Summary

  • LACNIC’s 2 November cumulative transfer file contains 46 October-dated IPv4 rows covering 74,496 addresses. Six rows dated 27 October send 24,576 addresses from SOBRALNET SERVICOS E TELECOMUNICACOES LTDA - ME to DB3 SERVICOS DE TELECOMUNICACOES S.A: 32.99% of that visible monthly volume.
  • A 7 October row records 512 different addresses from DB3 to Sobralnet Telecomunicações ltda. The two Sobralnet labels differ, the ranges do not overlap, and the public file supplies no corporate identity or deal linkage.

The most tempting calculation in LACNIC’s October transfer record is also the least defensible: subtract 512 from 24,576, write “net,” and move on. It produces 24,064. It also assumes that two differently written Sobralnet names identify the same legal counterparty and that two sets of distinct address blocks belong to one commercial arrangement. Neither assumption is in the file. A calculator cannot supply the missing contract.

Start with what LACNIC actually published. Its cumulative JSON file dated 2 November has 46 records whose transfer_date falls in October 2026. All 46 carry an IPv4 range; together the ranges cover 74,496 addresses without overlap. Forty-three records, covering 73,472 addresses, stay within LACNIC. The other three bring 1,024 addresses from RIPE to LACNIC. The 31 October snapshot has exactly the same October records. This is a stable visible cohort across those two releases, with 29 October the latest transfer date inside it.

Six entries do the heavy lifting. All are dated 27 October, all put SOBRALNET SERVICOS E TELECOMUNICACOES LTDA - ME in the source field, and all put DB3 SERVICOS DE TELECOMUNICACOES S.A in the recipient field. Each entry covers 4,096 addresses, the size of a /20. The six ranges are 177.184.128.0–177.184.143.255, 177.200.80.0–177.200.95.255, 186.225.32.0–186.225.47.255, 186.225.48.0–186.225.63.255, 187.60.32.0–187.60.47.255, and 189.84.112.0–189.84.127.255. Two 186.225 ranges touch and could be described together as a /19, but LACNIC has published two separate entries. Six records, 24,576 addresses: almost one address in every three in the visible October total.

Then comes the smaller arrow pointing toward a similar name. A row dated 7 October names DB3 as source and Sobralnet Telecomunicações ltda as recipient for 187.84.78.0–187.84.79.255, or 512 addresses. That block is disjoint from every 27 October block. The earlier recipient string is shorter and different from the later source string. The log publishes names, not a corporate identity graph. It does not say whether the two strings represent one legal person, related companies, or unrelated entries that happen to share a brand-like word.

LACNIC’s IPv4 policy manual says the public log records the transaction date, parties, addresses and relevant RIRs; intra-regional transfers require supporting legal documentation submitted to LACNIC. The public JSON does not expose that documentation, the price, a transaction identifier, a business purpose or the subsequent routing use of these addresses. Those are the missing facts that would change a conclusion about deal structure. The record supports a strong statement about registration and a weak one about commerce. Treating the latter as strong would make the registry’s sparse fields do work they were never equipped to do.

There is a real institutional stake beneath the bookkeeping.

A transfer log is often read as a market ledger, even though its entries are resource registrations. Here one set of six entries supplies 32.99% of the month’s visible address volume. An analyst can measure that concentration precisely. A board cannot infer six deals, one deal, a refund or a swap from the same rows. LACNIC has made address custody visible; the causal link between related-looking entries remains private. The public record is useful precisely when readers stop at its edge instead of painting a contract on the blank space.