Summary

  • As captured on 2026-10-18, the SwissIX homepage states that the exchange is present in 8 datacenters across Switzerland and offers 400G ports at three of its core sites: Colozueri.ch, Digital Realty ZUR1 and Equinix ZH4.
  • The participant list, the distribution of capacity across those eight premises and any traffic series were not obtainable from the public record in the same window; the member portal at portal.swissix.ch answers with a login prompt rather than data.

The public record of SwissIX, as captured on 2026-10-18, is short and specific. The homepage states that the exchange is present in 8 datacenters across Switzerland, and that its infrastructure "has been successfully upgraded", with 400G ports now offered at three of its core sites: Colozueri.ch, Digital Realty ZUR1 and Equinix ZH4 (swissix.ch). Those are the quantitative statements about the fabric that the page carries.

A port speed is a capability, not a volume. Saying that 400G ports are available at three sites establishes that the exchange can hand a member a 400-gigabit interface at those locations. It does not establish how many such ports exist, how many are lit, which members hold them, or what share of the exchange's traffic crosses each site. The same caution applies to the site count. Eight datacenters is a count of premises. Premises become redundancy only when the fibre paths between them are independent — separate duct, separate metro rings, ideally separate buildings that do not share a campus and a single entry trench.

The naming makes that question sharper than it looks. All three named core sites point at Zurich: ZUR1, ZH4 and Colozueri.ch. If the names reflect siting, the 400G capability is concentrated in one metropolitan area rather than spread across the country, and the eight-datacenter figure describes the reach of the fabric while three-site 400G describes where its capacity actually lives. That is a reasonable reading of the names and nothing more; the page does not say which of the eight sites carry which port speeds, and it does not say whether the other five offer 400G on request.

Where the numbers are. The portal at portal.swissix.ch returns a single readable line: "Log in to your SwissIX account to keep connected." (portal.swissix.ch). Participant data, per-site statistics and traffic figures — the three things that would let a member test a redundancy claim before buying — sit behind that login. This is not a scandal; gating member data is ordinary practice among exchanges. It is a boundary, and it has a price: the redundancy argument has to be taken on the operator's word until an incident tests it.

Independent records exist but did not resolve in this window. A PeeringDB exchange record for SwissIX exists at identifier 376 (peeringdb.com/ix/376), but the machine-readable endpoint for that record (peeringdb.com/api/ix/376) returned no retrievable fields when checked on 2026-10-18. Network count, facility count and IX-Federation count therefore remain unobtained. That is a limit of this check, not evidence that PeeringDB holds no such fields, and not evidence that the exchange publishes nothing anywhere.

Two self-descriptions deserve care. The homepage says "We are a non-for-profit association" and that SwissIX is the largest Internet Exchange Point in Switzerland in terms of port capacity and bandwidth. The first is a statement of legal form with no statute, registration extract or fee schedule behind it in the record examined here; the second is an unquantified comparative with no measurement of any Swiss exchange's port capacity or traffic attached. Both may well be true.

Neither is established by the page that asserts them, and a reader pricing a port should treat them as the operator's own characterisation rather than as findings.

The upgrade sentence has the same shape. "Our infrastructure has been successfully upgraded" carries no date and no prior baseline, so it records a direction of travel rather than a measured change. A member cannot tell from it whether 400G arrived this year or five years ago, or whether the upgrade changed the path set or only the optics at the same three sites.

Four things would settle the question. A published participant list showing the number of distinct networks and the ports each holds. A per-site traffic or capacity series showing how load divides across the eight premises. Documentation of separate fibre entry paths or distinct metro rings for the named core sites. And any dated statement of peak traffic that would let the site count be compared against delivered volume. None of the four was available on 2026-10-18.

Until one of them is, the impact mechanism is straightforward. A member buys a port at a distributed exchange on the assumption that a site incident degrades rather than severs its peering, and prices the port accordingly. Where the record states a site count but withholds per-site capacity, participant concentration and traffic, that assumption cannot be tested before an incident tests it. The redundancy risk stays with the member, and — this is the part that matters commercially — it stays there without a corresponding discount, because nothing in the published material lets the member argue for one.

SwissIX is not unusual in any of this. Most European exchanges publish a footprint and gate the numbers. The point is narrower: the question a buyer actually has and the record the seller actually publishes do not meet, and the gap is wide enough to be worth naming with a date on it. The full record for this subject sits in the BTW directory entry.