Databricks reads files through the S3 protocol, and that protocol does not have to point at Amazon S3. Files.com exposes an S3-compatible API, so a Databricks cluster — with the S3 endpoint set to your Files.com site — reads files directly from us, and Auto Loader picks up new arrivals as they land. There is no hyperscaler bucket to provision, sync into, or pay egress on.

Partners and upstream systems send files to your Files.com site over SFTP, FTPS, or HTTPS; an Automation validates each one, normalizes the format, and files it where it belongs. Auto Loader, pointed at the Files.com S3-compatible endpoint, streams those files into your tables — clean, current, and with no second copy anywhere. The same path runs in reverse for the results Databricks writes back out to partners.