Cloud storage sold for fractions of a cent per gigabyte carries a download charge that runs close to four times that rate, and a 10-terabyte pull ranges from $880 to more than $1,200 across the three dominant providers

Technology
September 17, 2026 · 3:30 PM4 min read

The price tag on the signup page says a fraction of a cent per gigabyte per month.

It sounds almost free. Millions of businesses, developers, and teams move their files in and think nothing more of it. Then the invoice arrives. The line that hurts is not the storage line. It is the data transfer out line, and almost nobody read that part of the pricing page. So why does downloading your own data cost so much more than leaving it there?

How the pricing gap between upload and download actually works

The mechanism is an intentional pricing asymmetry. Every major cloud lets you upload data for free and charges you to get it back out. The logic is straightforward: a provider’s network costs money to build and maintain, and outbound bandwidth is the expensive direction. But the gap between what you pay to store and what you pay to retrieve is far larger than any infrastructure argument can justify.

S3 Standard costs $0.023 per GB per month, while data transferred to the internet from that same storage starts around $0.09 per GB for the first 10 TB, a multiplier of nearly four times. A 1 TB storage bucket costs roughly $23.55 a month for storage alone, but a team pulling that same terabyte back out once faces a bill of roughly $87. Google Cloud Storage charges $0.12 per GB for egress, the highest of the major providers, and Azure lands at $0.087 per GB, making the asymmetry consistent across the board.

And that headline rate is only the beginning. Hidden charges, NAT Gateway processing fees, cross-region transfer, and API request costs routinely push the real cost of moving data well above the advertised egress rate.

What the dollar amounts look like at real workload scales

The per-gigabyte figures feel small until you picture a normal business operation. A backup job replicating a 2-terabyte file share to a second region every night runs to $1,200 per month in inter-region fees alone, a recurring cost that compounds every single month for a task most IT teams treat as table stakes.

The same 10 terabytes of outbound data costs $912.60 on AWS, $882 on Azure at $0.087 per GB, and more than $1,200 on the provider charging $0.12 per GB. Inbound data is free everywhere. The storage bill for that same 10 terabytes on AWS runs roughly $235, making the egress charge close to four times larger than the storage charge it supposedly accompanies.

Hidden surcharges stack on top. A 10-terabyte-per-month workload routing through a NAT Gateway adds roughly $0.045 per GB in processing fees, close to $460 per month in processing costs alone.

A typical multi-availability-zone app serving 50 terabytes a month carries a headline egress number near $4,400, but hidden charges for NAT Gateway processing, cross-AZ traffic, and API calls can push the actual bill substantially higher. That is a gap caused entirely by charges that never appear on the marketing page.

How consistent and verified these numbers are

According to independent egress-cost research verified against official provider documentation, moving data out of AWS S3, Google Cloud, or Azure storage costs $0.09 to $0.12 per GB, with AWS at $0.09 per GB, Azure at $0.087 per GB, and one zero-egress competitor at $0.00. These figures are published in each provider’s official pricing documentation and have been independently cross-referenced by multiple engineering and cloud finance teams.

The bills are not hypothetical. One managed services firm documented clients who walked in expecting a $4,000 monthly cloud bill and left with a $9,600 one because a new analytics tool started reading from object storage every 15 minutes. That kind of surprise egress spike is now reported widely enough that cloud cost management has become a dedicated software category.

Teams building AI pipelines are particularly exposed. The kind of chip-spanning AI workloads that move training data continuously between compute nodes can generate egress charges that dwarf the compute cost itself. Retrieval fees, API operation costs, and replication overhead routinely push real-world spend two to five times higher than raw storage rates.

Where the model breaks down and who it does not touch

The pricing asymmetry is not universal. Egress is billed everywhere except on zero-egress providers, and a cluster of newer competitors have built their entire pitch around a flat monthly storage fee with no download charge at all. Providers that colocate storage with compute, like Cloudflare R2, don’t need to charge for egress at all, creating a stark gap between them and the major hyperscalers charging up to $0.12 per gigabyte.

From January 12, 2027, switching-related fees, including data-egress charges incurred when a customer leaves, are banned for providers serving EU customers.

Hugo Rojas
Editor
Hugo Rojas is the editor of GCN. With a Master of Science in Engineering, he specializes in technology, data, and science, and brings a human-centered perspective informed by psychology.