Starlink : Markets and Marketing

[Eka]

Most of Comcast's subscribers are in the same boat--they don't have a reasonable alternative, and Comcast knows it.  Whether it's exorbitant pricing, poor reliability/speeds, bad customer service, or something else, relatively few people actually like their internet provider.  Presenting an alternative can only benefit them with better prices, better service, faster speeds, etc.

Other than cost I actually like my rural provider. 24Mb up and down, and rock solid. It took an ice storm plus generator self destructing to have an outage. Of course it's a locally owned and operated telco. Not a faceless national media conglomerate that worries more about bottom line than service. I'm near the edge of their service area. Farms just 6 miles east of me don't even have DSL. It's dialup, 3G, or satellite only for them. I know of a few point to point WIFI links that cross the line.

[magnemoe]

The supply of bandwidth is the same everywhere on Earth (minus the poles), ..

Is that true?
Isn't the supply of bandwidth greater closer to the max of the orbital latitude (~52°) and lower around the equator?

Yes, that's true.  But there will still be a lot of bandwidth available around the equator.

bandwidth will be kind of fixed, yes they will have satellites over low population areas support neighbor areas in more demand who will help coastlines but its still fixed.
And yes one of their customers is mobile phone providers as it makes off the grind cellphone towers far more practical.

[freddo411]

Comcast has over 26 million internet subscribers right now, the majority of whom hate it with a passion.

I would take the bet that at least 10% of the subscriber base would be willing to switch for a service that charged 30% more than what they are paying Comcast. That by itself represents over 2 billion dollars of gross annual revenue.

And yet Comcast added a quarter million wireless subscribers in Q4 and passed 2 million total wireless subscribers milestone.

I'm not sure what your point is.   Comcast is bundling mobile phone type service when it sells crappy, monopoly internet ... it's a sales strategy.   It doesn't indicate that people like Comcast.

[freddo411]

... snip ...

I think the economic discussion above gives some good reason to believe the economics work out for poor countries.  So the political question comes up -- will the governments in these countries stymie what should be an opportunity to greatly improve the lives of their people?  I think some will and some won't, but even having some let Starlink in to help their people will mean a lot of improvement in the lives of a lot of people.

Frankly, those in power tend to view their role as toll booth operator.   So if SL can funnel money to the rulers, it may be allowed to operate in $X country.   If there is an existing telecom operator that already funnels money to the ruler, this may impede SL's entry.

"Improving the lives of the people" is almost never seriously considered in these macro economic negotiations.

We will also hear SL talked about as "foreign colonialism" and such, as reasoning why local telecom monopolies should be maintained.

[vsatman]

Frankly, those in power tend to view their role as toll booth operator.   So if SL can funnel money to the rulers, it may be allowed to operate in $X country.   If there is an existing telecom operator that already funnels money to the ruler, this may impede SL's entry.

If we speak  about countries outside USA ana European Union  normally foreign satellite operator
have to
receive rights to use frequency in country X
receive license to sell telecom service in country X
found local company (sometimes with majority local companies as is now in USA)
hire local personal
pay local taxes ..
Here is one rule  about local market what locals say foreiners
"This is our cow and we will milk it!!!"
From this point of view Space X will sell capacity to local company for small money (one cents is better as nothing..)

[TheRadicalModerate]

bandwidth will be kind of fixed, yes they will have satellites over low population areas support neighbor areas in more demand who will help coastlines but its still fixed.
And yes one of their customers is mobile phone providers as it makes off the grind cellphone towers far more practical.

Presumably, a ground station will bind to a particular satellite and use it for as long as possible, before binding to a new satellite as the previous one drops out of range.  I would also presume that Starlink will have an algorithm for distributing load between however many satellites are in range of a given population of ground stations.  In equatorial regions, because of the geometry of the orbits, there will be fewer satellites in range to bind to, and they'll stay in range for less time.

So, even though the bandwidth of a single satellite is the same everywhere, the aggregate bandwidth of the system in an equatorial region is lower than the aggregate bandwidth in a mid-latitude region.  This will be true irrespective of how many satellites are deployed; there will always be fewer birds to load-balance across.

[TheRadicalModerate]

Definitely true, but keep in mind that a lot of backhaul in developing countries can be done via point-to-point wireless, so that's what you'd have to compete against. Seems like a lot of that equipment is now either solar or diesel powered for areas with intermittent electricity, which would be a requirement for Starlink as well of course.

Also do not underestimate the cellphone coverage already available in many of these regions - the push for mobile connectivity I was previously mentioning came a decade ago. Many of us probably don't live in developing regions so it's a good idea to double check our assumptions (I had to look up some numbers just now). In West and Sub-Saharan Africa, for example, a little under 50% of the population have mobile phones, and given that people care share phones for economically critical functions, it's not clear how strong the demand is for further coverage. Point I'm really making is that you're not competing with a market with no other solutions, just possibly more expensive solutions with less bandwidth, so even in the developing world there is a price point that needs to be matched.

Seems to me that the real issue here is the availability of end-user devices in the developing countries, so even if there are microwave or wired relay options available in an area that SpaceX wants to penetrate, it may still be worth SpaceX's while investing heavily in backhaul.  If all your customers have is a 2G, 3G, or LTE not-too-smart phone you have to be able to get them hooked up with that equipment.

I can see three business plays here:

1) SpaceX does backhaul deals with regional carriers and piggybacks off of their licensed spectrum.  The value proposition is simply that it reduces backhaul and infrastructure costs, allowing the provider to get into regions that were previously uneconomical to cover.

2) SpaceX buys low-power or mmWave spectrum and goes into the end-to-end microcell provider business.  Same biz case as #1, but now SpaceX is the vertical integrator (which is something that would no doubt appeal to Elon's engineering predilections).

3) SpaceX simply offers unlicensed nanocell access in addition to WiFi, and comes up with a billing scheme that allows very small communities a way to cost-share the price of a 10- to 20-ish meter tower with solar, battery, and an optional tiny gasoline- or diesel-powered generator.  This is very similar to #2 from an engineering standpoint, but it requires a lot more retail selling to get these communities to subscribe in the first place.

Note that the engineering for all of these are very similar.  The main difference is the marketing and regulatory nonsense required to deploy.  But all of these are better models in developing countries than trying to do deployment models that presuppose WiFi to get to the end devices.

[thirtyone]

Has anyone heard even a peep about SpaceX establishing any sort of partnerships with any service providers anywhere? It'd seem like the most efficient way to work in some developing markets is to use Starlink as a cellular backhaul, but it would seem a bit difficult to do that without partnerships or licensing of any sort in any region. I know most of Elon's companies operate with a philosophy of using as few middlemen as possible, but it seems maybe a bit excessive to try to also have to spend a good amount of resources managing local wireless equipment worldwide as well.

To be honest, part of me suspects that is *exactly* what they're planning on doing. But maybe there are just a lot of agreements going on behind the scenes.

OneWeb is, at least publicly, quite a bit more ahead - they have a few agreements with telecom companies for service in Africa, for example, and are apparently even partially owned by the government of Rwanda (they are an investor).

This is partly why I feel this topic is worth expanding into "markets for megaconstellations." Despite the relatively intense focus on Starlink, the problems (astronomy and space debris) and benefits (worldwide broadband coverage) largely apply to all megaconstellations currently being developed. Just a reminder - OneWeb is launching their first "production" batch of satellites in less than two weeks, so they're hardly a paper megaconstellation, like some of the other proposals.

[DigitalMan]

Has anyone heard even a peep about SpaceX establishing any sort of partnerships with any service providers anywhere? It'd seem like the most efficient way to work in some developing markets is to use Starlink as a cellular backhaul, but it would seem a bit difficult to do that without partnerships or licensing of any sort in any region. I know most of Elon's companies operate with a philosophy of using as few middlemen as possible, but it seems maybe a bit excessive to try to also have to spend a good amount of resources managing local wireless equipment worldwide as well.

To be honest, part of me suspects that is *exactly* what they're planning on doing. But maybe there are just a lot of agreements going on behind the scenes.

OneWeb is, at least publicly, quite a bit more ahead - they have a few agreements with telecom companies for service in Africa, for example, and are apparently even partially owned by the government of Rwanda (they are an investor).

This is partly why I feel this topic is worth expanding into "markets for megaconstellations." Despite the relatively intense focus on Starlink, the problems (astronomy and space debris) and benefits (worldwide broadband coverage) largely apply to all megaconstellations currently being developed. Just a reminder - OneWeb is launching their first "production" batch of satellites in less than two weeks, so they're hardly a paper megaconstellation, like some of the other proposals.

Isn't Google one of the investors?  It seems like Google could consume all their bandwidth for a while.

[TheRadicalModerate]

This is partly why I feel this topic is worth expanding into "markets for megaconstellations." Despite the relatively intense focus on Starlink, the problems (astronomy and space debris) and benefits (worldwide broadband coverage) largely apply to all megaconstellations currently being developed. Just a reminder - OneWeb is launching their first "production" batch of satellites in less than two weeks, so they're hardly a paper megaconstellation, like some of the other proposals.

Let's go down the list of high-runners for the ISP biz.  I've put these in what I think is rough most-to-least lucrative order for the satellite operators in the short term:

1) The ever-popular corporate VPN market.  Megaconstellations could be very cheap for far-flung enterprises with a lot of fan-out, and they're likely a lot easier to manage because they're so flat.

2) Aviation, maritime, long-haul trucking, and free-standing IoT.  This is something that's massively enabled with satellites, and quite possibly the market opportunity with the most pure growth.

3) Backhaul we've talked about.  As I said above, a system that can shave even $10K-20K off the capital investment for a small tower, or even marginal improvements in opex over microwave relays (which shouldn't be very hard) gets you into a lot of new markets.

4) Latency- and QoS-dependent apps.  The poster child here is high-frequency trading, where VLEO birds can handily beat trans- or inter-continental fiber systems, but this also includes telepresence, telesurgery, and other telerobotics applications.

5) Transit (i.e. edge-provider-to-edge-provider service).  This is usually an extremely high bandwidth business, and it'll be a while before Starlink or any other megaconstellation can compete here.  However, transit to weird, off-the-beaten-path places is hard to manage without over-investing in infrastructure, and it's really easy for a megaconstellation to soak up the overflow, which makes it easy to charge other transit providers a premium.

6) Content distribution.  A business where traffic engineering also occasionally goes wrong, and overflow capacity is nice to have at a premium.

7) Residential service, especially in rural and underdeveloped regions.  This has been talked about as the Starlink application, but I'm not putting this anywhere near the top of the list.

8) Live broadcast.  Hard to distinguish this from content distro with LEO birds.  This is sometimes extremely time-dependent, and a VLEO system may be hard to synchronize.  (You don't want the sports-betting people figuring out ways to use your system to past-post things...)  Anybody know how the GEO guys deal with this?

I'm sure I'm forgetting some biggies--and I'm also sure that hardly anybody will agree with my ordering.

PS:

9) 3.5) Military C³.  The switched phone system still has a facility called multi-level prioritization and preemption, which allowed the military to take over the private phone network to varying degrees.  I suspect that they paid a nice chunk of change for it.  The analogous internet technology puts super-high-priority diffserv markings on military packets and ensures that any other congestion gets dumped.  My guess is that that's worth a non-trivial amount of money to implement and to buy the option on the traffic.

PPS:

Per envy's comment below, I'd put military C³ at about 3.5 on the list.

[envy887]

9) Military C³.  The switched phone system still has a facility called multi-level prioritization and preemption, which allowed the military to take over the private phone network to varying degrees.  I suspect that they paid a nice chunk of change for it.  The analogous internet technology puts super-high-priority diffserv markings on military packets and ensures that any other congestion gets dumped.  My guess is that that's worth a non-trivial amount of money to implement and to buy the option on the traffic.

DoD applications are probably quite a ways up the lucrative-ness list. DoD is the only customer outside of SpaceX to have field tested the Starlink network - they have actually run multiple tests already, and those are just the ones we know about.

[TheRadicalModerate]

9) Military C³.  The switched phone system still has a facility called multi-level prioritization and preemption, which allowed the military to take over the private phone network to varying degrees.  I suspect that they paid a nice chunk of change for it.  The analogous internet technology puts super-high-priority diffserv markings on military packets and ensures that any other congestion gets dumped.  My guess is that that's worth a non-trivial amount of money to implement and to buy the option on the traffic.

DoD applications are probably quite a ways up the lucrative-ness list. DoD is the only customer outside of SpaceX to have field tested the Starlink network - they have actually run multiple tests already, and those are just the ones we know about.

Fair point--I kinda fell into the MLPP rat-hole. Just plain ol' vanilla internet access for the military is a non-trivial chunk of bandwidth.

It would be interesting to understand DoD's QoS requirements.  For ordinary surveillance and command interfaces, it's not a huge deal.  For things like controlling drones in real time, or for managing battlefield robots, it is.

I'd put the military at about 3.5 on my list.

[Tulse]

Is the US military going to be satisfied with running combat-related bandwidth on a civilian-owned network?  I can see using it for basic peacetime comms, but during a war I would think they would want their own dedicated constellation.

[TheRadicalModerate]

Is the US military going to be satisfied with running combat-related bandwidth on a civilian-owned network?  I can see using it for basic peacetime comms, but during a war I would think they would want their own dedicated constellation.

If they can get the bandwidth they want, when they want it, I'd guess that they'd be fine using a commercial megaconstellation.  No doubt the cybersecurity folks would want to go through it with a fine-tooth comb, but it's not like SpaceX or other operators are going to be slackers in this area.

The big ticket item for combat ops is that the bandwidth is absolutely there when you need it.  That may mean that the DoD will want to put their operators in the NOC with the commercial operators, and indeed have something MLPP-like in place in the constellation to preempt whatever would get in the way of the QoS they need.  That's stuff for which the operator can charge through the nose.

Current thinking in the Space Force (jeez, I've never used that in a sentence before...) is that lots of small and cheap is now better than a little bit of large and expensive.  Megaconstellations fill the bill there, but they're insanely expensive if they're dedicated only to military traffic.

[groknull]

Is the US military going to be satisfied with running combat-related bandwidth on a civilian-owned network?  I can see using it for basic peacetime comms, but during a war I would think they would want their own dedicated constellation.

The Air Force is trying to figure that out:
https://spacenews.com/air-force-enthusiastic-about-commercial-leo-broadband-after-successful-tests/
https://www.airforcemag.com/Global-Lightning-SATCOM-Project-Expanding-to-AC-130-KC-135/
...and many other articles.  Search on "Defense Experimentation Using the Commercial Space Internet (DEUCSI)", "Greg Spanjers", “Global Lightning”, or "Air Force Research Laboratory commercial satellites",

Path-agnostic communications is one area of focus, allowing reliable high bandwidth and availability and graceful degradation rather than failure.

The Air Force is concerned about the vulnerability of the existing dedicated military satellites:
https://spacenews.com/stratcom-chief-hyten-i-will-not-support-buying-big-satellites-that-make-juicy-targets/

[ChrisWilson68]

Is the US military going to be satisfied with running combat-related bandwidth on a civilian-owned network?  I can see using it for basic peacetime comms, but during a war I would think they would want their own dedicated constellation.

The military tends to like to own everything.  But in the case of Starlink it really would be irrational to refuse to use Starlink because of that.

There's no realistic downside to using Starlink.  It's a US owned and operated company.  Of course Starlink isn't going to cut off the military in time of war.  Even if they wanted to, the government could simply take over control of Starlink if they refused to cooperate voluntarily.  Starlink is going to be pretty secure to begin with, and the government can just add its own encryption layer on top of Starlink if it wants.

On the contrary, Starlink's thousands of satellites address the biggest issue with all other satellite communications in case of a war with an advanced adversary, such as China or Russia: a few satellites that could be destroyed.  It would be much, much harder to take out Starlink than to take out all other Western satellites put together.

Starlink is a best-possible case scenario for the Pentagon -- a US-controlled, world-wide communication system that is virtually impossible to destroy, all with no investment and no development risk for the government.  A private company is building it and just showing up with it ready to go.  It doesn't get any better than that.

[TheRadicalModerate]

There's no realistic downside to using Starlink.  It's a US owned and operated company.  Of course Starlink isn't going to cut off the military in time of war.  Even if they wanted to, the government could simply take over control of Starlink if they refused to cooperate voluntarily.  Starlink is going to be pretty secure to begin with, and the government can just add its own encryption layer on top of Starlink if it wants.

Let's be a little careful here.  While the massive redundancy in Starlink (or any other megaconstellation) makes it virtually impossible to shoot down, there are some architectural qualities that need to be carefully dealt with.

One of the major ways of securing a military network is by controlling access to its edge:  If you don't have the groundstations and keys to get into the satellites, you don't have any access at all.  That dramatically reduces the number of attack surfaces you have to monitor and secure.

You can define a secure edge in GEO satellite systems, because the space segment is just a bunch of bent pipes back to the ground station.  That means that if somebody gets control of one of your birds, they still don't have access to the core of your network unless they can hack the ground station.

With Starlink, you have neither of these security options at your disposal.  There are going to be millions of ground stations present, and it'll be almost impossible to know which of them are in the hands of the enemy.  Beyond that, if the enemy does manage to take over a bird, it's by definition in the core of your network, because each bird is simultaneously an access router and a piece of the core routing fabric.  If the enemy can start propagating bad routing tables from a compromised bird, they have a way of initiating a systemic attack.

These are fixable problems, as long as they don't need hardware support that isn't already on the birds.  For example, the military could request (and pay for!) a facility that allowed the entire system to be shut down except to a white list of military ground stations.  That makes it much, much more difficult for an attacker to gain access.  (Of course, you want to make real sure that the enemy can't shut the system down for anything other than its white list, but that seems doable...)

The access problem is likely fixable with post-launch software, but protecting the routing fabric is considerably more problematic without hardware.  You effectively have to firewall the ingress/egress systems from the core routing systems.  That's likely how the hardware has designed, with the routing systems being attached to either the inter-bird lasers or the ground-bounce systems.  But it's pretty hard to fix a hardware architectural mistake after the fact.

Perhaps the military will choose to run stuff in ground-bounce mode, simply because it makes the space segment more immune to systemic attacks?

[ChrisWilson68]

The military needs to connect hundreds of thousands of end users to one another.  Even if they were using a system built exclusively for the military, it's not realistic to assume an enemy will never get its hands on one of the end-user terminals.

Control of Starlink needs to be secured whether it's being used by the military or not.  Plenty of hackers will try to find security flaws in it.  In fact, this might be a security advantage because security flaws are more likely to be found and fixed before a war than they would be in a military-only system.

Musk is pretty tech savvy.  I'd trust him to make sure Starlink has a good security design more than I'd trust any big defense contractor.  And since Starlink is already being tested by the military, I'm sure all the security details have been shared with them.  If there were any security weaknesses the NSA or anyone else in the national security side of the government could find, I'm sure they'd tell SpaceX about them and they would be fixed.

The only security-related concern I'd have with Starlink would be physical security around the people and infrastructure that has access to control of the Starlink satellites.  Maybe the government can help here.  But, it is something that would be a legitimate concern, and maybe less secure than if it were owned by the government.

Still, it's not like the military is going to shut down all their other communication systems.  Starlink will be only one communication system used by the Pentagon, so there will be redundancy in case Starlink is taken off line.

[TheRadicalModerate]

The military needs to connect hundreds of thousands of end users to one another.  Even if they were using a system built exclusively for the military, it's not realistic to assume an enemy will never get its hands on one of the end-user terminals.

That's true, but it does point to the need to be able to both whitelist and blacklist terminals.  All networks will have to sustain attacks.  But it's a different story if they have to sustain massive attacks, from unknown vectors.

Control of Starlink needs to be secured whether it's being used by the military or not.  Plenty of hackers will try to find security flaws in it.  In fact, this might be a security advantage because security flaws are more likely to be found and fixed before a war than they would be in a military-only system.

Certain architectures are more vulnerable to attacks than others.  Balancing the operational requirements of a network with its security requirements is non-trivial, and dependent on what applications will run on the network.  For Starlink, a flat, integrated access/core architecture is cheap and extremely flexible.  But that flexibility comes at a certain security cost.

Musk is pretty tech savvy.  I'd trust him to make sure Starlink has a good security design more than I'd trust any big defense contractor.  And since Starlink is already being tested by the military, I'm sure all the security details have been shared with them.  If there were any security weaknesses the NSA or anyone else in the national security side of the government could find, I'm sure they'd tell SpaceX about them and they would be fixed.

Again, architecture matters.  Starlink is first and foremost a high-scale commercial network, and its architecture reflects that.  You can be as tech-savvy as you want, but design choices come with tradeoffs.

I don't want to give the impression that I think that Starlink is unsuited for military use, or that the military should be using purpose-built networks instead of trying to piggyback on commercial infrastructure.  Starlink provides massive value, and solves a major concern of space-segment networks literally for free, as far as DoD is concerned.

But I do think that the problem isn't as simple as you're making it out to be.  Your point about there being backup systems is well-taken, but part of bidding for heavy DoD work in this area is to be able to figure out how (and whether) the backup systems can take the load if there's a compromise of the the Starlink system.

[RedLineTrain]

Is the US military going to be satisfied with running combat-related bandwidth on a civilian-owned network?  I can see using it for basic peacetime comms, but during a war I would think they would want their own dedicated constellation.

As I understand it, the US military currently takes bandwidth on GEO sats, even those that are not US-owned.  So it shouldn't be a problem for Starlink.  But I'm sure SpaceX would build and launch a dedicated constellation for the military, if the military made it worth its while.