SpaceX’s Low Earth Orbit (LEO) satellite Internet service Starlink has arguably been a significant game changer for many users in the Pacific Islands who were previously accustomed to geostationary (GEO) and medium-earth orbit satellite services that charged in the hundreds of US dollars a month for a megabit per second of bandwidth. Starlink now reaches even the most remote of islands — and thanks to their inter-satellite laser link mesh, users in such remote places have reported seeing download rates of 100Mb/s and more.
Writing this post from Tarawa, Kiribati, on a Starlink connection, I see data rates more akin to what I’d expect on an ADSL connection right now, with ping round-trip-times to New Zealand (NZ) in the order of 100ms. While Starlink isn’t officially licensed to operate in Kiribati yet, the Communications Commission of Kiribati has been issuing satellite ground station licenses to about 1,500 operators of Starlink kits in South Tarawa alone — an area covered by merely two Starlink cells.
There are also an unknown number of kits that were brought in without licenses, likely numbering in the hundreds, many of them later models that more easily fit into suitcases and thus escaped customs attention. All Starlink kits here operate on roaming subscriptions from Australia, NZ, Fiji or elsewhere — as mentioned above, there is no way to subscribe for fixed service in Kiribati yet.
Meanwhile, Kiribati, like so many other Pacific Island nations, has sought connection to the international fibre network. A cable project is underway, funded by Australia, Japan, and the US. I’ve had the opportunity to marvel at the construction site for the cable landing station here and see the concrete pillars for its elevated foundations getting poured last week. The ETA for the cable is late 2025. For the I-Kiribati people in Tarawa, it can’t come soon enough.
Why? Because some Starlink users here have had mail from SpaceX asking them to take their units back to the economy they were registered in, formally pointing out that the conditions for the roaming plans don’t permit permanent use in a single location. Or register the unit locally as a fixed unit.
Given that Starlink isn’t currently able to offer fixed service in Kiribati, this seems a little mean to some of the recipients, especially because roaming service is a premium service in most places that costs more than fixed service. In NZ, they’ve recently ramped up the price for the roaming service by over 10%. So why this insistence on the small print if Starlink is clearly solving a problem here for so many folks and they are making good money with it?
The answer is a little complicated: There are only so many Ku-band beams that Starlink satellites can direct at Dishys (Starlink kit transceivers) on the ground in a cell. And they’re clearly getting close to reaching that number here, in a place where there is no comparable connectivity of similar quality yet.
It’s a problem we’ve already seen play out in the lifestyle block belts of Auckland and Sydney, where the population of resident IT-savvy folk on small plots beyond the reach of fibre pushed Starlink’s capacity to the limit at times. The usual response to this has been to restrict the sale of fixed location units, for example, in NZ, where Starlink lowered the hardware price for ‘rural’ areas but kept the price for ‘urban’ areas high — with downtown Hamilton meeting the ‘rural’ criteria at one time even as a city of 185,000 people, while the Auckland lifestyle block belt was classified as ‘urban’. The difference: Hamilton has fibre almost everywhere.
So that puts SpaceX in a difficult position: Even when they get a license to operate Starlink in Kiribati, they might not be able to sell enough fixed service here to meet the demand. So, what happens then?
SpaceX’s policy to date seems to have been to rather not sell than to sell bad service. Cue Starlink’s recently announced community gateways. These are Ka-band based, so they don’t compete for spectrum capacity with the end user, Dishys. At APNIC 58 and the Pacific IGF 2024 in Wellington, it transpired that SpaceX has offered Tarawa’s cable company the use of such a gateway as a stop-gap measure until the arrival of the cable, which will have ten times the bandwidth that the gateway will be able to offer. I understand that the Kiribati side is currently trying to determine a suitable site for such a gateway.
This shines a spotlight on a more widespread issue, however. For many islands, getting a fibre connection is an aspiration that represents a high bar financially. Island ISPs often spend years fundraising for cable connections, and even where donors are involved, there is usually an expectation that the island that receives cable will contribute significantly to the project in terms of finances.
This requires the island ISPs to have subscribers. Now. But if the more IT-savvy part of the population decides on a hand-in-mouth approach and ‘defects’ to Starlink now, the island ISPs lose the very customer base that would help them pay for the cable, locking all of the island into (presumably) higher-priced Internet for the foreseeable future.
Here’s the catch-22 — on remote islands with populations of just a few hundred, Starlink can currently provide adequate capacity, which will likely meet demand for the next decade. However, on islands with populations in the thousands but no undersea cable, Starlink’s capacity may struggle to keep up with future demand. Ironically, widespread adoption of Starlink could delay the arrival of the much-needed cable, which would provide a more sustainable long-term solution.
How do we solve this conundrum?
Watch Ulrich’s presentation on ‘How fast can Starlink really be?‘
at APNIC 58.
Dr Ulrich Speidel is a senior lecturer in Computer Science at the University of Auckland with research interests in fundamental and applied problems in data communications, information theory, signal processing and information measurement.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.
Where’s the presentation docs and ~300 megs of movies mentioned in the session?