This is the inevitable and foreseeable result of the scarcity of IPv4 addresses, and it perversely discourages IPv6 adoption. Once something has a cost, it has the potential to become revenue generating, and once that happens the incentive for companies changes to preserving the revenue stream. At that point, why would they make the effort to provide a free alternative?
I think this is good news for IPv6 deployment. As ISPs start charging more for IPv4, companies will finally have a financial reason to seek the alternative.
It's sort of like taxing carbon to make non-carbon energy more competitive.
It won't generate revenue but investment into ipv4 can be used to build a moat around your cloud business. Anyone who wants to compete with the big cloud vendors now needs not just a global network of data centers and good uplinks, but also a large pool of ipv4 addresses.
It would be very rare that any company passes the wholesale cost directly to the customer. There’s almost always some kind of markup, even for things like “administrative overhead”. Maybe that’s not widespread now, but the clear trend is reduced supply and increasing demand, so the costs will definitely go up.
I was just thinking this when reading the email Hetzner sent me. Would it be a good investment to buy 1000 IPv4 addresses now and sell them in a few years?
The thing that's saleable is routable IPv4 address space. That is, blocks of addresses which can just be announced somewhere by a new owner. I can't meaningfully sell say 81.2.89.126 even though that address is "mine".
The RIRs still manage this namespace. Their rules only allow transfers of space to LIRs that have a justified need for the addresses, the "sale" just allows you to bump their request to the top of the queue matched against your return of those addresses. At exhaustion (where most regions are now), the queue won't move unless either some kind soul gives back some addresses or, more likely they sell those addresses to somebody not at the front of the queue.
So, you can't really just buy 1000 IPv4 addresses. You would need to create an entity that needs 1000 addresses, that could buy them, and then it could use them, but then that's not really an "investment in IPv4 addresses" it's a company (ISP? Cloud provider maybe?) that you founded and provided some capital to in the form of the address space it needed.
Seems like a bad long term investment, since there's a plan for them to be worthless eventually. Economically speaking, if the market is rational, the price should tend down over time.
Of course the market may not be rational (it's obviously not super liquid, either), and it's very plausible the price creeps up over time before eventually crashing, or that we never get to widespread IPv6 adoption after all. Maybe you have some insight that they are underpriced at the moment and IPv6 adoption is further away than the market thinks. But I wouldn't contemplate this as an investment unless I had some plan to collect rent for the assets to make up for the expected eventual depreciation.
