Snowflake's Interactive Analytics workload (Interactive Tables and Warehouses) offers in-memory-class performance for Power BI DirectQuery mode at a fraction of the cost of scaling Microsoft Fabric SKUs. A demo using a 15-billion-row fact table on a Small interactive warehouse achieves sub-second P90 query latency with no user-defined aggregations. Detailed TCO comparisons show that combining a Snowflake interactive warehouse with a smaller Fabric SKU can be 47–76% cheaper than upgrading to the F1024 SKU required for Import or Direct Lake mode at that data volume. The post also rebuts claims by a Microsoft Fabric PM that non-Fabric sources like Snowflake are architecturally disadvantaged for DirectQuery, arguing that query latency and concurrency at the source matter far more than connector-level optimizations.
