Their idle is terrible compared to anything modern. The platform itself also draws a ton of power on top of the CPUs (60W with dual IOH). $300 is the annual cost for a constant 200W load at average US prices. The calculator on the site you linked is just plain wrong for a number of reasons.
There's no such thing as free electricity, so, it's always a consideration. When newer stuff has an ROI of months, then yeah, it's electronic waste, nevermind it being an ancient platform that is no longer being supported. It's not about my opinion as one cannot argue with the math.
If modern idle was so great, then why are newer power supplies so much more wattage? I won't disagree on 60w for the rest of it, motherboard, storage controllers, nics, etc., but I highly doubt these same power draws are zero on a newer platform, so your point is moot. I also doubt that the efficiency you're touting is also bringing power costs to zero, so it doesn't cost zero--old or new--and the difference is not as steep as you claim it is. If your claim is the math proves your point, then as they would say in school, "show your work." I have no doubt
your use case can be shown by your math, but I have yet to find a single all encompassing formula for calculating all use cases of all things homelab. If you have such a formula, it would be useful for the entire community to have it.
I guess just because
you have to pay for electricity, no else can have it for free. A very arrogant stance imo. If your entire argument is just your opinions presented as facts, then nothing outside of your experience seems to exist--rois and use cases that don't apply to you simply do not exist--even when they do. Such self-centering isn't the reality of a very diverse broader community.
There's a lot of people like yourself contributing to the poisoning of the world by continuously spreading this fud about powering old gear being somehow insanely expensive. (There's even more of this on reddit's homelab communities where people are told to overbuy for most use cases and then end up downsizing at a loss.) This fud is simply not to true to those of us that see both sides of the coin and understand both opex and capx investments in relation to homelabbing.