A Snowflake credit is a unit of compute consumption. Compute cost equals the number of credits consumed multiplied by the account's dollar rate per credit. Storage, data transfer, and some serverless services are billed separately.
A credit does not correspond to a query, row, or gigabyte. It measures the time spent running a virtual warehouse at the rate assigned to its size. A suspended warehouse does not consume credits. Total warehouse consumption depends on warehouse size and active time.
Storage is billed separately at roughly $23 to $40 per terabyte per month, depending on on-demand or capacity pricing, cloud, and region. The charge applies to compressed data and does not use credits.
Snowflake offers four editions with different credit prices. Representative US on-demand rates run about $2.00 per credit on Standard, about $3.00 on Enterprise, and about $4.00 on Business Critical, with Virtual Private Snowflake higher still. Rates vary by cloud, region, and capacity contract. The account order form contains the applicable rate, which may be below list price under a committed-spend agreement.
The same workload costs more on Business Critical than on Standard when credit consumption is identical. Higher editions add compliance and security features rather than faster compute.
Every warehouse size doubles the credits per hour of the size beneath it. The hourly rates are 1 credit for X-Small, 2 for Small, 4 for Medium, 8 for Large, 16 for X-Large, 32 for 2X-Large, 64 for 3X-Large, and 128 for 4X-Large. Snowflake bills per second after a 60-second minimum each time a warehouse resumes, so a warehouse that runs for a three-second query is billed for a full minute.
A Medium warehouse handling ETL at 4 credits per hour for 9 hours a day over a 21-day month consumes 756 credits, near $2,268 on Enterprise at $3.00 per credit. A Large running on the same schedule consumes 8 credits per hour, 1,512 credits, close to $4,536. A Large that completes the workload in half the wall-clock time has the same credit cost, while a smaller runtime reduction increases total consumption.
Cloud Services compute handles authentication, query planning, and metadata. It is free up to 10% of daily compute credits and billed above that threshold. Heavy metadata operations can increase this charge.
Serverless features consume credits separately from warehouses. Cortex AI functions, Snowpipe, serverless tasks, and search optimization each have their own rates. METERING_DAILY_HISTORY divides consumption by service type and can distinguish warehouse usage from serverless usage:
-- credits by snowflake service type, last 30 days
select service_type, sum(credits_used) as credits
from snowflake.account_usage.metering_daily_history
where usage_date >= dateadd('day', -30, current_date)
group by service_type
order by credits desc;
It depends on your edition and contract. Representative US on-demand list rates run about $2.00 on Standard, $3.00 on Enterprise, and $4.00 on Business Critical, and a capacity commitment can lower that. Your order form carries the rate that applies to you.
No. A warehouse consumes credits only while it runs. The catch is the 60-second minimum on every resume, so a warehouse that repeatedly starts and stops for tiny queries can accumulate minimum charges.
The usual causes are a resized warehouse, longer idle periods, Cloud Services usage crossing the 10% threshold, or a serverless feature such as Cortex or search optimization that began metering. Break credit consumption down by warehouse and by serverless feature to locate it.
No. Storage is a separate monthly charge, roughly $23 to $40 per terabyte on compressed data. Credits cover compute alone.