The archive · Business Models · Product decision · 2006
Amazon rents storage by the gigabyte — S3 opens web-scale infrastructure to any developer
In March 2006 Amazon launched S3: developers pay only for what they use — $0.15 per GB-month of storage, no minimum, no upfront.
Amazon.com
What it had to solve
In 2006, scalable storage meant building and running your own data center — an upfront investment that shut out small developers. Amazon had already built one of the world's largest storage infrastructures to run its own retail site, and its Web Services division decided to rent that capability out by the unit.
How it works
In 2006, scalable web storage meant owning infrastructure. A startup that wanted Amazon-grade reliability and scale had to build its own data center — upfront capital, hardware, and operations out of reach for most small developers. Amazon, meanwhile, already ran one of the world's largest storage infrastructures for its own retail site.
On March 14, 2006, Amazon launched S3 (Simple Storage Service). It was not a consumer product but a web service: a standards-based REST and SOAP interface that let any developer store and retrieve objects from 1 byte to 5 gigabytes, with a unique URL for each file, private or public. There was no minimum fee — developers paid only for what they consumed: $0.15 per gigabyte per month of storage and $0.20 per gigabyte of data transferred.
The design goal was to make web-scale storage a utility — something developers could take for granted, scaling up and down without capital. TechCrunch called it game changing and noted it enabled entire classes of companies that could not have existed before because of infrastructure costs. Early users included NASA's Stardust@Home science project, which used S3 to store and deliver 60 million images to 100,000 volunteers.
Why it lands
- Amazon metered its own capital asset, so any developer could buy web-scale storage in gigabytes instead of building a data center.
- No minimum fee and no upfront investment removed the two barriers that had filtered out small developers.
- A simple API with low, transparent prices meant the product needed no sales force or contract negotiation.
- Running S3 on infrastructure Amazon already used for itself kept marginal cost low enough that small customers were profitable.
- Launching as a developer service made S3 the seed of a platform that later added compute and other metered services.
What it did
TechCrunch called the launch game changing and said S3 enabled entire classes of companies that could not have existed before because of infrastructure costs. Early customers included NASA's Stardust@Home project, which used S3 to deliver 60 million images to 100,000 volunteers, and podcast transcription service CastingWords. S3 became the foundation of Amazon Web Services.
What you can take
If the asset is too expensive for most buyers to build, sell access to it by the unit: metering turns a capital asset into a product anyone can afford, and the low entry price creates the market.
Since then
S3 was the first block of what became Amazon Web Services: EC2 compute followed later in 2006, and the pay-as-you-go model spread across storage, compute, databases, and AI. AWS went on to become Amazon's profit engine and the template for cloud pricing — metered, no minimums, no upfront — that Microsoft Azure and Google Cloud copied.
Sources
spotted an error? The archive wants to know.
Your turn
You just read one. Describe the brief you are staring at, and see who has been given the same problem.
Free account · 3 free questions · no card