Download websites that use Amazon S3

About Amazon S3

Amazon S3 is not something a website runs. The service is where a website keeps things, which makes it unusual among the entries in a technology dataset. Nothing renders because of it. No visitor interacts with it. A detection simply means that somewhere in the delivered page, a reference points at object storage operated by Amazon rather than at the site's own server, and that single fact carries a surprising amount of information about the organisation behind the domain.

The information is architectural. Serving images, video, documents, downloads or an entire static site out of a storage bucket is a decision somebody made about separating content from compute, and it usually travels with the rest of a cloud posture. Teams that put assets there tend also to be teams with build pipelines, infrastructure defined in code, a bill from a cloud provider and an engineer who understands what a region is. For anyone selling cloud tooling, cost management, media processing, security posture review or migration services, that inference is the qualification step, and it is visible from outside the company without asking anyone a question.

Detection is where honesty is required, because this is one of the categories that undercounts hardest. Attribution normally depends on a storage hostname appearing in the page source. The moment a team places a content delivery network in front of the bucket and serves everything from its own domain, which is the recommended pattern and the common one at any real scale, the evidence disappears from the page entirely. Private buckets, backups, data pipelines and application internals never appear in a fetch at all. So a large customer can be completely invisible here while a small site that linked directly to an uploaded file is plainly counted.

The consequence is worth stating without softening. The population we can show is biased toward direct linking, which correlates with simpler setups, older pages, smaller teams and content management plugins that upload media straight to a bucket. Sophisticated infrastructure is systematically underrepresented. As of September 2026 we detect it on 1,232,634 sites, and the correct reading of that figure is a floor on visible usage rather than an estimate of the customer base, which no external crawl can measure.

Even as a floor, the geography behaves sensibly for a CDN service on offer since 2006 under a Proprietary arrangement. The country we record most often is United States at 922,341, with .com as the suffix we see most often, tracking where cloud adoption started and where the developer market is deepest rather than describing where any particular byte is physically stored.

The reason to take the list rather than check domains one at a time is that architecture questions are population questions. Which sites in a market expose storage directly. Which content management systems push media into buckets by default and therefore leak the hostname. Which businesses combine object storage with a particular framework, commerce platform or delivery network. Each of those is answerable by filtering a file and unanswerable through a search box, and the rows include organisations you would never have known to look up, which is the part a lookup tool structurally cannot provide.