If you've worked with S3 long enough, you already know this — it's not really a file system.
We treat it like one. We talk about folders, files, paths. But under the hood, it doesn't behave like a traditional file system — and that gap shows up pretty quickly when you try to use tools that expect one.
S3 was never designed for that.
No real folders, no native file operations — just data stored in a way that looks like files, but behaves differently. And a lot of tools, especially in ML, analytics, or even some legacy workflows, don't work directly with that model.
So in practice, we worked around it.
We'd copy data out — usually into something like a file system — do the processing there, and then push it back to S3.
Think of it like storing things in a warehouse, but having to bring everything to a workbench before you can actually use it.
It worked — but it meant duplicate data, extra pipelines, and more moving parts than necessary.
AWS recently introduced something that changes that — S3 Files.
In simple terms: your tools can now work with S3 data directly, as if it were a file system, without the data ever leaving S3.
Before: you had to move data into a file system before your tools could touch it.
After: your tools work on S3 data directly — AWS handles the translation underneath.
That's the shift.
You'll feel this most in places where tools expect file-based access — ML workflows, data processing systems, shared datasets used across teams.
Instead of copying data into another storage layer, those tools now work on S3 directly.
It removes the glue work that used to sit between object storage and file-based processing.
There’s also a scale angle here: thousands of compute resources can read from the same data simultaneously, with high throughput, no duplication required.
S3 Files makes S3 feel like a file system — but it doesn't change what S3 fundamentally is.
File operations are still being translated into S3 requests under the hood. There's abstraction involved.
That's worth knowing, not worrying about. This isn't about turning S3 into a file system.
It's about removing one of the most common friction points when working with it.
If you've ever had to move data back and forth just to make your tools work — this is a welcome change.
Curious how others are thinking about this — does it simplify your workflows, or just shift the abstraction?
Read more https://aws.amazon.com/about-aws/whats-new/2026/04/amazon-s3-files/
Top comments (0)