This is interesting in that if IP addresses are allowed to trade freely (currently 4 quid per IP), I'd argue that we'll never run out of them. And this whole transition to IPv6 may never be necessary.
We'll never run out in the sense that it will always be possible to buy one if you have enough money. Just like we haven't "run out" of, say, original Apple ][s - if you really want one, and can pay enough, you can still get them in the open market.
Doesn't mean they're available at a practical price for most endeavours though. At some point it's going to become cheaper to use IPv6 than to buy from the limited pool of IPv4 addresses. Indeed for some use cases it's already there.
In the next ~20 years there will be tens of billions of things connected to the Internet, vastly exceeding the v4 limitations.
The ability to have unique addresses for trillions of objects, may enable new technologies we haven't yet thought of.
How about every house, or Sim character, in your own virtual reality universe having a v6 ip address? Silly, sure, so what.
It's technologically absurd to stick to such a limited, primitive system as IPv4, with a mere four billion addresses. It'd be identical to arguing that we don't need more bandwidth, storage, transistors, et al. because, seemingly, we already have a lot. When in fact, every time we expand such restrictions, we discover and invent vast new technology fields.
And the notion that v4 addresses should cost any meaningful sum, when we can each have a million v6 addresses for little to no cost - that makes it clear what the right direction is. Plenty should always be chosen over scarcity in computing resources.
By that logic if we had 10 IP addresses we still wouldn't run out of them if we just could trade them freely. It is pretty obvious why that does not really work.
Well, presumably if blocks of IP addresses get radically smaller, routing tables will get large enough that not every system can handle them. Given how sluggish people are about upgrading to IPv6 people obviously aren't upgrading often, so there must be a lot of legacy equipment out there.
So it's not like I can just buy an one or two IP addresses, like I could in a liquid marketplace.
If you are starting a new cloud provider and offering something like containers (say LXD), you have a big problem on your hands. IPv6 and v4 aren't really interchangeable from an end user perspective. Also, techniques are NAT64 are shocking when you look at them in detail (my jaw dropped at least ... not all protocols can be supported with NAT64, proxies need to be stateful etc.) We are in for a mess.
The economies of scale works backwards here: the more you break up a block of addresses, the more inconvenient it is to route to them and the less they're worth. I'm not sure how easy it is to transfer anything smaller than a /24
Bad jokes aside, I think the IPv6 thing is for the prospective benefit of sundry bits of hardware hippity-hopping around, which would be price sensitive.
Doesn't mean they're available at a practical price for most endeavours though. At some point it's going to become cheaper to use IPv6 than to buy from the limited pool of IPv4 addresses. Indeed for some use cases it's already there.