Blog
Leasing or buying IPv4: how the numbers actually compare
Three routes, three different commitments
There are three ways to get IPv4 address space in 2026, and they commit you to very different things.
Apply to the registry. In the RIPE NCC service region this means becoming a member and joining the waiting list for a single /24, with no date attached (is that worth it?). Elsewhere the free pools are gone or rationed. It is the cheapest route on paper and the slowest in practice.
Buy on the transfer market. A block is transferred into your name under the registry's transfer policy and stays yours. Capital up front, permanent afterward. What you may do with it, and what it costs to keep, depends on what kind of block it is.
Lease. You get the right to use and announce space registered to someone else, usually with the routing paperwork included. An operating cost, reversible, and quick.
Most comparisons stop at capital expenditure against operating expenditure. That is a real distinction, but two other things decide it more often: what kind of block you would be buying, and the yearly cost of registry membership.
PA, PI or legacy: what you would actually own
A block on the transfer market is one of three kinds, and they are not interchangeable. The kind decides who may hold it, whether its addresses may go to your customers, and what it costs to keep.
PA (provider aggregatable) is the space the RIPE NCC allocates to its members, the LIRs, so that they can assign it to their own customers. Only a member may hold it: the RIPE NCC's transfer policy lets an allocation move only to another member (ripe-807). Buying PA therefore stacks two costs, the purchase and a membership you keep paying for as long as you hold the block.
PI (provider independent) was assigned to an end user for its own network. You can hold it without being a member, through a sponsoring LIR, which makes it much cheaper to keep. The catch is in the policy: PI "cannot be used to make further assignments to other parties" (ripe-826, section 7.0). If your customers get addresses from the block, as they do at a hosting company or an ISP, PI is the wrong tool. It is cheaper than PA, but it is not the same thing at a lower price.
Legacy space was handed out before the regional registries existed. RIPE policies apply to it only where a policy says so (ripe-639, section 1.2), so in practice it is best thought of as PI without PI's restrictions: you can hold it through a sponsoring LIR, or with no contract at all, and assign it to customers. That freedom and a supply that only shrinks are why it usually sells at a premium (legacy space explained).
| PA | PI | Legacy | |
|---|---|---|---|
| Who may hold it | A RIPE NCC member (an LIR) | Any organization, through a sponsoring LIR | Any organization, with or without a contract |
| Addresses to your customers | Yes | No, your own network only | Yes |
| To keep it, each year | €1,800 per LIR account in 2026 and €1,894 from 2027, after a €1,000 sign-up | €199 sponsorship with us | Nothing, or €199 sponsorship with us for RPKI |
| Fits | Networks that assign to customers and would be an LIR anyway | Your own servers, offices and services | Customer assignments without a membership |
The membership figures are the RIPE NCC's charging schemes for 2026 and 2027. Our sponsorship is €199 per year per resource, plus a one-off €149 setup fee.
The fixed cost that decides most of this
What matters about the RIPE NCC's annual fee is not its size but its shape: it is fixed. It does not scale with how much address space you hold. You pay the same whether your LIR holds a single /24 or a /16.
That has a straightforward consequence. The smaller your requirement, the more the membership fee dominates the total, and the worse the registry route looks. For an organization holding a /16 the fee is a rounding error against the value of the space. For an organization that needs one /24 it is very likely the largest line in the budget, larger than what the same block would cost to lease for a year.
This is why the lease-versus-buy question so often has a different answer than people expect at the small end. Anyone comparing "lease a /24" against "become an LIR and get a /24" is not comparing an operating cost against a capital cost. They are comparing an operating cost against a larger recurring operating cost, plus a sign-up fee, plus a wait of unknown length.
You do not always need your own LIR
Here is the part that changes the arithmetic, and it gets missed constantly.
Owning address space does not always make you a RIPE NCC member. A bought PI block has to be covered by a contract with an LIR that maintains the registry relationship, but that LIR does not have to be yours, and legacy space needs no contract at all. A sponsoring LIR holds the relationship while the resources stay registered to your organization, and you keep full control: announce the space from wherever you like, through whichever upstream you choose, and manage reverse DNS as you see fit.
PA is the exception, and it is the one that matters if you assign to customers. To hold a PA block you must be a member, so unless you are one already, a PA purchase brings the sign-up and the annual fee with it.
So the honest framing depends on the block. For your own network it is "lease, or buy PI and pay a couple of hundred a year to keep the registration in order". For addresses that go to customers it is "lease, buy legacy at a premium, or buy PA and carry a membership".
What buying is good at
Buying makes sense when your requirement is stable and you can fund it.
- The cost nearly stops. After the purchase and the transfer, the ongoing cost of PI is a sponsorship fee, small and predictable, and legacy can cost nothing at all. PA keeps its membership.
- It is an asset. Address space can be sold again; lease payments do not come back. Prices do move, though: a /24 sells for less than half its 2022 peak.
- Nobody can take it back. The registration is yours for as long as you pay the registry's fees and keep the contract it requires. There is no renewal conversation and no counterparty risk.
- It suits scale. The larger the block, the better buying looks, because the fixed costs spread further.
The obvious drawback is that a purchase is permanent, which is an advantage when your requirement is stable and a liability when it is not. Even a single /24 is a meaningful one-time outlay, and that capital is committed.
What leasing is good at
Leasing makes sense when the requirement is uncertain, temporary, or moving. Each step, from the first message to live addresses, is in how leasing IPv4 works.
- No capital and no membership. Nothing to fund up front, nothing to join.
- Your customers are covered. Our Terms let you sublease leased space to your own customers, which is the freedom PI does not give you.
- It is fast. No registry transfer is involved, so leased space is usually live within 24 hours of payment. A purchase depends on the registry: we file transfers with the RIPE NCC within 24 hours of payment, and the RIPE NCC completes most "within one or two business days" once the documents are complete, while ARIN and APNIC add steps of their own (transfer rules compared).
- It flexes. Month to month, or one to four years. A monthly lease ends on 30 days' notice, and when a fixed term ends you can resize the block or hand it back.
- It is reversible. You are not committing to a view about where IPv4 prices go next.
The drawback is that the payments continue and you build no asset. Over a long enough horizon on a stable requirement, leasing costs more in total than buying.
A decision you can actually make
Strip it back to four questions.
Will your customers get addresses from it? If yes, PI is out, and you are choosing between a lease, legacy at a premium, and PA with a membership. If no, PI through a sponsoring LIR is the cheapest way to own.
How certain is the requirement? If you are confident you will still need this space in five years, buying is likely cheaper overall. If there is a real chance you will not, leasing prices that uncertainty for you.
How quickly do you need it? Days means leasing. Weeks can accommodate a purchase.
Would you be an LIR anyway? If yes, PA is the natural thing to buy, and, if your LIR account has never had IPv4 from the RIPE NCC, the waiting list is worth joining because you pay the fee regardless. If no, do not take on a membership's annual fee to solve a problem that sponsorship or a lease solves for a fraction of it.
What should not drive the decision is the assumption that owning address space always requires membership. It does for PA and not for PI or legacy, and once you know which kind you need, the comparison gets much simpler.
Where RTM Networks fits
We do both, which means we have no particular stake in which one you pick. We hold PA and legacy blocks at RIPE NCC, lease from a single /24 upward, buy and sell in the RIPE NCC, ARIN and APNIC regions, and act as sponsoring LIR for PI and legacy space, so you can own it without running a registry.
Tell us the block size, the region, whether customers will get addresses from it and roughly how long you expect to need it, and ask for both routes: we will quote each, so you can compare them against each other and against the cost of membership. If your case is one where buying is clearly better, we will tell you that, and the same the other way around.
Contact
Advies over uw eigen situatie?
Dit artikel is algemeen; uw netwerk niet. Vertel ons waarmee u werkt, dan kijken wij er goed naar. Wij reageren meestal in 1-3 kantooruren.



