Skip to main content

Rates last reviewed: June 2025.

How Snowflake Pricing Works

Snowflake separates compute and storage. Compute is charged in "credits" consumed while warehouses run; storage is charged per TB per month. This page explains Snowflake editions, cloud services compute, regional pricing, and the usage behaviors that can drive costs up unexpectedly.

Snowflake Editions: Standard, Enterprise, Business Critical

Snowflake editions are an important part of pricing. Standard is the entry-level offering, Enterprise adds concurrency scaling and security controls, and Business Critical includes HIPAA-ready compliance, stronger audit capabilities, and higher platform resilience. Edition selection affects which features you can use and how you should model your operating cost.

Credits and Warehouse Sizes

Snowflake exposes warehouse sizes using XS, S, M, L, XL (and larger). Each size represents a fixed number of virtual resources and consumes credits at a predictable rate. A common sizing progression doubles capacity as you step up (approximate example):

Size Approx. credits / hour Best for
XS (X-Small) 1 credit / hr Light testing, small ad-hoc queries
S (Small) 2 credits / hr Small production workloads
M (Medium) 4 credits / hr Medium concurrency analytics
L (Large) 8 credits / hr High-concurrency BI or ETL
XL (X-Large) 16 credits / hr Very large, compute-heavy workloads

These values are illustrative and follow the common doubling pattern Snowflake uses: each larger size is typically ~2× the prior size. Your contract and account edition (Standard, Enterprise, Business Critical) will determine the exact behaviour and available sizes.

Compute pricing: credits × credit price

Compute cost is the product of the number of credits consumed and the price per credit. Many publicly cited default examples use a credit price in the low single digits (for example $2–$3 USD / credit), but actual price varies by cloud, region, and any discounts you negotiate with Snowflake.

Storage costs

Storage is charged separately from compute. A commonly used list-price baseline is around $23 / TB / month for on-demand storage; long-term or capacity pricing can be lower if you commit to capacity or use cheaper long-term tiers. Storage covers both your active data and any fail-safe or time-travel retention you configure.

Cloud Services Compute: the Snowflake 10% rule

In addition to warehouse credits, Snowflake bills a small amount for cloud services compute. This covers metadata processing, query parsing, result caching, and the Snowflake control plane. In practice, many teams treat this as an additional ~10% on top of warehouse credit consumption for active analytics workloads. If you only model warehouse credit charges, you can understate the true bill.

Cloud provider & region differences

Snowflake runs on AWS, Azure, and Google Cloud, but list prices and effective costs differ by cloud and region:

Snowflake Hidden Costs and Billing Surprises

Many cost surprises are not bugs in Snowflake but configuration and usage patterns that increase compute consumption, storage retention, or cloud services overhead. The most common culprits:

1. Warehouses left running (auto-suspend)

If a warehouse is not configured to auto-suspend or uses a long suspend interval, it will continue consuming credits while idle. Set a short auto-suspend (for example, 60–300 seconds) where appropriate, and schedule persistent warehouses only when needed.

2. Auto-resume + bursty query traffic

Auto-resume is convenient — warehouses start on demand — but frequent resume cycles with short active periods can add overhead. It’s better to tune auto-suspend and auto-resume together and use a slightly larger warehouse for consistently bursty workloads.

3. Multi-cluster warehouses (scale-out)

Multi-cluster warehouses add additional compute clusters when concurrency rises. Each additional cluster consumes the same credits/hr as the warehouse size, so concurrency spikes can multiply compute costs quickly. Use concurrency scaling and max-cluster controls to limit surprises.

4. Continuous ingestion and Snowpipe

Snowpipe and similar serverless ingestion patterns make ingestion simple, but high-volume continuous ingestion will incur additional compute and potentially serverless charges. If you stream many small files or high-frequency loads, batch them where possible or monitor Snowpipe usage closely.

5. Frequent full refreshes, materialized-view maintenance, and search optimization

Full-refresh ETL jobs, frequent materialized view refreshes, or aggressive clustering and search optimization can run large compute jobs repeatedly. Optimize incremental pipelines (dbt incremental models, partitioning) to reduce repeated full-table work.

6. Time travel & retention settings

Longer time travel and fail-safe retention increase storage usage. Review retention settings (day-level retention) and purge large staging tables when no longer needed to reduce storage costs.

Practical tips to control costs

Example: quick cost snapshot

Using illustrative numbers: a Small warehouse (≈2 credits/hr) running 8 hours/day for 22 days at $2.50/credit would cost:

$2 credits/hr × 8 hr/day × 22 days × $2.50/credit = $880 / month (compute)

Add storage: 5 TB × $23/TB-month = $115 / month. Total (example): ≈ $995 / month.

Want to estimate your real bill?

Vendor list prices and regional differences make generic calculators noisy. Use your actual workload inputs — queries, active warehouse hours, edition, and storage retention — to get a realistic estimate.

Estimate your Snowflake bill with your actual query volume and warehouse size

Note: Numbers on this page are illustrative. Snowflake pricing varies by contract, cloud provider, and region — always confirm rates in your Snowflake account or negotiated contract.

Compare platforms

Read the other guides to compare compute, storage, and real-world cost behavior: