Summary

  • ARIN and RIPE NCC recorded 2.13 million IPv4 addresses moving outside records explicitly classified as mergers and acquisitions in September, 39.3% above August. Unique recipients rose only 4.2%, from 216 to 225.
  • The increase survives a like-for-like concentration check, but registry records do not show prices, contracts, settlement, demand or deployment. The evidence points to more recorded movement, not a proven boom in independent purchases.

The denominator is the story

InterLIR, an IPv4 marketplace operator, compiled the August and September 2026 records and reported 2,134,784 addresses outside explicitly classified M&A in September, against 1,532,416 in August. Groups sharing a date, source and recipient increased 5.1%, from 237 to 249. Unique recipients across the two registries rose 4.2%.

Those are three different measures. Address volume counts the size of the prefixes in the records. A group is InterLIR’s analytical grouping, not a contract. A recipient count says how many named organisations appear on the receiving side, not how many independent buyers entered the market. A single large block can therefore move the first measure far more than the other two.

The M&A filter is also material. August’s unfiltered total included about 5.30 million addresses moving from Frontier Communications to Verizon Business. That corporate transfer was larger than September’s entire unfiltered activity. Excluding records explicitly labelled as M&A makes the month-to-month comparison more useful, but it cannot remove corporate movements that appear under other classifications.

A concentration effect, but not only one recipient

CloudIP received six /16 blocks from IBM and three from NCR Voyix in September, together representing 589,824 addresses. That was 43.5% of ARIN’s September RESOURCE_TRANSFER volume. The figure was large enough to dominate the headline comparison, but it does not explain all of the increase.

InterLIR ranked recipients separately in each month and then removed the same number of leading recipients from both sides. On that basis, combined volume still grew 15.7% after removing each month’s largest recipient, 11.4% after removing the top three, and 17.4% after removing the top ten. These are different cuts of a concentrated distribution, not a single estimate of underlying demand. In ARIN alone, excluding the largest recipient from each month leaves growth of 9.6%; removing CloudIP only from September instead produces a 7.7% decline against all of August. The one-sided comparison changes the denominator.

The regional split tells a similar story. ARIN volume rose 63.4%, while its recipient count remained at 92. RIPE volume rose 10.9%, and its recipient count increased from 125 to 133. Large blocks supplied 70% of September’s combined volume, up from 62% in August. At the same time, 74 registry entries in one Canada Post transfer accounted for 101,888 addresses, so a rise in entries does not necessarily mean a rise in separate transactions.

A transfer log is not a price tape

ARIN and RIPE publish transfer records to describe changes in registration, under their respective policies. ARIN’s policy says number resources are not sold under its administration; RIPE’s policy requires registration records to be updated when a transfer is completed. Neither kind of log identifies a negotiated price or proves that a distinct commercial purchase occurred. A record also does not establish how quickly a resource will be routed, whether it can move to another registry, or what operational work a recipient must complete.

InterLIR’s lease-price series is separate from the transfer data. It reports September averages of €0.43 per address per month for ARIN and €0.42 for RIPE, but it is a sample rather than a market-wide index. It does not measure IPv4 sale prices, and changes in the sample could affect its averages. The firm’s marketplace business gives readers a reason to examine methods and denominators carefully; it is not evidence that the figures are wrong.

For a network operator, the report is most useful as a prompt to ask which registry, block size and transaction path match a specific need. Monthly totals cannot answer whether an available block is suitable, what it should cost, or whether an approval will be granted. Public records make some movements visible; the portability and operating value of a resource still depend on the relevant rules and the receiving network’s circumstances.

Sources