Amazon S3 cost calculator

What actually drives your S3 bill

Most people budget for Amazon S3 as though it were a disk they rent by the gigabyte. It isn't. S3 charges on at least three independent axes, and for the majority of teams the one they think about least — data transfer — is the one that dominates the invoice.

1. Storage: what is sitting in the bucket

You pay $0.023 per GB per month on S3 Standard (first 50 TB, US East N. Virginia) for the data at rest. The subtlety is that this figure is cumulative. If you upload 10 GB every month and never delete anything, you pay for 10 GB in month one, 20 GB in month two, and 120 GB in month twelve. Your storage line grows even if your usage is perfectly flat, which is why an annual estimate can never be a monthly estimate multiplied by twelve.

2. Requests: every operation is billed

Writing an object is a PUT ($0.005 per 1,000). Reading one is a GET ($0.0004 per 1,000). These are cheap enough that for a document-sharing workload they round to noise — but they are not free, and applications that poll, list buckets in a loop, or generate thumbnails on read can run up a request bill that is larger than their storage bill. If your object count is high and your object size is small, requests deserve attention.

3. Data transfer out: the line that surprises people

Sending data from S3 to the internet costs $0.09 per GB — roughly four times the monthly cost of storing that same gigabyte. A 10 MB PDF stored for a month costs you about two hundredths of a cent. Sending that same PDF to 500 prospects costs about 44 cents. The file you store once and share widely is the expensive one, and it is exactly the file a sales or marketing team creates.

AWS gives you 100 GB of egress per month at no charge, aggregated across your entire AWS account. That allowance is real, but it is shared — if anything else in your account serves traffic, it is likely already spent. The calculator leaves it off by default for that reason, with a checkbox to apply it.

How to estimate your own numbers

You need three figures, and rough ones are fine — the goal is an order of magnitude, not a forecast to the cent.

• Uploads per month. Count new files, not total files. A marketing team producing case studies, one-pagers and decks typically lands somewhere between 5 and 50 per month. Automated pipelines can be orders of magnitude higher.
• Average file size. Text documents and spreadsheets are usually under 1 MB. Design-heavy PDFs and slide decks commonly run 5–20 MB. Video is in a different class entirely — a single recorded demo can exceed 500 MB, and if video is a meaningful share of your uploads, calculate it separately rather than averaging it in with documents.
• Views or downloads per month. This is the multiplier that matters most, because it drives egress. Be honest here: a link shared in an email campaign to 2,000 recipients with a 20% open rate is 400 downloads, not one.

If your files vary wildly in size, run the calculator two or three times — once for documents, once for video — and add the results. Averaging a 500 MB video in with a hundred 2 MB PDFs produces a number that describes neither.

What this calculator does not include

This is a deliberately scoped estimate covering the three charges that account for the vast majority of a typical S3 bill. It is not a full AWS billing model, and it does not include:

• Storage classes other than S3 Standard. Infrequent Access, Glacier Instant Retrieval and Deep Archive are cheaper at rest but add retrieval charges and minimum storage durations — moving rarely-touched data there is often the single biggest saving available, but the arithmetic changes.
• Regions other than US East (N. Virginia). Rates vary by region; some are materially more expensive.
• Lifecycle transitions, replication and versioning. Versioning in particular is a common source of surprise: every overwrite keeps the old object, and you keep paying for it.
• Volume tiers past the first 50 TB, where the per-GB storage rate steps down.
• CloudFront. Putting a CDN in front of S3 changes the egress economics substantially, and is usually the right move for widely-shared content.

For an authoritative, line-by-line figure, use the official AWS Pricing Calculator. Use this one to get to a number quickly and understand which lever matters.

Why storage cost is the wrong thing to optimise

Teams that go looking for S3 savings almost always start by trying to store less. For a document workflow that is usually the smallest lever available. If the calculator above shows your bill is mostly data transfer — and for file-sharing use cases it very often does — the questions worth asking are about distribution, not storage: how many times is each file being sent, is it being re-downloaded because nobody can tell whether it changed, and do you have any visibility into who actually opened it.

That last point is where raw S3 stops being a complete answer. S3 will tell you a GET happened. It will not tell you which customer opened the proposal, how long they spent on page four, or whether the link you sent three weeks ago is still being forwarded around their procurement team. For infrastructure that is exactly right; for a revenue team it leaves the most useful part unanswered.