Data Pipeline Cost Calculator

Pick an option per layer, see the monthly bill and where it can be cheaper.
Estimated monthly cost — — per day · 30.4 days/month · US on-demand list prices

Where the money goes

Your pipeline

Layers

Frequently asked questions

DataPipeCal is a free data pipeline cost calculator for data engineers: it prices each layer of a pipeline across AWS, GCP, Azure, Snowflake, Databricks and 300+ tools, shows the formula, and suggests cheaper alternatives.

How much does a data pipeline cost per month?

A 1 TB/day pipeline on AWS with Glue, S3, Lambda, MWAA and Athena is about $2,000/month at US list prices. The same volume through Fivetran into a Snowflake Medium warehouse runs to tens of thousands. Change the inputs above to price your own volume, retention and tools.

Which is cheaper for Spark: Glue, EMR, Databricks or Dataproc?

Per vCPU-hour at list price: EMR on EC2 Spot (~$0.024) < Dataproc Serverless ($0.06) ≈ EMR Serverless (~$0.076 with 4 GB RAM) ≈ Databricks Jobs classic (~$0.074 incl. EC2) < Glue 6.0 ($0.077) < Glue 5.x ($0.11). Spot, Graviton and Photon change the ranking — open any option's details for the exact formula.

What is the cheapest data lake storage?

Hot data: Backblaze B2 ($6.95/TB-mo), Wasabi ($7.99/TB) and Cloudflare R2 ($15/TB, free egress) undercut S3 Standard ($23/TB), GCS Standard ($20/TB) and Azure Blob Hot ($20.8/TB). Cold data: S3 Glacier Instant Retrieval, GCS Coldline and Azure Cold are ~$4/TB-mo plus retrieval fees.

Is Snowflake or Databricks cheaper?

Snowflake bills credits per warehouse-hour ($2–4/credit); Databricks bills DBUs ($0.15 Jobs, $0.35 serverless, $0.70 SQL serverless) plus VMs for classic compute. Scheduled Spark jobs are usually cheaper on Databricks Jobs compute; BI on curated data is often cheaper on a small auto-suspending Snowflake warehouse. Pick both in the Stack dialog to compare on your numbers.

How are costs calculated?

Your inputs (TB/day, retention, compression, complexity, runs, invocations, query volume…) are turned into derived quantities — GB stored, vCPU-hours, rows, stream MB/s — and each option applies its vendor's unit prices to those. Every line shows its formula; "est." marks quote-only or unrendered prices.

Where do the prices come from and how fresh are they?

US-East on-demand list prices from vendor pricing pages and the Azure Retail Prices API, collected 18 November 2026, with no discounts, commitments or free tiers unless stated. Each option links to its source page. Prices change — verify before committing budget.

Built and maintained by , data engineer. Open source on GitHub — corrections welcome. Last price update: .

US-region on-demand list prices researched 18 Nov 2026 from vendor pricing pages, no free tiers, discounts or commitments. est. marks low-confidence or contract-dependent prices. Verify on the vendor pricing pages (linked in each layer's details) before committing budget.

Which technology will you build on?

Pick one or more — each layer will only offer options for your stack. Cloud-agnostic tools (Airflow, Kafka, Fivetran…) always show.

Cloud

Data platform

Not decided yet?

Advanced inputs