Ground stations are not antennas

AWS Ground Station's user-guide document history records, dated 14 April 2026, documentation for two new operations: ListAntennas and ListGroundStationReservations. Separate entries the same day cover Dedicated Antennas and contact versioning.

Both new operations are GETs. For anyone proxying this API read-only — forwarding GET and refusing everything else, so that a console cannot be one fetch away from booking billable antenna time — that means the new surface is reachable with no change to the proxy at all. The guarantee is about the method, not about an endpoint list, which is why the list growing does not weaken it.

The part worth repeating

Until April there was no operation that returned an antenna list. If you wanted a sense of the size of the network, ListGroundStations was what existed, and counting its results was the only thing available.

A ground station is a site. A site hosts one or more antennas. So a count of ground stations presented as a count of antennas is not a rounding error in the label — it is an undercount, by a factor that varies per site and that nobody can see from the number.

That was forgivable shorthand while the honest number did not exist. It stopped being forgivable in April, and it stopped being merely pedantic on the same day, because Dedicated Antennas makes the distinction commercially real: an antenna is now a thing you can hold exclusively, which means it is a thing worth counting correctly.

We were doing this. Our own data-sources panel reported a ground-station count under the word "antennas". We have corrected the noun, and the true count is a fan-out away for whoever wants it.

If you have a dashboard with an antenna count on it, this is worth ten seconds of checking where the number came from.

Where to read it

• https://docs.aws.amazon.com/ground-station/latest/ug/doc-history.html
• https://docs.aws.amazon.com/ground-station/latest/APIReference/API_ListAntennas.html
• https://docs.aws.amazon.com/ground-station/latest/APIReference/API_ListGroundStationReservations.html

Also this week

The catalogue-number transition is now documented by a third major provider. AWS wrote the 100,000-object catalogue transition into its own documentation in May, following CelesTrak in July and N2YO in July. The conclusion has been the same each time and can now be stated as a settled rule rather than rechecked: systems that carry catalogue numbers as numbers, from a structured feed, are unaffected. Systems that parse them out of fixed columns in legacy two-line element text are not, because that format cannot represent a number above 99,999. If you still have a TLE text parser on a live path, it is on a clock.

Checked and unchanged: all six ITU-R propagation recommendations we cite, ICNIRP 2020, and IEEE Std C95.1-2019.

Checked and not verifiable: the CCSDS publications host has refused our fetcher for eight consecutive days. That is recorded as unchecked rather than unchanged. A 403 is not a 404, and the difference matters — one means nothing has changed, the other means we do not know.