage, database, and execution, VAST can go after larger seven figure platform deals, which already show up in its enterprise contract sizes and cloud provider agreements.
The next step is a split market. Cloud first analytics teams will keep using Databricks and Snowflake, while GPU heavy enterprises and sovereign clouds will want one system that keeps data local and feeds both analytics and AI pipelines. If VAST keeps winning those environments, it stops being compared to Pure or NetApp and starts being evaluated as a full data platform.