IPv4 transfers in 2025: old blocks carried the volume

Summary

Prefixes allocated in the previous seven years accounted for about 20% of recorded 2025 transfer transactions but less than 2% of transferred address volume.
Blocks first allocated 13–25 years earlier supplied more than 55% of volume. Those first allocated at least 20 years earlier supplied about 40%, appearing in 28% of the transaction records.

The IPv4 transfer market has two ledgers. One counts the paperwork; the other counts the addresses. In 2025, they told very different stories.

An ARIN-hosted analysis of Regional Internet Registry records puts prefixes allocated within the preceding seven years in about one fifth of transfer transactions. Their share of transferred addresses was under 2%. Older inventory carried the heft: blocks allocated 13–25 years earlier made up more than 55% of volume. The 20-years-and-older cohort alone supplied about 40% of the addresses, while appearing in 28% of recorded transfers. That last group sits inside the 13–25-year cohort; the percentages are overlapping cuts, not pieces to add together.

The headline changes with the denominator. Count transactions and newer prefixes look busy. Count addresses and the older cohorts take the room. “Busy” and “big” are not synonyms, however much a market chart would like them to be.

These are registry records, not a cash register. The count includes transfers within and between RIRs, merger-and-acquisition transfers and transfers on other grounds. It records no price beside each age cohort and cannot tell whether a transfer was a negotiated sale. The age is the allocation date recorded in the registry, not the current holder’s time in possession. ARIN records an intra-region transfer as a return followed by a new allocation, resetting that date; other RIRs preserve the original date when processing a transfer.

That convention matters whenever “old address” is used as shorthand for “long-held asset.” The chart measures registry history. It does not provide a chain of commercial ownership. Its cleanest finding is narrower and more useful: transfer counts describe how often records moved; address volume describes how much numbered space those records covered. In 2025, the older cohorts dominated the second measure.

For anyone tracking IPv4 liquidity, that is a reason to keep the two columns apart. A rising count can coexist with a small amount of newer space moving. A large address volume can sit in a relatively smaller share of records. The annual totals do not tell an operator whether a particular block fits a network, but the age split does show why “number of transfers” is a poor stand-in for capacity on the move.