Not to sound combative here, because I love you guys and I don't want to see you fail. I'm just approaching this purely from a non-biased perspective. Are you comfortable sharing the costs for infrastructure? Because I think you'd find collectively there are enough tech-minded people here who would gladly help you optimise your setup if there is room for improvement. Or make recommendations to improve things.

Really, the 2.5x is the part I'd want broken down though, because with zero egress on R2 the cost can't be bandwidth, so it's coming from somewhere else, and your own list points right at it: transformations. If you're doing on the fly resizing and optimisation, that's almost certainly the driver, not storage.

If that's the case, the cheaper pattern isn't R2 instead of your hoster, it's R2 for storage plus a transform layer that caches variants. Generate each size once, push it to CDN cache, never reprocess it. imgproxy or thumbor on a small box, or even Workers, in front of R2, with aggressive cache headers. Then transforms are a one-time cost per variant instead of per request, and egress stays zero. A naive setup that reprocesses or caches poorly absolutely can come out 2.5x, so I believe your number. I just don't think it's measuring the cheap version.

If financial figures are too much, usage numbers would suffice: rough storage volume, monthly request count, egress GB, cache hit rate, transform volume. With those we can get a clearer picture of the scale.

The redundancy point I'll give you. That's the real argument and it's a good one, especially for RPC. A second independent path matters and someone does pay for api.hive.blog. The only caveat is that a DHF-funded operator being the redundancy is its own kind of dependency, so for images specifically I'd still rather see mirrored, replaceable hosting than one funded host, even if that host is you.

I think this highlights that Hive has a serious problem on a core level. If we're relying on ecosystem apps to ask for funding to sustain core critical redundancy of the chain outside of the chain itself, I think it's worth flagging that as a serious issue with Hive itself and one that could compromise it down the track if our market cap keeps slipping and HBD's pegged value continues to fall.