Snowflake pricing is a consumption model with three separate meters: compute (measured in credits and billed per second while virtual warehouses run), storage (per compressed terabyte per month), and a cloud services layer (free up to 10% of daily compute credits). You don’t pay a flat fee — you pay for what you use, and the dollar cost of a credit depends on your edition, region, and cloud provider.
Understanding these three meters is the key to controlling spend. Once you know how they work, see Snowflake cost optimization for how to cut the bill, and SaaS cost management for the bigger picture.
At a glance
- Compute is the dominant cost (~80% of a typical bill), charged in credits.
- A credit costs roughly $2 (Standard), $3 (Enterprise), or $4+ (Business Critical) for US on-demand.
- Warehouse size doubles credits at each step: XS=1, S=2, M=4, L=8, XL=16 credits/hour…
- Storage is about $23/TB/month on compressed data — usually a small slice of the bill.
The three things Snowflake charges for
1. Compute (virtual warehouses) — billed in credits
Virtual warehouses are the compute engines that run your queries, load data, and perform transformations. They consume credits only while running, billed per second with a 60-second minimum each time a warehouse starts or resumes. When a warehouse is suspended, it costs nothing.
Warehouses come in T-shirt sizes, and each step up doubles both the compute power and the credits burned per hour:
| Size | Credits/hour |
|---|---|
| X-Small | 1 |
| Small | 2 |
| Medium | 4 |
| Large | 8 |
| X-Large | 16 |
| 2X-Large | 32 |
| … up to 6X-Large | 512 |
This doubling is why oversizing is so costly — and why right-sizing is the highest-leverage optimization.
2. Storage — per terabyte per month
Storage is billed on the average monthly volume of compressed data (Snowflake’s compression often shrinks raw data 3–5×). The on-demand list price is around $23/TB/month for AWS US East. Note: older guides cite $40/TB — that rate is outdated. Storage is real but usually minor next to compute.
3. Cloud services — mostly free
The cloud services layer handles authentication, query planning, and metadata. It consumes credits too, but Snowflake only bills the portion that exceeds 10% of your daily compute credits. Most teams stay under that threshold and pay nothing extra — though heavy use of serverless features can push you over.
What a credit actually costs
A credit’s dollar value depends on three things:
- Edition: Standard ~$2, Enterprise ~$3, Business Critical ~$4 (US on-demand). Higher editions add features like multi-cluster warehouses, materialized views, and longer Time Travel.
- Region: US regions are cheapest; non-US regions carry a premium (often 30–60%).
- Cloud provider: AWS, Azure, or GCP — broadly comparable within a region.
- On-demand vs capacity: committing to annual capacity (typically $25k+/year) cuts per-credit rates, often 20–45% at volume.
The “hidden” cost drivers
Beyond warehouses, several features quietly consume credits with no obvious warehouse attached:
- Snowpipe — continuous ingestion, billed per file notification (lots of tiny files = expensive).
- Automatic Clustering — re-clusters tables continuously in the background.
- Materialized views and Dynamic Tables — consume credits on every refresh.
- Cortex AI functions — billed per token, unbounded by warehouse size; a single large job can cost thousands.
- Data transfer — egress across regions or clouds.
These are the line items that surprise teams on their first real invoice — and the reason an account with little data can still run a large bill.
A simple mental model
Snowflake’s cost has almost nothing to do with how much data you store and almost everything to do with how many warehouses run, how large they are, and how long they stay on. A single Medium warehouse left running 24/7 on Enterprise burns 4 credits/hour × 24 × 30 × ~$3 ≈ $8,640/month — more than many companies’ entire storage bill.
That’s also the good news: because the model rewards management directly, disciplined teams cut bills 30–50% without losing capability. See Snowflake cost optimization for the playbook.
Frequently asked questions
How does Snowflake pricing work?
Snowflake uses a consumption model with three meters: compute (credits, billed per second while warehouses run), storage (per compressed TB per month), and cloud services (free up to 10% of daily compute credits).
How much does a Snowflake credit cost?
For US on-demand, roughly $2 on Standard, $3 on Enterprise, and $4+ on Business Critical. Non-US regions cost more, and annual capacity contracts reduce the rate.
Is Snowflake priced by data volume?
No. Storage (per TB) is usually a small part of the bill. The dominant cost is compute — credits consumed by virtual warehouses while they run.
What is a Snowflake credit?
A credit is the unit of compute consumption. Warehouses burn credits per hour based on their size (doubling at each step), billed per second with a 60-second minimum.
Why is Snowflake more expensive than expected?
Usually because warehouses are oversized or never suspend, and because serverless features (Snowpipe, automatic clustering, Cortex AI) add charges with no visible warehouse.
Written by the FinOpsDaily team. Last updated: 2026. Figures reflect Snowflake’s published 2026 pricing and vary by edition, region, cloud, and contract — verify current rates before deciding.
