Understanding Credit Consumption
Snowflake · Cost Management & Monitoring
Theory
Most Snowflake usage is measured in credits, a unit priced according to your contract and edition. Virtual warehouses consume credits per second while running, based on size. Serverless features such as Snowpipe, automatic clustering, materialized view maintenance, search optimization, serverless tasks and Cortex functions also consume credits. Cloud services usage consumes credits too, but it is billed only for the part that exceeds 10% of that day's warehouse usage.
Storage is billed separately, as a monthly charge on the average compressed data stored, including Time Travel and Fail-safe data. Moving data out to other regions or clouds can add data transfer charges. ACCOUNT_USAGE views such as WAREHOUSE_METERING_HISTORY show where credits went.
Why we use it
Without knowing what consumes credits, bills grow from idle warehouses, oversized compute or forgotten serverless features. Knowing the cost drivers lets you control spend without hurting performance.
When to use it
Review credit usage regularly, especially after adding warehouses or enabling serverless features. Use warehouse metering views for compute, METERING_DAILY_HISTORY for all services, and storage views for growth in table, Time Travel and Fail-safe bytes.
Compute Credits
Every running virtual warehouse burns credits per second (after an initial minimum), scaled by its T-shirt size — an X-Small burns roughly 1 credit/hour, a Small 2, doubling up the size ladder.
Storage Costs
Billed separately from compute, purely on compressed data volume per month, at a flat per-terabyte rate that doesn't change with how often the data is queried.
Serverless Feature Costs
Features like Snowpipe, automatic clustering, and Search Optimization run on Snowflake-managed compute outside any warehouse you control, and are billed as their own line item.
How it works
- Bill or usage looks high
- Query METERING_DAILY_HISTORY by service type
- Is warehouse compute the main cost? Yes: Check WAREHOUSE_METERING_HISTORY: idle time, size, auto-suspend. No: Check serverless features or storage.
- Fix the cause: suspend, resize, drop unused features or data
- Add resource monitors and budgets to keep it in check
Example
SELECT WAREHOUSE_NAME, SUM(CREDITS_USED) AS total_credits
FROM SNOWFLAKE.ACCOUNT_USAGE.WAREHOUSE_METERING_HISTORY
WHERE START_TIME >= DATEADD(MONTH,-1,CURRENT_TIMESTAMP())
GROUP BY WAREHOUSE_NAME ORDER BY total_credits DESC;Key points
- Compute, serverless features and storage are billed separately
- Each larger warehouse size doubles credits per hour
- Cloud services credits are charged only above 10% of daily warehouse use
- Time Travel and Fail-safe data count toward storage
- ACCOUNT_USAGE views expose historical credit spend