Snowflake Credit
A Snowflake credit is the unit in which Snowflake measures compute consumption. Snowflake’s documentation describes the billed cost of compute as the number of credits consumed multiplied by the price of a credit. The separation matters: how many credits a workload uses is an engineering question, and what each credit costs is a commercial one that depends on the account type, region, and contract.
Product details below were checked against Snowflake’s documentation in September 2026. No prices appear here, because they vary.
What consumes credits
| Source | How credits accrue |
|---|---|
| Virtual warehouses | While a warehouse runs, at a rate set by its size — for Gen1 standard warehouses, 1 credit per hour for X-Small, doubling with each size up. Billed per second with a 60-second minimum each time a warehouse starts. A multi-cluster warehouse consumes that rate for each running cluster. |
| Serverless features | Features that run on Snowflake-managed compute rather than a user’s warehouse. Shown as their own line items. |
| Cloud services | Coordination work such as authentication, metadata, and query compilation. Charged only when daily cloud services usage exceeds 10% of that day’s warehouse usage. |
A small worked example shows the arithmetic. A Medium warehouse (4 credits per hour) that runs for two hours in total over a day consumes 8 credits. Run as two clusters for the same time, it consumes 16. Resized to Large and finishing the same work in one hour, it consumes 8 again — if the work really does finish twice as fast, which is a condition to test rather than assume.
What is not billed in credits
Not every cost is a credit, and budgets built only on credits miss them.
- Storage is charged separately, calculated monthly from the average number of on-disk bytes stored each day. It grows with data and retained history, whether or not anyone runs a query.
- Data transfer out of an account to a different region on the same cloud, or to another cloud platform, is charged per terabyte. Incoming data is not charged.
Using credits to manage cost
Because credits are a usage measure independent of price, they are the right unit for engineering decisions: comparing warehouse sizes, attributing usage to teams, and setting quotas. A resource monitor works in credits, for example, suspending warehouses when a quota is reached. The price per credit is a separate lever, handled through the account and contract rather than through query tuning.
The useful habit is to look at credits by source rather than as one total. Warehouse credits respond to sizing, suspension, and scaling policy. Serverless credits respond to which managed features are enabled. Cloud services charges respond to patterns such as very frequent metadata operations. A single total hides which of those moved.
How each layer of Snowflake’s architecture is billed, and which controls act on each, is worked through in Snowflake Architecture: Three Layers and What Each One Bills For.
References: Snowflake Documentation, Understanding overall cost; Snowflake Documentation, Understanding compute cost; Snowflake Documentation, Overview of warehouses.
Discover more from Insightful Data Lab
Subscribe to get the latest posts sent to your email.