-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
doghousePVD
My grandfather’s car
That would be great! I have one on the way, arriving perhaps Monday. Can’t wait to see if it’s round or rectangular.
I am suspecting SpaceX's market will be severely limited by the number of terminals it can support in a cell, which I remember reading somewhere is in the low thousands per cell. So if you add up all the cells it can serve across the world, we are talking about a few million subscribers? Or am I totally off with those numbers?
I believe the users are limited by two things, one is the number of satellites and the other is how many ground stations are in a particular cell. As they put up more satellites and bring online more ground stations, the amount of terminals that can be supported will go up. I have no idea what the current limiting factors are and it might be different in each cell I suppose. I signed up the day it became available and am still waiting... still says "late 2021".
Contrary to popular belief, Elon isn’t an idiot. SpaceX has done the math and know they can run a profitable Internet company.
Now that you have cleared that up, what is the practical max no. of subscribers Starlink can support across the world?
I am so excited to get Starlink, and about 5 more houses where I live would also follow up once I told them it was good or not.
All I want for Christmas is a Dishy, cmon Santa Elon!
All I want for Christmas is a Dishy, cmon Santa Elon!
Sorry for my snippy reply. It’s just that I get tired of people second guessing the economics of Starlink, not the least anyone from Viasat
I wrote about this elsewhere, but each Starlink sat has about 16 Gbps of downlink (customer) bandwidth. Due to statistical effects, an ISP usually needs about 3 Mbps to 5 Mbps per customer backhaul capacity. So, each Starlink sat can accommodate about 5,000 customers at any one time. So now we get into really sketchy numbers of utilization. A satellite traveling over the Pacific Ocean won’t have 5,000 customers talking to it right now. BUT as Starlink starts signing up airlines, shipping companies, and pleasure boaters, they may actually hit that utilization number even over the oceans. Remember that a single airplane is equivalent to 300 customers or so. And don’t forget military contracts. Starlink happens to be a VERY resilient network, much better than a GEO sat once the missiles start flying.
So, eventually (like five years out), I do see each satellite looking after, say 4,000 customer or customer equivalents throughout the world. With an eventual 40,000 satellite constellation, that works out to be 160M customers. Times $100/month works out to an eventual revenue of $192B per year.
Oops. That’s too much. SpaceX itself have estimated that Starlink revenues will be in the $30B range. So knock down whatever numbers you want by a total factor of 6.
Btw, SpaceX total steady state revenue today is around $3B to $4B. So Starlink is expected to be 10X current size of SpaceX launch revenues. At least until Starship gets going...
I was just logging in to post the same...more waiting.Just got an email from Starlink - got super excited!
Email says thanks for your support, check your account for availability...
Login to account, estimate has moved from mid/late 2021 to mid 2022
doghousePVD
My grandfather’s car
ucmndd
Well-Known Member
Also got the mid-2022 bump. 
I’ve been looking forward to completing the Elon self-sustainability kit… car + solar + Powerwalls + Starlink so I can ride out the interminable power outages here. When the grid goes down our dinky rural cable ISP goes down with it, so Starlink is the way…
I’ve been looking forward to completing the Elon self-sustainability kit… car + solar + Powerwalls + Starlink so I can ride out the interminable power outages here. When the grid goes down our dinky rural cable ISP goes down with it, so Starlink is the way…
That actually brings up a good question.Also got the mid-2022 bump.
I’ve been looking forward to completing the Elon self-sustainability kit… car + solar + Powerwalls + Starlink so I can ride out the interminable power outages here. When the grid goes down our dinky rural cable ISP goes down with it, so Starlink is the way…
How hardened are the current StartLink gateway locations? Solar? PowerWalls (or larger batteries), I presume they at least are on a Fiber Loop vs a collapsed loop or leg, but probably not at the point of redundant fiber at most of the early gateways (yet?)?
-Harry
They all have standard generator backup. They are often at long haul fiber repeater sites in the middle of nowhere. So they have at least two different paths to the Internet typically.
Sort of. While maybe there was/is some still-in-buildout limitation to the number of terminals in a cell (I hadn't heard that, but I suppose its possible?), there's really no practical limit to number of UTs.
I've waxed on this upthread, but the ultimate bounding condition is how much power the constellation can put on any point on the ground, known as power flux density (PFD). This is a level that's regulated by the FCC/ITU, and then...more or less anyway...Starlink will need approval from local regulators in all of the countries they wish to operate. When combined with the comms protocol implemented on the user links and the footprint of a beam (= the cell), this regulated power limit essentially sets the maximum data rate within an area that must be shared by all the users in that area.
So then, the real limit there is what kind of service Starlink chooses to offer their users. More users, less service...less users, more service. This is actually quite independent of the number of sats—at least in context of 'max data' to any one area on the ground, having more sats doesn’t really do anything once that max rate can be reached reliably and consistently over the year with a given sized constellation (a year is an important metric as it cycles each sat/orbit through their worst case configurations).
Kinda. Ish.
The big thing with each satellite is duty cycle. Each satellite can certainly provide full (PFD limit) power in any one beam; what they can't do is blast every beam with full power all the time. Intuitively this should make sense--a satellite that could blast its full coverage area with max power over an entire orbit (let alone an entire year) would have an impractically overbuilt EPS (and thermal system) and would spend most of its time wasting that overbuilt-edness over the oceans.
What ends up happening is that the satellite's power system is sized such that the total power available on the sat will basically "hop" around each of the couple-thousand beams (on the order of many microseconds), providing more power and/or dwelling longer on the high traffic cells, and throttling back on the low traffic cells...but at a macro level, at some much-less-than-100% duty cycle interval.
When you add sats to the constellation, what ends up happening is each sat effectively has to service fewer cells, and so the sum of those sats can service the same number of cells with longer dwells and more power. As a thought experiment for the above concepts, if each sat can form 2000 beams, a sat flying over an area with no redundant coverage (from another sat) would have to split its power across those 2000 beams. Some hot spots will certainly be close to full dwell at the PFD and some will be colder or even unoccupied (and so not lit up at all), but ultimately the limit to service at this point in the thought experiment is how much power the sat can push to each of the beams that need it. One way to increase service would be to make the power systems larger--more solar arrays and bigger batteries means that sat can dwell on more beams for a longer period of time. But that concept starts to become inefficient rather quickly. What SpaceX has decided to do is just put more sats over that same 2000 beam area (more or less--obviously they don't fly right on top of each other). So now there's effectively 4 sats worth of power (or whatever) that has to serve those 2000 beams. There's a lot of upside to this, from production efficiency (more units = lower per unit cost) all the way to constellation redundancy (if one sat goes out when there's no other sats, the 2000 cells are screwed, if one sat out of 4 goes out, there's only a small gross reduction in the max available service).
But again what more sats can NOT do is increase the upper limit cap in any one cell, which is again defined by PFD. Short of getting more PFD approved by the regulators, the only real solution is bigger satellites with more antenna elements and, most importantly, the ability to form smaller beams. Those smaller beams still won't increase PFD but they do allow for more efficient partitioning of the otherwise bigger cell, and, at least typically, enables efficient use of available spectrum.
The ground stations are pretty irrelevant with a ~steady state space infrastructure though, similar to the supercharger network buildout, supply vs demand may not be fully aligned during buildout. The limiting factor here is primarily the fact that because each sat's gateway link is a MASSIVE amount of data (effectively, the sum of the all of its user data links) the ground antennas need to be parabolic dishes pointed at (and tracking) a single sat. One could imagine in a buildout phase that there may be a surplus of sats relative to ground antennas, and so those sats are effectively useless, at least at that instant.
The location of the ground antenna also does not need to be in a cell, it needs to be in the footprint of the entire sat. The ground stations can be many hundreds of miles apart and as the satellite flies across them, each ground antenna just 'hands off' the sat to the next ground antenna at some point in time.
The problem is that it it really has nothing to do with the number of satellites. First, the theoretical limit to the infrastructure is, at least within reason, pretty large. That limit is all about 1) PFD and 2) the sum of service Starlink wishes to offer its subscribers in any particular area.
But--and this was hashed out upthread too--the real practical limit to number of subscribers is a function of 1) PFD and 2) people that don't have superior terrestrial service (where superior is a broad brush), 3) people that will be offered service (so, not Jai-Nuh) and 4) people with enough wealth to pay for service.
As noted upthread, there's basically 10M subscribers in the US that have historically paid for satellite internet. So, as a first order analysis, one can figure that Starlink's cap in the US would be 10M subscribers, or ~$12B. It is of course impractical to assume Starlink will be able to capture that market share long term, given the perpetual expansion of terrestrial+MNO service and the inevitable competition from other eventual constellations (Kuiper, notably, since they're a company that doesn't need their satellite service to be a direct profit center). I'd speculate that ~5M subscribers is about the practical cap in the US, or ~$6B revenue.
For various reasons its also quite fair to assume that the majority of subscribers/revenue will come from the US, so we're likely looking at less than 10M global subscribers, so let's call it a round $10B revenue. That's some big bucks for sure, but far from what more dreamy-eyed folks would rather see.
One all needs to add upside from opening up new product families, primarily satellite backhaul on terrestrial cell towers, and maybe global esports expansion (= global low latency). I honestly don't know much about the latter beyond people telling me that esports is apparently 1) A Thing and 2) going to be A Huge Thing, but the former will end up being a bit of a deal with the devil--the more coverage people can get from their existing connectivity resources, the less they'll need from Starlink....
Thanks much for the most informative post. We could really use starlink, our options are terrible and we, sadly, did not get one in the initial beta but I was signed up from the get go. Getting to be that my son must stay up to midnight to get homework submitted because our bandwidth is so poor before then (cell phone data). SIGH
How is that assumption 'quite fair'?
Ignoring the competition is coming argument:
US: 5 million out of 330 million people (FCC pegs unserved at 5.2 million)
Rest of world: 5 million out of 7.5 billion people
You could make a claim the monthly cost may be adjusted in developing regions, but a 0.07% take rate seems pessimistic. Plan for India next year is 200k terminals.
Then there are military, airline, cruise ship, and back haul service revenue streams.
Current satellite users is not a valid figure to base numbers off of beyond setting the minimum addressable market size. DSL and cellular are both better than current geostationary satellite service but worse than Starlink.
SSonnentag
埃隆•馬斯克
My account now says February 2022. I‘m anxious to get some decent internet service in my rural location. 
