Do you have a source for that?

yes sorry, what i meant to say was "the expected usage of oil over time" When a lot of the early to late 90's "we're running out of oil stuff was happening, a lot of predictions would've been based on continued increased usage of oil. Rather than it just randomly plateauing. It's likely that the predicted curve would've have been significantly more exponential than presented.

And we're also talking on a more local scale here, so this would be more centric around a single country, or north america specifically. Or perhaps assuming that third world countries would start industrializing or something. There are any number of factors that could have influenced the potential consumption predictions.

another interesting tidbit, this was also just after the time we thought we were going to build a lot of nuclear power, so arguably that influenced the older variants of the graph as well as the modern consumption of oil for power production, for example.

IPv4 addresses are a static pool, yes. But we’re continually using them more efficiently, the same as Oil.

Yeah but idk about this one. Perhaps at the scale of CDNs and proxy distribution, but generally, i don't see this being very possible, simply because in order for a site to be supported strictly by IPV6 it must be supported by all connecting clients, and considering that most clients today are uh, not IPV6. If you want your service to work, it's going to need to be IPV4. I mean sure internal communications, but those aren't real so you can use any subnet range you want, it makes no difference.

but you could quite happily host 1,000 or 10,000 websites on a single server.

it depends on what you classify as a server, what you define a website as, and how you define the usage of it, but yeah generally, ignoring the fact that this is irrelevant, it's about that simple.