Update 2021-05-25 - I now consider this article Starlink, Part 3 of a series. For the entire series, click on the Starlink tag.
Discussions ensued after my article Why Starlink Will Win Rural Broadband went live :-)
Starlink Business Models in Not-As-Developed Countries
There was considerable discussion about Starlink business models in not as developed countries. Despite the amazing growth and penetration of cellular, not everyone has cellular available.
I learned a lot about satellite geo / techno politics from several books about the Iridium system, especially Eccentric Orbits. A satellite system operator has to be flexible because, within its borders, each country can dictate the particulars of operating a satellite system. It's a balancing act on both sides - countries don't want the satellite operator to just turn off the sats as they pass over their country - they're too damn useful (China, Russia, and the US all use Iridium). On the other hand, the satellite operator has to price things so people don't game the system like buying Iridium minutes in a very low cost country and using them in a high cost country. Etc. A LEO satellite system has LOTS of moving parts.
A point I didn't make in my previous article is that the most expensive parts of Starlink - the satellites and the launch, is going to be paid for by the "developed world" customers. Once those satellites are in orbit, any additional revenue beyond those primary customers (that justified the satellite construction, launch, and operational expenses) is mostly "found revenue". That said, there are expenses of serving every customer, like bandwidth to the ground stations and construction and operation of the ground stations. I can also imagine there will be local repair facilities for all the Dishy McFlatfaces that will inevitably get damaged. Thus, it's reasonable (to me) to have different pricing and business models between developed countries and not as developed countries customers.
For example, in a not as developed country, put one Starlink terminal in a small, disconnected village, essentially as backhaul. Then use that bandwidth either in a Wi-Fi network, or a cellular network. Starlink would still get its $100/month revenue from that terminal, it just wouldn't be their original business model of one family or business per terminal.
One motivation for Starlink to do so is that individual countries have authority over spectrum usage within their borders. Thus they can make conditions on satellite service providers like Starlink to provide such "small village" service, as a condition of providing its normal service.
External Financial / Business / Capacity Models
Another point of considerable discussion was various financial / business / spectrum capacity modeling that concludes that Starlink isn't really viable because the financial / business / capacity model doesn't work out, thus Starlink won't be able to service enough customers to pay back its investment, etc. I understand that most who are creating these models know what they're doing, and such models generally work by accounting for as many variables as they can. But...
What strikes me, and I can't quite quantify it, is that a LEO satellite constellation is a very different analysis than a GEO satellite. You can do this kind of modeling with a GEO satellite because their coverage area per spot beam is static. You can know the customer density within each spot beam. You know the price of the satellite and the launch costs to get it to GEO. You can hope that the GEO satellite will last a decade or more.
But with a LEO satellite constellation, every satellite can potentially be earning revenue as it moves through its orbit. As each satellite flies through its orbit, it can sequentially provide service to residential, utility, mobile, and different countries. Each Starlink satellite will be earning revenue all over the planet, not just the in the US. And there will be more satellites being added for quite a while. Depending on its orbit, the customer mix of any particular satellite will vary minute to minute. Dynamic.
Thus, in mid-2021, I don't put much stock in external models of Starlink because Starlink as a system is highly dynamic and SpaceX / Starlink is opaque about so many details that factor into such models. The only thing Starlink is required to disclose is spectrum usage, the satellites themselves (to satisfy authorities about launch and orbit), and the ground terminals and user terminals. Nothing about internal pricing, capacities, enhancement of system capabilities, planned upgrades, etc.
I say "dynamic" because (no surprise) Starlink seems to embody its parent company SpaceX's ethos of "iterate fast, see what breaks, fix it, make it better, repeat". Every batch of Starlink satellites can be better and more capable. As they evolve Starlink, they're changing assumptions that those external models are built on. Just one example that I can pick on is models that have been quoted at me blithely state "Up to X customers can be serviced by the amount of spectrum available to one Starlink satellite, and X customers aren't enough to pay back the price of the satellite." Um... no. No one external to Starlink can say with certainty how many customers can be served by a Starlink satellite. I'll explain some other Starlink scenarios that would generate additional customers / revenue.
For example, to increase system capacity without additional spectrum, Starlink can basically do what cellular companies do to increase system capacity without additional spectrum - shrink the coverage area of a base station and add more base stations for more aggressive frequency reuse. In Starlink's case, they can tighten the beam footprint and add more satellites. That's more expensive, but making more satellites and doing more launches doesn't seem to be a limiting factor for Starlink.
Starlink Isn't Significant Competition For...
There are a lot of businesses that aren't going to see Starlink coming for them until it's too late. Of course the canonical example of not seeing it coming was that AT&T didn't consider the Internet as competition... until it was.
We're going to see a lot of stories like this one from Light Reading: Starlink's threat to wired broadband 'minimal' – analyst. Yep! And that is the plan. Actually the Light Reading story is a pretty good one, and it sounds like the subject of the story, Craig Moffat's analysis of Starlink is better than most. Moffat discloses many of the "unknowns" about Starlink, but does so expertly, versus my amateurish analysis. (Yes, I'm happy to admit that.)
When I mentioned in a discussion that it would be easy for Starlink to add Internet of Things (IOT) services to its capabilities, one commenter mentioned that there are satellite startups addressing the IOT market with small inexpensive satellites and corresponding low prices (like $10/month). The commenter concluded by saying that those IOT satellite operators don't think that Starlink is competition for them. My reply:
I'll further guess that a fair number of Starlink terminals won't require significant bandwidth all the time. I'm guessing Starlink will quickly get into the Internet of Things monitoring... pipelines, electrical grids, etc. One of the big differences between Starlink and, say, VSAT [Very Small Aperture Terminal] systems is that there's enough bandwidth in Starlink to allow video for security. I'll guess electric utilities would like to have a video feed of all their key sites that's not dependent on cellular or microwave.
In a later discussion I amplified on this point by extrapolating that, in direct competition for those IOT satellite systems, Starlink may well develop a small, inexpensive ground station that doesn't use a phased array antenna for situations that don't need continuous connectivity. It you have modest data requirements such as uploading a day's worth of data from a remote reservoir, it's quite feasible to point a static antenna skyward and just wait for a flyover by a Starlink satellite. Starlink's bandwidth is sufficient that only a few seconds of connectivity would be needed for such an application.
One of the reasons I'm skeptical of business models predicting that Starlink isn't viable is the possibility of additional revenue from hosted payloads on Starlink satellites. Part of the business plan for Iridium Next was hosted payloads. There's nothing to prevent Starlink from doing so. Starlink can use such potential revenue in its business modeling, but no one external could know about such a possibility until Starlink is required to disclose particulars about the payload or communications requirements.
Starlink as Bypass for the Internet / Governments
One commenter mused that Starlink might be a good way to bypass government censorship or monitoring (including by the US). I don't think that's practical (yet). At the moment, Starlink is operating as a "bent pipe" which means there has to be a ground station to provide Internet bandwidth to/from the satellites. Thus there's an effective censorship point within the country. Not to mention that Starlink uses spectrum which (theoretically) has to be licensed for use in each country. Starlink could, technically, ignore that dictum, but it would be... inadvisable to do so. It would be easy to jam Starlink, and Starlink wouldn't want to burn its bridges in any country.
Then there's the fact that Starlink is built (satellites, ground stations, user terminals) in the US by a US company. The Starlink system is managed by a US-based company, from the US. So, do the math about what the US government could do if (when) the US government sees the need. I know nothing about Starlink's internal protocols, but I've seen nothing mentioned about privacy considerations such as built-in encryption. (Perhaps relying, mostly on end-to-end encryption such as https.) I think I read that Starlink, or Elon Musk, said that internally Starlink doesn't use TCP/IP, only that only the users and the Internet connections see TCP/IP... so that's interesting.
It will be different in a few years when Starlink starts to use inter-satellite links. Then, it will be a much more resilient system, much less dependent on ground infrastructure. Once the inter-satellite links are working, that will theoretically allow communications between any two Starlink user terminals anywhere on earth to transit only the Starlink system. Again, it doesn't benefit Starlink to flout various government regulations - better to choose not to provide service within a particular country if a country's regulations are too onerous.
Iridium uses inter-satellite routing now, and that's one of its strengths. If you don't think that inter-satellite links are practical, remember that Iridium had inter-satellite links figured out on its original satellites in the 1990s. We've had some technical evolution since then.
Bottom line - I, personally, wouldn't count on my communications over Starlink being secure enough to conduct illegal activities (including illegal speech in certain countries). At least, don't consider Starlink to be any more secure than using conventional Internet or cellular connectivity.
Starlink Speeds Are Decreasing; Starlink Still Has Drop-outs
A commenter made the point (in support of their thesis that Starlink may not be viable) that reported speeds of Starlink users are declining. Yes, speeds have gone down as more users have come on the system. Starlink is still optimizing the satellite orbits, tuning all the software, and continually adding ground stations. Even if Starlink halved the average reported speed (that I've seen from first hand accounts) from 100 Mbps to 50 Mbps, it would still be considered a roaring success and very welcome because the average speeds, quality, and reliability of the other options for Broadband Internet Access in rural areas are so poor.
Starlink Capacity Concerns
As I confessed earlier, all of my "analysis" about Starlink is amateurish. I have no hard information about Starlink other than what little has been published in the open, no inside sources (Oh, I wish...), etc. So this is really going to be a leap further out.
I think that the ultimate solution for Starlink capacity concerns, at least for Ground Terminal to Satellite, and inter-satellite, will be lasers. For ground-to-satellite, there have been experiments. There's also been amazing progress in adaptive optics to focus long laser paths. Most of that is from military experimentation. Eventually someone will commercialize it. Given that very few thought that it was feasible to return a rocket first stage to the ground by landing it upright, nor to actually launch and operate a worldwide network of Internet access satellites... I wouldn't count SpaceX out for developing practical ground-to-LEO-satellite laser links.
For inter-satellite links, lasers are more practical as there's little attenuation (but not none - there are still air molecules at LEO) to worry about. Mostly it's a matter of aim, or at least a balance of a tightly focused beam (requires precision aiming) or beam spreading (higher attenuation, but aiming requirements are reduced). Mostly, it's an exercise of an n-dimensional matrix choosing which satellite to communicate with every millisecond and constantly updating aim and other parameters. I think that the requisite compute power suitable for use on satellites is now possible, and perhaps even practical.
Starlink as a Godsend to Parents who are Boaters
It's been practical for some time to provide Internet connectivity, and eventually Broadband Internet Connectivity to boats and ships out of line of sight of land. Inmarsat was the first to provide that capability (I used Inmarsat systems in the early 1980s), and now Iridium offers reasonable speeds for connectivity (but not exactly "Broadband" at 300+ kBPS). But it's never been cheap. But now Starlink intends to make mobile user terminals, as long as the vehicle can accommodate the same form factor as Dishy McFlatface.
That gave me the idea for a funny extrapolation about Starlink - every pleasure craft on, for example, the Great Lakes (that can go, say, 100 miles from shore) owned by a family with teenagers is going to get a lot more popular because now those boats can be equipped with a Starlink terminal. When Mom, Dad, or the grandparents propose a long boat trip away from land (or into remote areas like sparsely populated islands) the teenagers won't scream in agony because via Starlink they'll be able to stay in touch with their friends, even doing social videos. The parents or grandparents will be a lot happier with a fixed bill from Starlink rather than the Oh My God! bill from Inmarsat or Iridium after a week with teenagers onboard.
I can imagine a similar scenario for commercial vessel personnel on the US coasts and inland waterways, including the Great Lakes. All of the sudden, they can get Netflix! As a former full-time resident on a commercial vessel, I can testify that having access to Netflix onboard would be a very, very welcome addition to shipboard life.
Thanks for reading!
Steve Stroh
Bellingham, Washington, USA
2021-05-13
Copyright © 2021 by Steven K. Stroh