Eitsop:

 

is this a fair cost for providing this?

 

This is one heck of a big question, and I don't think it's one which has a yes or no answer.

Some insights to consider...

If an organisation wants to obtain IPv4 address space, they need to contact their local RIR (Regional Internet Registry), in our case, APNIC (Asia-Pacific Network Information Centre).

All RIR's now have policies in place to restrict acquisition of IPv4 from the RIR, due to exhaustion of the IPv4 address space worldwide. In APNIC's case, a new provider is only able to receive a total of a /23 worth of space from APNIC. This means a new entrant to the market can only obtain 512 IPv4 addresses in total, for everything from their network, backbone, ancillary services, through to customer addressing.

If an organisation wants to acquire more IPv4 address space, they must then either lease or purchase IPv4 address space from another provider.

In the case of leasing address space, of course this would therefore come with a direct monthly cost to the acquiring organisation which would be passed on.

If purchasing - well, it's not cheap. Looking at recent past sales, even a /24 (256 IPv4 total) is anywhere from $33 USD ~ $42 USD currently ($8,448 USD ~ $10,752 USD) just to obtain the space - ouch!

Of course this does mean that providers who have been around for a while, likely obtained a significant resource allocation from the RIR, and therefore had a lower initial cost of acquisition of address space - as well as sitting on a bit of a goldmine!

Once you have the IP space, you then need to pay annual membership fees to your RIR, such as APNIC. The initial /23 allocation for example comes with an annual fee of $1,625 AUD at time of writing. This fee increases as the resources held within your APNIC account increase.

All of the above is before a single piece of network infrastructure is touched, but gives some insight as to why CG-NAT is being rolled out to conserve address space and limit how many dollars need to be spent in purchasing new subnets.

All of this is before touching a single piece of network infrastructure, which of course there are costs to supporting.

Depending on systems and processes in use on a network, providing static IP addressing can also sometimes (moreso in older networks), be more complex to provide from a provisioning perspective than dynamic or CG-NAT addressing, as these can be provided from a dynamic pool on a local BNG, not requiring singular routes to be installed into the routing table.

I could go on and on, but this is a relatively consise way of saying that yes, there are both direct and indirect costs, one-off and recurring, for providers to be able to provide IPv4 addressing. These direct and indirect costs of providing addressing can also fluctuate wildly between providers, depending on the systems and processes they're using, and whether they're long-time holders of large amounts of IPv4 space, or needing to regularly acquire IPv4 space at significant cost on the market.