I believe you, I can see how it would make sense that AWS would create a tier to exploit the spare capacity in S3 disk bandwidth, just like they did for EC2 spa

I believe you, I can see how it would make sense that AWS would create a tier to exploit the spare capacity in S3 disk bandwidth, just like they did for EC2 spare VM capacity with spot instances. Still it doesn't make intuitive sense to me how the performance AND the price can be so far. That's why I'd love a longer write-up if you know of any. It's also weird that the retrieval gives you a regular fast S3 object you can then access. Given that it's already on that hardware, is a copy even happening?