Is SpaceX overvalued?

[Tywin]

How could a plunge in valuation affect SpaceX?


https://twitter.com/TMFAssociates/status/1527686900059779077

[Tywin]

https://twitter.com/LionnetPierre/status/1530917766928244737

Interesting tweet...

[RedLineTrain]

TSLAQ is strong with these ones.  Avoid.

[Swedish chef]

TSLAQ is strong with these ones.  Avoid.

In principal i agree but I would like to argue that this person don't belong to the TSLAQ-crowd, this is an entirety different beast. His job is to preserve the status quo for the European Space Industry and judging from his tweets he does a dang good job.

[su27k]

No

[meekGee]

What price a planet?

[spacenut]

SpaceX includes Starlink.  They are already starting to make money off of Starlink.  They also launch a lot of commercial satellites.  They supply the ISS as well as crew to and from ISS.  They have contracts for a Lunar Starship as well as private flights.  They are on par to launch around 60 Falcon 9's/Falcon Heavies this year.  They may be undervalued. 

[DanClemmensen]

SpaceX Dominates the launch industry. They launch more than two thirds of the world's total payload mass to orbit. Falcon Heavy Is the most capable active heavy lift launcher. All of this is without considering Starship or Starlink.

[M.E.T.]

So the FUD has now subtly moved on from “Starship is not a real thing”, to “Can they launch enough Starships with just two launch pads”.

Once they have four launchpads, expect it to move seamlessly to “The Next Thing”.

[RedLineTrain]

TSLAQ is strong with these ones.  Avoid.

In principal i agree but I would like to argue that this person don't belong to the TSLAQ-crowd, this is an entirety different beast. His job is to preserve the status quo for the European Space Industry and judging from his tweets he does a dang good job.

He likes TSLAQ tweets, he uses TSLAQ thought patterns.  Walks like a duck, talks like a duck...

[JayWee]

Even retweeted our dear friend, the one and only, ESG Hound.

[DanClemmensen]

So the FUD has now subtly moved on from “Starship is not a real thing”, to “Can they launch enough Starships with just two launch pads”.

Once they have four launchpads, expect it to move seamlessly to “The Next Thing”.

This is especially ludicrous when you recall that SpaceX's goal is multiple launches per day from each launch pad. Even with "only" one launch per day, you get 365/yr, which is more than twice all the world's current yearly launches. Starship will average more that 20 times the average payload of all the world's launches, so at least 40 times all the world's current payload mass to orbit and potentially more than 200 times, all from a single launch pad. You only need the second launch pad for redundancy.

It would be more reasonable to ask where the customers are for this capacity. Just to calibrate, F9 launches about 2/3 of the world's current payload mass, and more than half of this is Starlink, at a current rate of < 30 Starlink launches/yr to launch V1.5 Starlinks. But it would take < 4 Starships to launch this same tonnage, or < 12 Starships to launch the same number of  V2.0 Starlinks.

Again to calibrate: one launch tower needs a total of one SH to maintain a launch cadence of two launches per day. Let's be generous and make that three SH per tower to account for redundancy and periodic maintenance.  You need more SS because each mission is at least one day from launch to landing: let's again be generous and use ten SS to maintain a one-a-day launch cadence.

Conclusion: capacity is not the problem. Finding enough customers is the problem.

[JayWee]

So the FUD has now subtly moved on from “Starship is not a real thing”, to “Can they launch enough Starships with just two launch pads”.

Once they have four launchpads, expect it to move seamlessly to “The Next Thing”.

It would be more reasonable to ask where the customers are for this capacity. Just to calibrate, F9 launches about 2/3 of the world's current payload mass, and more than half of this is Starlink, at a current rate of < 30 Starlink launches/yr to launch V1.5 Starlinks. But it would take < 4 Starships to launch this same tonnage, or < 12 Starships to launch the same number of  V2.0 Starlinks.

Have you missed the news that Starlink v2 is 1250kg and 7m long and that the constellation is planned to be 30k strong? That means about 300 Starship launches. Then add replacement sats.

[DanClemmensen]

So the FUD has now subtly moved on from “Starship is not a real thing”, to “Can they launch enough Starships with just two launch pads”.

Once they have four launchpads, expect it to move seamlessly to “The Next Thing”.

It would be more reasonable to ask where the customers are for this capacity. Just to calibrate, F9 launches about 2/3 of the world's current payload mass, and more than half of this is Starlink, at a current rate of < 30 Starlink launches/yr to launch V1.5 Starlinks. But it would take < 4 Starships to launch this same tonnage, or < 12 Starships to launch the same number of  V2.0 Starlinks.

Have you missed the news that Starlink v2 is 1250kg and 7m long and that the constellation is planned to be 30k strong? That means about 300 Starship launches. Then add replacement sats.

No, I had not missed that news. I accounted for the 3x mass increase by tripling the number of launches all the way up to 12. I was trying to relate the Starship launch rate to the world's existing launch rate to show that a single launch tower provides about 40 times more capacity than the world currently uses even at only one launch per day. To add those 300 launches, SpaceX can increase the launch cadence from one launch per day to two launches per day for ten months. Replacement: for a 5-year average lifetime you need 60 launches/yr.

[oldAtlas_Eguy]

I took a look at the subscriber growth patterns for Starlink and projected that with the probable Gen2 sat launch cadences over that same period. I also went conservative with my numbers and also included significant per subscriber price reductions to maintain subscribers and to grow the numbers due to the growing competition.

The answer for the amount of revenue generated by Starlink over the ten year period from 2023 to 2032 was $160B. The usual is that valuations generally based on the revenue produced in the next 10 years. But there is also considerations given to cash flow (profit margin percentages) on average for that same period. Another considerations is the rate of growth of the company size. Starlink and all the rest of SpaceX is currently showing signs of a revenue and expansion rapid growth pattern like Amazon over its first 20/30 years of existence. Because of this and that the slowdown in this expansion increasing slope expansion rate will not  likely ease until the mid 2030's  Its actual value to an investor is much higher than the basic generalization of value.

Such that investors that have invested or about to invest into SpaceX are not likely to cash out for at least 10 years if not as much as 20 years. Just look at how stubbornly the investors are holding onto their investments into SpaceX and not cashing out. In most cases they are even investing more capital. We are also not talking small investments but investment amounts of a $B by a single investor.

All of this together shows that a valuation of $225B is very undervalued for this company. But because of it being a space business and is a high risk leader in setting trends investors are being cautious. Otherwise you could see a a valuation of as high as $400B or even more. Risks cause discounts of the possible valuation into lower investor price points in line with the amount of risk they are willing to take.

If Starlink continues its expansion. Starship starts delivery of Gen2 Starlink sats. More consolidated large customers for various launch, satellite construction and communications services by governments (foreign and domestic) and defense organizations (domestic and western aligned nations). That by end of 2023 with all these things to some degree come to pass The valuation at the end of 2023 will have a very large jump. Because most of the risks will have disappeared. A valuation of significantly >$300B.

[ZachF]

The original tweet being quoted is especially funny because Tesla actually does print cash now and generate large profit (about $10k per car sold) with a large growth rate, while Amazon’s cash flow has actually been negative…

[Widford]

SpaceX currently charges 1.1m minimum for a cubesat launch on a rideshare mission. Does this mean that if starship's 2m per launch goal is achieved, they could break even by launching just 2 cubesats?

[DanClemmensen]

SpaceX currently charges 1.1m minimum for a cubesat launch on a rideshare mission. Does this mean that if starship's 2m per launch goal is achieved, they could break even by launching just 2 cubesats?

The $2 million was the cost of propellant at that time, pretty much, and energy prices, which are highly volatile, were very low. The cost of LNG has quadrupled since then, and we can guess that the cost of energy to extract LOX has gone up a lot since then also. However, The prices will go back down eventually, and will become a lot less volatile as renewables deliver an increasing percentage of the supply.

[marcus79]

Did not know what TSLAQ was. I find it a bit strange and disconcerting how today people's evaluation of the facts tends to follow their preferences in one flavour or the other.

Personally, I do not like Musk as a public persona at all, but I have a lot of respect for SpaceX and its achievements. They really have been a game changer in providing affordable, high-volume medium to heavy lift.

However, I think the commercial case for super heavy lift has yet to be made. At present the commercial fate of Starship is tied to the commercial performance of Starlink. Vice versa the commercial performance of Starlink is tied to the technical performance of Starship. This is fundamentally different from the Falcon series, where there was an established market demand for medium to heavy lift. So the valuation is more like a potentiality, where for investors a lot depends on the confidence they have in Musk and the willingness to take a risk. So there are three kinds of risk: Starship's technical performance, Starlink's commercial performance, and investor confidence and ability to invest.

The only other business for super heavy lift of course is the government's human exploration program. Low flight rates, a limited budget, and political uncertainty/meddling have historically made it hard to sustain super heavy lift. Let alone two heavy lifters, one extremely inefficient heritage design, one innovative risky design. I see a lot of risk in Artemis, too. I don't think it could bail out Starship if Starlink is not commercially viable.

The inherent problem is that super heavy lift requires (super) heavy fixed costs. Someone needs to pick up the tab for that. Of course with a high flight rate this can be driven down, with full reusability, but this needs customers, a lot of them. Super heavy lift never had real commercial customers. The Soviets tried hard with Energia, but it was just pipe dreams and a big comsat they tried to develop in vain in the waning days of Gorbachev. If that comsat had been so potentially profitable, I am sure it would not have been discontinued.

If you took the comsats from Falcon 9 and put them in Starship, you run up to the fact that it is not optimised for it. Not sure if one Starship can replace five Falcon 9 flights, then it might (barely) work. But probably a design tailored to such payloads is going to be more economical and competitive.

Therefore the valuation of the company, in my view, is tied to one's view as to whether achieving the potentiality of Starlink is possible. Perhaps it goes even deeper and depends on one's view whether Musk can pull off something like this. Maybe the valuation depends, as it did with other tech companies, on his aura.

Personally, I wish they'd focused on achieving full reusability first for medium lift, with an expendable stage for heavy lift.

[dondar]

Re-usability "costs".  It costs not only in money but what is more important it is expensive in weight. (see tyranny of the rocket equation).
The important part is that most of these costs are "fixed", that is they scale "badly".  I.e. if you scale up the incorporation of re-usability becomes easier and easier from engineering POV. Musk hit a jackpot with the design of Starship and the star-ships will be the "last chemical rockets" to fly. I am not exaggerating.

Your opinion on Musk is formed by "Teslaq" in very big part.
 Generally it is expected that somebody collects info and then this somebody forms an opinion. If you would did that, you would find what all of us already know that the factual load of the "teslaq" crowd is ZERO. i.e. the mere attention to what these people say is waste of time.

About "market", "load" etc.

The investments can have different purposes and "effective management/accounting" has very limited usability as evaluation mechanism.  The value you see is the result of the available cash on the market (a lot) coupled with the lack of alternatives. (the number of engineering companies is pretty much zero. You have Musk companies, AMD and NVIDIA. And this is it). What is also important to remember about SpaceX is that the investments are "personal". Closed group of people invest as much as needed and formalize it in a way which is comfortable from legal POV. They don't intent to extract these money any time soon. It means   the "stock value" is important only for the employees and the limited secondary market which is  extremely oversubscribed.

From practical POV Starship sizes offer very open window for the actual human presence in space. The projects and movement in this direction will arise at the moment Starship goes orbital. As I was writing already the developmental delay is ~ 4 years.

[su27k]

SpaceX currently charges 1.1m minimum for a cubesat launch on a rideshare mission. Does this mean that if starship's 2m per launch goal is achieved, they could break even by launching just 2 cubesats?

Sort of. $2M is the target marginal launch cost of Starship, this would include propellant cost, plus the labor for preparing the fully reusable Starship for a new launch. So if they can achieve this low cost, they can sell an entire Starship launch at $2M and breakeven, or more likely they sell the Starship launch at whatever price customer can bear, and pocket the difference as profit.

Note however the current $1.1M price for their rideshare service is not for a single cubesat, it's for 200kg of payload to orbit, you can fit a lot of cubesats in 200kg.

[marcus79]

Expanding the human presence in space does not pay the bills.

Of the $2 million figure, $900,000 is fuel. Even if it is more expensive, that will unlikely be the problem. However, the $ 1.1 of the salaries and fixed costs assume a flight rate of 3 flights per day. That translates into a cool $ 1,205 million operating cost per year. for a single Starship. And this is hypothesized for a future scenario when 1,000 Starships are flying to sustain the colonisation of Mars. In the beginning the costs are likely to be a bit higher.

Near term, Musk says he wants to get to $ 10 million in 2-3 years. He needs the Starlink V2 flights to amortise the fixed operating costs to achieve that. Also a high flight rate is necessary to amortise the building costs of Starships. Unit costs for the Starship have been estimated at $ 200 - 250 million and more for the booster. I'm not even thinking about the development costs.

https://www.fool.com/investing/2020/02/16/are-elon-musks-spacex-promises-even-possible.aspx

https://www.quora.com/How-much-will-it-cost-to-build-1-starship

If the flight rate is less, the amortisation of fixed, unit and development costs is less.
So Starlink is absolutely essential to lower the per flight costs, giving it hundreds of flights.
If not, I cannot see it competing for traditional medium/heavy payloads. The competitors will have rockets tailored to this market, indeed SpaceX would be kind of its own best competitor with Falcon 9 and Heavy.

Maybe there will be new kinds of commercial payloads that require super heavy lift, but this will take time. The government will only require so many missions per year, partly because they will keep spending on SLS for a bit.

The same logic applies to the Long March 9 btw, even in its new reusable design. They keep trying to find missions for it like space solar power and so on.

I'm not saying Starship will not somehow manage to tread this dangerous waters, just that the valuation of SpaceX is pretty volatile and to some extent in the eye of the beholder.

[su27k]

Expanding the human presence in space does not pay the bills.

Of the $2 million figure, $900,000 is fuel. Even if it is more expensive, that will unlikely be the problem. However, the $ 1.1 of the salaries and fixed costs assume a flight rate of 3 flights per day. That translates into a cool $ 1,205 million operating cost per year. for a single Starship. And this is hypothesized for a future scenario when 1,000 Starships are flying to sustain the colonisation of Mars. In the beginning the costs are likely to be a bit higher.

The $2M is the marginal launch cost, it doesn't include fixed cost. To breakeven on a new launch, they only need the launch revenue to exceed the marginal launch cost.

To breakeven on their overall cashflow, they'll need total revenue to exceed total expenses, the latter includes fixed costs. But these numbers are much harder to estimate, since there're a lot of unknowns. But we do know the $ obligated to SpaceX from their government contracts in the past 12 months is over $1.9B, and they spent $430M at Boca Chica last year with 1,600 employees. So extrapolate the latter to 10,000 employees gives an annual expenditure of close to $3B, and they just need less than $1B per year from non-government contracts to break even.

I think it's safe to say that the estimate of $200M to $250M for each Starship is wildly off mark.

[marcus79]

Expanding the human presence in space does not pay the bills.

Of the $2 million figure, $900,000 is fuel. Even if it is more expensive, that will unlikely be the problem. However, the $ 1.1 of the salaries and fixed costs assume a flight rate of 3 flights per day. That translates into a cool $ 1,205 million operating cost per year. for a single Starship. And this is hypothesized for a future scenario when 1,000 Starships are flying to sustain the colonisation of Mars. In the beginning the costs are likely to be a bit higher.

The $2M is the marginal launch cost, it doesn't include fixed cost. To breakeven on a new launch, they only need the launch revenue to exceed the marginal launch cost.

To breakeven on their overall cashflow, they'll need total revenue to exceed total expenses, the latter includes fixed costs. But these numbers are much harder to estimate, since there're a lot of unknowns. But we do know the $ obligated to SpaceX from their government contracts in the past 12 months is over $1.9B, and they spent $430M at Boca Chica last year with 1,600 employees. So extrapolate the latter to 10,000 employees gives an annual expenditure of close to $3B, and they just need less than $1B per year from non-government contracts to break even.

I think it's safe to say that the estimate of $200M to $250M for each Starship is wildly off mark.

Ok, maybe the Motley Fool article was muddying the waters somewhat as it said the $ 2 million included salaries (fixed costs, unless they hire temp workers). They also said SN8 cost $ 216 million.

Anyway, the operational costs are the most important from the point of view of how much a flight costs (for SpaceX).

[JayWee]

Ok, maybe the Motley Fool article was muddying the waters somewhat as it said the $ 2 million included salaries (fixed costs, unless they hire temp workers). They also said SN8 cost $ 216 million.

Anyway, the operational costs are the most important from the point of view of how much a flight costs (for SpaceX).

From the fool article:

If you assume that Starship, with four times Falcon's payload, costs roughly four times as much to build, this implies that building a single Starship might cost roughly $216 million

That's a very bad assumption. Stainless steel is $4-$5/kg, Falcon9's AlLi alloy around $40/kg.
Additionally, they base their calculation on the "list price". Obviously, that's not the "build cost". Garbage In, Garbage Out.

[alugobi]

Golly, Batman, all these peeps saying they know how much SX stuff costs.

I don't know who to believe.

[DanClemmensen]

Ok, maybe the Motley Fool article was muddying the waters somewhat as it said the $ 2 million included salaries (fixed costs, unless they hire temp workers). They also said SN8 cost $ 216 million.

Anyway, the operational costs are the most important from the point of view of how much a flight costs (for SpaceX).

From the fool article:

If you assume that Starship, with four times Falcon's payload, costs roughly four times as much to build, this implies that building a single Starship might cost roughly $216 million

That's a very bad assumption. Stainless steel is $4-$5/kg, Falcon9's AlLi alloy around $40/kg.

Also, the "roughly four times" is suspect. This Falcon 9 block 5 can launch 22.8 tonne to LEO if expended, but that's after ten years of tweaking the design. A recoverable launch is only 16.25 tonne. Falcon 9 V1.0 was at 10.4 tonne. We don't know where Starship will end up, But the effective Starship payload will be between 100 and 150 tonne when is reaches the maturity level of Falcon 9 Block 5, and this is between six and nine times the payload for a recoverable F9 launch. Is he now going to multiply his $54 mission by nine to get $486 million?

<sarcasm>Gee, it's a big new rocket with twice the liftoff thrust of the SLS. Surely it will cost twice the $4 billion SLS manufacturing cost. </sarcasm>

One of the major innovations of Starship is a transition of manufacturing from the aerospace model to the shipyard model. His cost computation is laughable.

[alugobi]

Quote from: Motley Fool

    If you assume that Starship, with four times Falcon's payload, costs roughly four times as much to build, this implies that building a single Starship might cost roughly $216 million

Stick to subjects you understand, Fool. 

No, it doesn't imply that.  Sheesh.

[wannamoonbase]

So the FUD has now subtly moved on from “Starship is not a real thing”, to “Can they launch enough Starships with just two launch pads”.

Once they have four launchpads, expect it to move seamlessly to “The Next Thing”.

Once they have a dozen ships on Mars it could become, but a dozen isn’t enough ships.

SpaceX is not under valued, but I say this based on Starlink.  I think Starlink can be a $500B to $1T company.  Maybe more.   And it could throw off hundreds of billions in profits each year for use to settle Mars.

Starlink could be revolutionary

[JayWee]

I'm actually quite surprised at the subscriber growth. It's almost 100k/month now.

[meekGee]

I'm actually quite surprised at the subscriber growth. It's almost 100k/month now.

Got mine the other day.  It's as plug and play as it can ever get.

The service is so revolutionary, and the implementation is so flawless - it's going to be everywhere.

But SpaceX's valuation should be much higher than Starlink's.  They're going to be the ground floor investors in a new planet.  The upside is mind boggling.

[vapour_nudge]

What is the value of SpaceX if, God forbid, Elon Musk was no longer around or at the helm?

[alugobi]

In Shotwell we trust.

[dondar]

I'm actually quite surprised at the subscriber growth. It's almost 100k/month now.

"Exponential growth" is a fantasy in the real world. If you want to estimate future market you have to look for the uncovered market. There is considerate incoming market in south East Asia (Indonesia, Tailand, Vietnam where the major clients will be local goverments), South America and US. The growth beyond 500k is bound by the proper development of inter-sat links (software is a prime target, I am extremely curious how they will solve routing paths) and proper bandwidth-ed ground stations (apparently SpaceX has some issues in some areas of US already). Basically yet another year to wait for the proper Starlink growth.

[DanClemmensen]

I'm actually quite surprised at the subscriber growth. It's almost 100k/month now.

"Exponential growth" is a fantasy in the real world. If you want to estimate future market you have to look for the uncovered market. There is considerate incoming market in south East Asia (Indonesia, Tailand, Vietnam where the major clients will be local goverments), South America and US. The growth beyond 500k is bound by the proper development of inter-sat links (software is a prime target, I am extremely curious how they will solve routing paths) and proper bandwidth-ed ground stations (apparently SpaceX has some issues in some areas of US already). Basically yet another year to wait for the proper Starlink growth.

The two issues are related. You need ISL to shift the teleport traffic away from the congested teleports.

No in-space routing at the IP layer. Use frame forwarding at the frame layer using a specialized frame layer. Because ISL topology changes are known in advance table updates are not reactive. Instead, forwarding tables have scheduled updates what change with microsecond precision. Similarly, the user terminals switch between beams in a precisely-timed fashion, as do teleport-satellite links: all of this is handled by the specialized frame layer and its forwarding tables.  I never worked for SpaceX so I do not know that they are doing it this way, but I did do a preliminary system design as part of a business proposal for a different constellation in 2014.

[Robotbeat]

Starlink could easily get to a similar revenue level as Comcast. (And yes, SpaceX could bundle TV, etc, like other cable or satellite providers.)

Comcast is worth about $200B, has about $120 billion in revenue from 30 million subscribers. Yes, it can operate in dense cities better than Starlink, but… It operates in just one country.

I think Starlink alone could get to $200B, maybe more as the global economy grows. Maybe $500 billion to a trillion in a couple decades, along with the rest of SpaceX. Starlink v1.5 will double in count, v2 will increase that count by 10 and bandwidth by almost 10 per satellite. They have over 500,000 subscribers now and by the time v2 is filled out could handle 100x that, especially as permission coverage increases globally and they further upgrade Starlink.

So I don’t think SpaceX is overvalued. I think it’s about right ($125B-$150B?) given uncertainties and possible future competition (which the launch side of SpaceX could actually benefit from).

[ajmarco]

Starlink could easily get to a similar revenue level as Comcast. (And yes, SpaceX could bundle TV, etc, like other cable or satellite providers.)

Unless you are separating the telecom portion out from those numbers please be aware the Comcast owns NBC Universal. So as a whole company they are a cable provider, streaming service, production company, theme park operator, etc...

[SimonFD]

Surely the question is moot as SpaceX is not a publicly traded company and so it doesn't have a 'value' in the same way as Tesla does (because it IS publicly traded).

i.e. There is no 'value' for it to be over anything


edit for grammar and spelling

[Arcas]

Surely the question is moot as SpaceX is not a publicly traded company and so doesn't have a 'value' in the same way as Tesla does (because is IS publicly traded).

i.e. There is no 'value' for it to be over anything

Every time they sell stock, they have a "value". What's the difference between the "value" of a private company and that of a public company other than the number of shareholders?

[JayWee]

What's the difference between the "value" of a private company and that of a public company other than the number of shareholders?

Speculation.

[Robotbeat]

SpaceX has to regularly sell stock to allow employees, who receive stock options as part of their compensation, to cash out. This is where headline valuations come from. It’s not part of the public stock market, you have to be an “accredited investor” to have even a shot at buying stock in SpaceX.

[Robotbeat]

Starlink could easily get to a similar revenue level as Comcast. (And yes, SpaceX could bundle TV, etc, like other cable or satellite providers.)

Unless you are separating the telecom portion out from those numbers please be aware the Comcast owns NBC Universal. So as a whole company they are a cable provider, streaming service, production company, theme park operator, etc...

That’s a very good point, but they still get most of their revenue from the telecom side.

[dondar]

I'm actually quite surprised at the subscriber growth. It's almost 100k/month now.

"Exponential growth" is a fantasy in the real world. If you want to estimate future market you have to look for the uncovered market. There is considerate incoming market in south East Asia (Indonesia, Tailand, Vietnam where the major clients will be local goverments), South America and US. The growth beyond 500k is bound by the proper development of inter-sat links (software is a prime target, I am extremely curious how they will solve routing paths) and proper bandwidth-ed ground stations (apparently SpaceX has some issues in some areas of US already). Basically yet another year to wait for the proper Starlink growth.

The two issues are related. You need ISL to shift the teleport traffic away from the congested teleports.

No in-space routing at the IP layer. Use frame forwarding at the frame layer using a specialized frame layer. Because ISL topology changes are known in advance table updates are not reactive. Instead, forwarding tables have scheduled updates what change with microsecond precision. Similarly, the user terminals switch between beams in a precisely-timed fashion, as do teleport-satellite links: all of this is handled by the specialized frame layer and its forwarding tables.  I never worked for SpaceX so I do not know that they are doing it this way, but I did do a preliminary system design as part of a business proposal for a different constellation in 2014.

laser Inter-sat links==mult-node dynamic relay network. At the scale of Starlink they will end with semi-distributed model. You need to do dynamic allocation.

[DanClemmensen]

I'm actually quite surprised at the subscriber growth. It's almost 100k/month now.

"Exponential growth" is a fantasy in the real world. If you want to estimate future market you have to look for the uncovered market. There is considerate incoming market in south East Asia (Indonesia, Tailand, Vietnam where the major clients will be local goverments), South America and US. The growth beyond 500k is bound by the proper development of inter-sat links (software is a prime target, I am extremely curious how they will solve routing paths) and proper bandwidth-ed ground stations (apparently SpaceX has some issues in some areas of US already). Basically yet another year to wait for the proper Starlink growth.

The two issues are related. You need ISL to shift the teleport traffic away from the congested teleports.

No in-space routing at the IP layer. Use frame forwarding at the frame layer using a specialized frame layer. Because ISL topology changes are known in advance table updates are not reactive. Instead, forwarding tables have scheduled updates what change with microsecond precision. Similarly, the user terminals switch between beams in a precisely-timed fashion, as do teleport-satellite links: all of this is handled by the specialized frame layer and its forwarding tables.  I never worked for SpaceX so I do not know that they are doing it this way, but I did do a preliminary system design as part of a business proposal for a different constellation in 2014.

laser Inter-sat links==mult-node dynamic relay network. At the scale of Starlink they will end with semi-distributed model. You need to do dynamic allocation.

I worked with a bunch of very experienced system architects with many collective years of experience in networking, and I learned that it was a common (nearly universal) tendency to try to apply existing network models to the satellite constellation problem. Basically, these models fail in this environment. The constellation topology is indeed highly dynamic: generally several orders of magnitude more setup/teardown events than in terrestrial networks. However, except for a tiny percentage of them the topology changes are also absolutely predictable down to the microsecond. An appropriate specialized frame layer can support this dynamic topology without the need to signal each topology change. Of course you must signal changes caused by failures, restorals, and new satellites. Yes, you need dynamic bandwidth allocation, at least on the RF links (user beams and teleport beams). You may or may not need to consider bandwidth constraints on the ISL. Yes this allocation must be distributed or the system won't scale, but it may or may not be distributed to the teleports or to PoPs on the ground instead of the satellites.