Elastic compute
Spin warehouses up and down in seconds. Pay only for the queries you run, with auto-suspend and granular workload isolation.
Product
SQL Warehouse built for AI-native analytics
A serverless, elastic SQL engine that scales from gigabytes to petabytes, with cost controls and governance baked in.
Open table formats
First-class support for Iceberg, Delta and Parquet. Query your existing lakehouse without copying data.
Governed access
Row, column and masking policies enforced uniformly across BI tools, notebooks and AI agents.
AI-ready SQL
Native vector search, embeddings and model inference functions you can call directly from SQL.
Performance you can measure
Benchmarks customers actually see in production — not lab numbers.
10x
Faster than legacy MPP
On standard TPC-DS at 10TB scale.
70%
Lower TCO
Auto-suspend + per-second billing.
<200ms
P95 BI latency
On governed semantic models.
99.99%
Multi-region SLA
Active-active warehouses.
How it works
- 01
Connect your lakehouse
Point the warehouse at Iceberg, Delta or Parquet stored in S3, GCS or Azure.
- 02
Define governance
Apply row, column and masking policies once — enforced for every consumer.
- 03
Query from anywhere
BI tools, notebooks, agents and apps connect over standard SQL endpoints.
- 04
Scale on demand
Workloads scale up under load and auto-suspend when idle to control spend.
Frequently asked questions
Do I need to migrate my data?+
No. SQL Warehouse reads your existing Iceberg, Delta and Parquet tables in place — no copies, no ETL.
How is concurrency handled?+
Each warehouse can scale across multiple clusters with workload isolation, so heavy queries don't starve dashboards.
Can I bring my own BI tool?+
Yes. We expose ANSI SQL with JDBC, ODBC and HTTP endpoints, plus drivers for Tableau, Power BI, Looker and Metabase.
How does pricing work?+
Per-second compute pricing with auto-suspend. Storage is billed separately at cloud-object-store rates.
Bring AI-ready SQL to your data
Connect your lakehouse and run your first query in under 10 minutes.