Cloud Storage Cost Calculator

Twenty-three dollars, and the charges that are not in it

Storing 1,000 GB on AWS S3 Standard costs about $23.00 a month — $0.023 per GB — which is roughly ₹1,978 at ₹86 to the dollar, or $276 a year.

That is the storage line. On a real workload it is frequently the smallest part of the bill, because object storage charges for four different things and only one of them is measured in gigabytes stored.

The four cost components

1. Storage volume. Per GB-month, prorated. Around $0.023/GB for S3 Standard, with Azure Blob Hot and Google Cloud Standard in a similar range.

2. Requests. Every PUT, COPY, POST, LIST and GET is billed. PUT-class requests cost roughly $0.005 per 1,000 and GET-class about $0.0004 per 1,000. Irrelevant for a few thousand large files; dominant for an application writing millions of small objects. A workload storing 50 million tiny files can spend more on requests than on storage.

3. Data transfer out. Retrieving data to the internet costs roughly $0.09 per GB after the free tier — nearly four times the monthly cost of storing that same gigabyte. Serving 1 TB a month from a 1 TB bucket costs about $81 in egress against $23 in storage.

4. Retrieval and early-deletion fees on cold tiers. Archive classes are cheap to store and expensive to read.

Storage classes, and where the trap is

Standard / Hot — frequent access, highest storage price, no retrieval fee.

Infrequent Access / Cool — roughly half the storage cost, but a per-GB retrieval charge and a 30-day minimum billing duration.

Archive / Glacier — around a tenth of the cost or less, with retrieval times from minutes to hours and a 90 to 180-day minimum.

The trap is the minimum duration. Moving an object to Glacier and deleting it a week later bills you for the full 90 or 180 days anyway, and lifecycle rules that aggressively tier young objects routinely cost more than they save. Tier on actual access patterns, not on age alone.

There is also a per-object metadata overhead on cold tiers — roughly 40 KB charged at Standard rates per archived object — which makes archiving millions of small files uneconomic regardless of their size.

Reading your estimate

The figure covers storage volume only, at a representative rate. Actual pricing varies by provider, region and class, and all three major providers price the Mumbai region above their cheapest US regions.

Two further exclusions worth naming: replication, where cross-region redundancy roughly doubles storage cost and adds inter-region transfer; and versioning, which is the single most common cause of an unexplained bill increase — every overwrite retains the previous version, and a bucket with versioning enabled and no lifecycle rule grows without limit while appearing unchanged in the console.

GST at 18% applies for Indian customers.

Common mistakes

Enabling versioning without a lifecycle policy. Old versions accumulate silently and are billed in full.

Ignoring incomplete multipart uploads. Failed large uploads leave orphaned parts that are billed until a lifecycle rule aborts them.

Archiving small files. Per-object overheads and minimum durations make this a net loss below a few hundred KB per object.

Budgeting storage and forgetting egress. For anything user-facing, transfer usually exceeds storage cost.

Comparing providers on the per-GB rate alone. Request pricing, free egress allowances and minimum durations differ enough to reverse the ranking.