How IPv4 transfers work: moving address space between holders
The free pool of IPv4 ran dry years ago, and a market quietly took its place: millions of addresses now change hands every year through a registry process most buyers and sellers go through exactly once. Here is how a transfer actually works, from agreement to registry update, what the paperwork changes, and the longer, more expensive list of things it does not.
Why transfers exist at all
The regional registries stopped having meaningful IPv4 free pools in the 2010s, but demand did not stop, so the registries adopted policies letting address space move between organisations. The market that resulted is usually dated to 2011, when Microsoft bought Nortel's addresses out of bankruptcy at $11.25 per address, the first headline sale of what had previously been treated as unownable. Fifteen years on, transfers are routine: brokers arrange them, escrow services settle them, the registries process them, and each registry publishes statistics on the space that moved.
A transfer is not a sale of property in the ordinary sense, and the paperwork says so: what moves is the registration of the space, the right to be recorded as its holder and to have it routed as yours. That distinction rarely matters day to day. It matters at the edges, which is why the process runs through the registries rather than around them.
The three kinds of transfer
- Market transfers are the common case: one organisation agrees to transfer space to an unrelated other, usually for money, under the registry's transfer policy. Most of what brokers handle is this.
- Merger and acquisition transfers move space because the company holding it was bought or restructured. The registry updates the holder to match the corporate reality, with evidence of the acquisition in place of a sale agreement.
- Inter-RIR transfers move space between registry regions, say from an ARIN organisation to a RIPE NCC one. Both registries are involved, both policies apply, and not every registry pair supports it, so a cross-region deal needs checking before it needs pricing.
The process, start to finish
Details differ by registry, but the shape is consistent, and the shape is the useful part.
- Agree the deal. Exact prefixes, price, timing, and who fixes what if the space turns out to carry problems. Get the prefixes in writing; you cannot check a description.
- Pre-checks on both sides. The registry verifies the seller actually holds the space with the standing to transfer it, and that the buyer qualifies under its policy; some registries assess need, and each publishes its own fees. Brokers earn their cut mostly here, keeping a deal that was agreed in a week from dying in the paperwork.
- Settle. Money typically moves through escrow against the registry's confirmation rather than a handshake, because the registry update, not the invoice, is the moment the space changes hands.
- The registry updates the record. The block appears under the new holder's organisation, the transfer enters the registry's published log, and the buyer sets about the part nobody briefed them on, which is the rest of this piece.
What a transfer changes
The registration record: holder, organisation, contacts, and the registry the space sits under if it crossed regions. The published transfer statistics gain a line. The registry also revokes the seller's resource certificate and any ROAs issued under it. That is the whole list. It is a real and legally meaningful change, and it is far smaller than most buyers assume.
What it does not change
Everything keyed to the addresses rather than the holder survives the transfer untouched, and this list is where the money gets lost.
- Reputation. Blocklist entries, fraud scores and classifications stay with the addresses. No listing operator is told a transfer happened.
- Routing paperwork. ROAs are the exception the registry does handle: the seller's are revoked as the transfer completes, so the block falls to RPKI-unknown until you publish your own, and a wrong origin AS or max length in that new ROA is what leaves your announcement RPKI-invalid at every filtering network. Old IRR route objects and stale reverse DNS carry over the same way.
- Geolocation. The databases keep placing the block where its previous holder was until you publish otherwise.
- History. The block's origin history, its quiet years, its flagged months, remain part of its record, which is precisely why reading that record before the money moves is the cheapest step in the whole transaction.
A useful mental model: the transfer edits one ledger, the registry's. The internet keeps a dozen other ledgers about your addresses, and every one of them updates on its own schedule, or only when you push it.
The paper trail it leaves
Transfers are among the best documented events in a block's life. The registry's record changes holder, the published transfer logs name the block, and the routing table shows the old origin ending and the new one beginning. Read together, they are how you tell a legitimate handover from a hijack wearing one's clothes: a real transfer leaves matching entries across all three, in the right order, and an impostor leaves gaps.
That trail is also your due-diligence record in the other direction. How many times a block has moved, how recently, and whether the party selling it appears in its chain at all are questions the trail answers in minutes. A block on its fourth holder in three years is not necessarily dirty, but several people in a row deciding to get rid of it is a fact you want priced in rather than discovered, and it is exactly the kind of fact a dated history puts in front of you before you become the fifth.
Look up any prefix on the front page to see its holders, its transfers and its routing history on one dated timeline. The report names the source behind each event, and says plainly when nobody has checked something.