Snowflake

Snowflake is a cloud data warehouse and lakehouse for analytics, data engineering and AI teams, with usage-based pricing in compute credits plus storage.

Snowflake gives an organization one governed copy of its data and runs SQL, Python and AI work against it with no servers or clusters to maintain. Snowflake Inc., a listed company headquartered in Menlo Park, California, builds and operates the service. Storage and compute are separate, so virtual warehouses can be resized, paused or added while the data stays where it is. The service is hosted on AWS, Azure and Google Cloud. Compute runs in virtual warehouses. Multi-cluster warehouses start extra clusters when queries pile up, and Adaptive Warehouses size resources for each query. Time Travel reads or restores a table as it stood earlier, and zero-copy cloning makes copies of tables, schemas or databases that share the source's storage. JSON, Avro, ORC, Parquet and XML load into VARIANT columns, while Apache Iceberg tables and external tables reach files kept in a data lake. Snowpipe Streaming and Openflow bring data in, Secure Data Sharing hands live tables to other accounts, and Horizon Catalog applies masking and row access policies. Snowpark runs Python, Java and Scala beside the data, Streamlit turns it into apps, and Cortex AI, CoCo and CoWork add AI agents. The platform is aimed at data engineering, analytics and platform teams that want a warehouse or lakehouse without running infrastructure. It also suits companies that exchange live data with partners, subsidiaries or customers, and teams that buy third-party data sets on Snowflake Marketplace. Regulated industries handling sensitive data are served by the Business Critical edition, which adds Tri-Secret Secure. Developers who prefer Python to SQL work on the same tables through Snowpark rather than a separate system. A typical deployment starts with pipelines, where Snowpipe Streaming lands events from apps and devices and dbt shapes them into reporting tables. Records from SAP, Salesforce and Workday can be brought into the same account, so finance and sales reports draw on one source. Teams that run Apache Spark read Snowflake-managed Iceberg tables from outside the platform instead of keeping a second copy. A developer clones the production database to test a change without duplicating its storage, and resource monitors alert on or suspend a warehouse once spend passes a set threshold. Snowflake fits a data team whose analysts, engineers and AI builders need to work from the same governed tables.