Snowflake Hybrid Tables are as much as 8x faster by default

Snowflake says its new Hybrid Tables engine is open to all customers and on by default, and existing workloads get faster with no change. Hybrid Tables keep live data for apps and AI agents in the Snowflake database, and teams do not operate a different database. Snowflake says throughput is as much as 8x higher, median response time is 3x lower and slow requests are 2x faster.

How it works

The new engine speeds up lookups and writes that come again and again, and bulk inserts. Snowflake also removed one billing item, which lowers the number of charges. Apps continue to work with no change.

The numbers

Snowflake and its customers gave these results, many from the preview.

• Snowflake says costs fell 15% on average, and 40% or more for high volume work.
• Writer, an AI platform, says write time fell about 15x after it moved to a Hybrid Table.

What is not known

The figures are the best case for some workloads. The source gives no test set up and no external test.

Insight: Two data camps move toward each other

Snowflake adds live app data to its analytics database. In November 2026, Google let AlloyDB, its database, query Iceberg tables, an open format for history data. This reads as a race from two sides to hold live data and history data in one system. Builders must test each offer on their own work, since each vendor gives its own figures.