A fourth edition, Virtual Private Snowflake, appears in the table too. It is sold only through an order form with a separate yearly fee, so it is outside this guide.
Snowflake storage and data transfer cost
Storage costs $23.00 per TB per month on Standard storage in AWS US East, On Demand. In EU Frankfurt it is $24.50. Snowflake bills the average amount stored over the month, after compression.
Capacity customers pay less per TB as their contract grows. In US East the first discounted tier is $21.47 per TB, down from $23.00.
Data coming in is free. Data going out is not. In AWS US East, Snowflake lists data transfer to the same cloud and same region at $0.00 per TB. Transfer to a different region on AWS is $20.00 per TB. Transfer to another cloud or the internet is $90.00 per TB. Rates differ by region, so look up yours in Table 4(a).
Serverless features and cloud services
Some features do not use your warehouse. Snowflake docs name Search Optimization and Snowpipe as examples. They run on compute that Snowflake manages and bills in compute-hours, rounded up to the next second. The credits per hour differ by feature, and each appears as its own line on your bill.
The cloud services layer is billed at 4.4 credits per hour of use, according to the consumption table. You pay for it only when your daily cloud services use is more than 10 percent of your daily warehouse credits. Serverless use does not count toward that 10 percent allowance.
Example of the rule: if your warehouses use 100 credits in a day and cloud services use 15, the first 10 are free. You pay for 5. This is an illustration of the rule, not a measured bill.
Worked example: a Medium warehouse for 6 hours a day, 22 days
This example uses the credit counts and prices above. Medium is 4 credits per hour. The warehouse runs 6 hours a day on 22 working days and is suspended otherwise.
- Hours per month: 6 hours x 22 days = 132 hours.
- Credits per month: 132 hours x 4 credits = 528 credits.
- Compute cost: 528 credits x the price of one credit.
| Region and edition (AWS) | Price per credit | Compute per month |
| US East, Standard | $2.00 | 528 x $2.00 = $1,056.00 |
| US East, Enterprise | $3.00 | 528 x $3.00 = $1,584.00 |
| US East, Business Critical | $4.00 | 528 x $4.00 = $2,112.00 |
| EU Frankfurt, Standard | $2.60 | 528 x $2.60 = $1,372.80 |
| EU Frankfurt, Enterprise | $3.90 | 528 x $3.90 = $2,059.20 |
Now add storage. Say you store 2 TB in US East. That is 2 x $23.00 = $46.00 a month. On the Enterprise edition, the total is $1,584.00 + $46.00 = $1,630.00 a month. That total leaves out cloud services above the 10 percent allowance, serverless features, and data transfer.
These totals use the listed On Demand prices. A contract with Snowflake can change them.
Hidden costs that raise a Snowflake bill
Most surprises come from compute time you did not need. Here are the four to watch.
Idle warehouses
A Medium warehouse left on all month uses 4 x 24 x 30 = 2,880 credits. At $3.00 that is $8,640. The same warehouse used for 132 hours costs $1,584. That is about 5.5 times more for the same work. Set auto-suspend to a short idle time.
Auto-resume and the 60-second minimum
Every resume bills at least one minute. Say a team sends 100 short queries a day, and the X-Small warehouse suspends fully between each one. That is 100 x 60 seconds = 6,000 seconds, or about 1.67 credits a day, even if each query runs 10 seconds. This is a worked illustration, not a measurement. In real use the idle time before auto-suspend adds to it.
Query retries and failed jobs
If a job fails and a script runs it again, the warehouse runs again and bills again. Each run bills as normal. Fix the cause of failing jobs and cap how many times a script retries.
Cloud services and data transfer
Heavy metadata work can push cloud services past the 10 percent allowance. Sending data to another region or cloud adds a per-TB fee. Both are easy to miss if you only watch warehouse hours.
Snowflake cost vs bigquery
The two bill different things. Snowflake bills warehouse time. BigQuery's default plan bills the bytes your queries scan. Neither is cheaper in every case.
According to Google's BigQuery pricing page (read 16 November 2026), on-demand queries cost $6.25 per TiB scanned, and the first 1 TiB each month is free. Active logical storage is listed at $0.000031507 per GiB per hour. At 730 hours in a month, that is about $0.023 per GiB, or roughly $23.55 per TiB (my calculation). The first 10 GiB of storage each month is free. Google lists prices by region, so check yours.
| Example month | BigQuery (on-demand) | Snowflake (US East, Enterprise) |
| Query 5 TiB of data | (5 - 1) x $6.25 = $25.00 | Depends on warehouse hours, not bytes |
| Store 2 TB | (2,048 - 10) GiB x $0.023 = about $46.87 | 2 x $23.00 = $46.00 |
TB and TiB are slightly different sizes, so the storage rows are close but not exact. The query row shows the real difference. A few light queries cost little on BigQuery on-demand. A warehouse that runs all day costs the same whether the queries are heavy or light. Heavy, steady query use can favor a warehouse. BigQuery also has a capacity plan that bills slot time instead of bytes. Its default is per-second billing with a one-minute minimum.
For a full feature comparison, read our Snowflake vs BigQuery vs Redshift guide. If you are weighing Databricks, see Databricks vs Snowflake. If you are new to the product, start with what Snowflake is.
Who should not choose Snowflake
Skip Snowflake if you will not watch your usage. It bills compute time, so an unwatched warehouse becomes a surprise bill.
- You run a few small queries a month. BigQuery's free first TiB of queries and 10 GiB of free storage may cover you.
- Your team has no one to set auto-suspend, size warehouses and review credit use.
- You move large amounts of data between regions or clouds. At $20.00 to $90.00 per TB in AWS US East, egress adds up fast.
- You need a fixed monthly price with no usage link. A Capacity contract is the closest, but it still bills by use.
- Your data and tools already sit in one cloud provider's own warehouse, and you have no reason to add a second platform.
Snowflake can still be the right call if you need to share data or run mixed workloads. Just budget for credits, not for a flat fee.
Sources
Credit counts, credit prices, storage, cloud services and data transfer rates come from Snowflake's Service Consumption Table (effective 12 November 2026). Billing rules come from the Snowflake docs pages on compute cost, overall cost, data transfer and trial accounts. BigQuery figures come from Google Cloud's BigQuery pricing page. All were read on 16 November 2026.
Snowflake's own pricing page may show different regional prices than the tables above. Confirm your region and edition before you commit.
Photo: Direct URL (https://commons.wikimedia.org/wiki/Special:FilePath/Front_of_server_racks_at_NERSC.jpg?width=1200) — verify license before use