Launching Starlink with Starship

[Robotbeat]

Also, they can launch larger and more massive Starlinks this way when payload increases.

[abaddon]

IMHO in the near to medium term how Starship launches Starlink will depend on how orbital re-entry (for Starship) and reuse (for both Starship and Super Heavy) go.  If they manage to fully recover the launch stack relatively early on and are able to reuse them, it makes sense to launch them more frequently and not-have-to-be full.  If they are regularly losing Starships (or worse, or Super Heavys), they'll want to pack them to the gills rather than build a bunch more expendable boosters/ships.

Eventually, assuming success, I agree maxing out the payload is less important than flight rate.

For this reason, I also would be surprised if we see bigger/heavier Starlinks before Starship stack recovery (and subsequent reuse) is successful at least say half the time.

[Robotbeat]

IMHO in the near to medium term how Starship launches Starlink will depend on how orbital re-entry (for Starship) and reuse (for both Starship and Super Heavy) go.  If they manage to fully recover the launch stack relatively early on and are able to reuse them, it makes sense to launch them more frequently and not-have-to-be full.  If they are regularly losing Starships (or worse, or Super Heavys), they'll want to pack them to the gills rather than build a bunch more expendable boosters/ships....

Good point.

[1]

Shuttle got 27.5t to LEO (200km, 28 deg latitude) but just 12.7t to polar and 16t to ISS orbits. ISS is fairly close to Starlink's orbit. Shuttle is probably closest to these early, heavier Starships than any other vehicles. So this is a pretty reasonable comparison and illustrative of what kind of knockdowns they might have for the earlier, heavier vehicles to different orbits without refueling and with extra margin for landing.

I think the situation is a bit better than this. Remember that the shuttle, regardless of fuel reserves, needed to shut down its main engines early enough to ensure immediate reentry of the external tank. I think a lot of that performance degradation at >200km altitude was due reliance on the comparatively crappy OMS pods to do the final orbital insertion. An extra second or two of SSME burn might have done wonders for the shuttles polar/iss payload numbers.

Starship, of course, shouldn't have this problem. I'd guess that the payload numbers to various LEOs will be much closer to one another.

[GregA]

SpaceX can save a few months of precession drifting by launching more times into more planes. Since equal satellite distribution is necessary to provide equal bandwidth, and customers won't like wildly varying bandwidth as more or less densely populated constellation planes come overhead, they can grow the Starlink customer base faster with more frequent launches. Faster customer growth means more revenue.

I've wondered about launching to multiple shells and planes from one starship launch.

It might not be as big a problem though. Aside from the efficiency of not expecting satellites to move themselves to the correct shell or plane, I assume SpaceX will frequently update the satellites and it's best to keep the same version at each shell (and plane, to a lesser extent). They'll want to be launching big batches of the same satellite to the same altitude / shell.

For now, I think that means they don't want to start any more orbital shells until they have inter satellite links ready to go, to avoid having 2 or 3 shells where a portion of the satellites only work with a ground station, while the rest of the satellites work globally providing they can get a link to another laser-connected sat. Later when Starship is launching Starlink they'll still have newer upgraded satellites with more or faster laser interconnects. If they're paired up with lesser satellites they won't get the benefits, or they'll need to retire the older satellites earlier and I assume they'd rather keep an entire older shell up, next to an entire newer shell - so if my reasoning is right they'll be working on one shell at a time, mostly, and even when it comes time to replace old satellites it'll be whole shells.

[TheRadicalModerate]

...
4) Takes longer for each flight's payload to reach operations.  Eh?  The only time that would be true for Starlink is if there were suddenly a burning need for a half a load, and you couldn't wait for the second half to be manufactured.  Do you really see that being an issue any time soon?  Even if you got to the steady-state maintenance phase, where you were only launching spares in anticipation of upcoming retirements or failures, if you've planned things so that you have a burning need for 200 right now instead 400 two weeks from now, you've done something horribly wrong.

It's at least plausible that more frequent launches with less payload could be worthwhile. It would take some modeling of the vehicle costs and constellation design and revenue to find the optimal point, and I don't think we have good enough data to tell for sure.

SpaceX can save a few months of precession drifting by launching more times into more planes. Since equal satellite distribution is necessary to provide equal bandwidth, and customers won't like wildly varying bandwidth as more or less densely populated constellation planes come overhead, they can grow the Starlink customer base faster with more frequent launches. Faster customer growth means more revenue.

This is a similar effect to all-electric GEO commsats paying more to get inserted closer to the operational orbit. A month of lost revenue using ion thrusters to get into service is a lot of money.

Hadn't thought about optimizing loads to different RAANs; good point.  However, the phase 2 VLEO deployment has every bird with its own RAAN, so there will be no batching of those.  I frankly don't understand what they're doing there, but it makes more sense than the (yet to be licensed) phase 3, where you've got three planes of 7178 birds at three different (low-ish) inclinations.

[TheRadicalModerate]

IMHO in the near to medium term how Starship launches Starlink will depend on how orbital re-entry (for Starship) and reuse (for both Starship and Super Heavy) go.  If they manage to fully recover the launch stack relatively early on and are able to reuse them, it makes sense to launch them more frequently and not-have-to-be full.  If they are regularly losing Starships (or worse, or Super Heavys), they'll want to pack them to the gills rather than build a bunch more expendable boosters/ships.

Eventually, assuming success, I agree maxing out the payload is less important than flight rate.

For this reason, I also would be surprised if we see bigger/heavier Starlinks before Starship stack recovery (and subsequent reuse) is successful at least say half the time.

The only case I can see for doing the smaller payloads / more launches strategy is if they absolutely must do x number of tests (where x > 100) before y date, even if it means launching test masses instead of live payloads.  In that case, you might as well spread out the Starlink constellation over the whole test program, because you'll reduce your insurable loss and (per envy's comment up-thread), you can minimize time to the proper RAAN.

But I think people are way, way hung up on this "hundreds of missions before people fly" comment.  If you believe this is word-for-word accurate, then that would be a huge constraint on the LSS/HLS program and, sure enough, you'd want to launch Starships empty at a furious rate if you had to.  But I don't think that's what Elon intended: it's much more likely that he meant that there would need to be hundreds of launches before people launched, and that's very different.  LSS doesn't launch with people on it; they transfer in NRHO.  And even if SpaceX could talk NASA into replacing SLS with LSS (please, please, please...), they could still do it by transferring people in LEO from/to a D2.

I do agree that this decision hinges critically on the actual amortized cost of an SH/SS launch.  If reuse rises rapidly to 100/SH and 20/SS, then per-launch costs could easily be lower than $5M/launch, and a test program where 50 of the 100 launches you planned had no value other than the test data would only cost you $250M.  But if the early versions are 10/SH and 2/SS, then launches could cost (internal cost) $17-20M, and your 50 throwaways now cost $850M-$1B.  A billion here, a billion there, and pretty soon you're talking about real money...

[Alvian@IDN]

But I think people are way, way hung up on this "hundreds of missions before people fly" comment.  If you believe this is word-for-word accurate, then that would be a huge constraint on the LSS/HLS program and, sure enough, you'd want to launch Starships empty at a furious rate if you had to.  But I don't think that's what Elon intended: it's much more likely that he meant that there would need to be hundreds of launches before people launched, and that's very different.  LSS doesn't launch with people on it; they transfer in NRHO.  And even if SpaceX could talk NASA into replacing SLS with LSS (please, please, please...), they could still do it by transferring people in LEO from/to a D2.

What I (and most people) meant is the crew Starship that will actually launch all the way from Earth, which is Dear Moon (which I don't think they have to wait for significant delays from 2023 just because of waiting 100th uncrewed launches to occur). Of course he could set a launch on Dragon then docked with Dear Moon Starship like you mentioned (which will decrease the crew to 4, and likely make the trip more expensive), which revoke the thought of when they are actually planned to stay on Dragon for the rest of the Moon trip in the first place just 2-3 years ago

[TheRadicalModerate]

But I think people are way, way hung up on this "hundreds of missions before people fly" comment.  If you believe this is word-for-word accurate, then that would be a huge constraint on the LSS/HLS program and, sure enough, you'd want to launch Starships empty at a furious rate if you had to.  But I don't think that's what Elon intended: it's much more likely that he meant that there would need to be hundreds of launches before people launched, and that's very different.  LSS doesn't launch with people on it; they transfer in NRHO.  And even if SpaceX could talk NASA into replacing SLS with LSS (please, please, please...), they could still do it by transferring people in LEO from/to a D2.

What I (and most people) meant is the crew Starship that will actually launch all the way from Earth, which is Dear Moon (which I don't think they have to wait for significant delays from 2023 just because of waiting 100th uncrewed launches to occur). Of course he could set a launch on Dragon then docked with Dear Moon Starship like you mentioned (which will decrease the crew to 4, and likely make the trip more expensive), which revoke the thought of when they are actually planned to stay on Dragon for the rest of the Moon trip in the first place just 2-3 years ago

Yes, that's what I think Musk meant.  I just don't think he meant it for missions where you didn't need to launch the crew on Starship.  At the very least, LSS/HLS as currently specified falls in that category.

DearMoon might be a high-risk exception, but the risk is huge for a reward that isn't much more than a bunch of cool publicity.  Given that, I think they'll either reduce the risk by not doing a crewed launch or EDL, or they simply won't do it.  But that's off-topic.

What's on topic is how Starlink launches are structured on Starship.  If you really need hundreds of launches before any crews can be carried (either launched or rendezvoused in LEO via D2), and you absolutely positively need something crewed before late 2024, then fairly small (~50t) Starlink loads are probably the way to go, and you have to spend a boatload of money to push the schedule to be acceptable for HLS, to say nothing of a Mars attempt in the 2024 window.

But it's a boatload of money to do that--especially if reusability increases fairly slowly.

[oldAtlas_Eguy]

But I think people are way, way hung up on this "hundreds of missions before people fly" comment.  If you believe this is word-for-word accurate, then that would be a huge constraint on the LSS/HLS program and, sure enough, you'd want to launch Starships empty at a furious rate if you had to.  But I don't think that's what Elon intended: it's much more likely that he meant that there would need to be hundreds of launches before people launched, and that's very different.  LSS doesn't launch with people on it; they transfer in NRHO.  And even if SpaceX could talk NASA into replacing SLS with LSS (please, please, please...), they could still do it by transferring people in LEO from/to a D2.

What I (and most people) meant is the crew Starship that will actually launch all the way from Earth, which is Dear Moon (which I don't think they have to wait for significant delays from 2023 just because of waiting 100th uncrewed launches to occur). Of course he could set a launch on Dragon then docked with Dear Moon Starship like you mentioned (which will decrease the crew to 4, and likely make the trip more expensive), which revoke the thought of when they are actually planned to stay on Dragon for the rest of the Moon trip in the first place just 2-3 years ago

Yes, that's what I think Musk meant.  I just don't think he meant it for missions where you didn't need to launch the crew on Starship.  At the very least, LSS/HLS as currently specified falls in that category.

DearMoon might be a high-risk exception, but the risk is huge for a reward that isn't much more than a bunch of cool publicity.  Given that, I think they'll either reduce the risk by not doing a crewed launch or EDL, or they simply won't do it.  But that's off-topic.

What's on topic is how Starlink launches are structured on Starship.  If you really need hundreds of launches before any crews can be carried (either launched or rendezvoused in LEO via D2), and you absolutely positively need something crewed before late 2024, then fairly small (~50t) Starlink loads are probably the way to go, and you have to spend a boatload of money to push the schedule to be acceptable for HLS, to say nothing of a Mars attempt in the 2024 window.

But it's a boatload of money to do that--especially if reusability increases fairly slowly.

Boatloads of money.

SpaceX is spending on the launch of each (just for the launch) Starlink sat from an estimated low of $250K to a high of $500K. For a Starship launch that carries just 180 Starlink sats that is a value of  $45M to $90M. 100 starship launches of Starlink at current expenditure amount for Starlink deployment is a value of $4.5B to $9B.

Is that enough for a boatload of money?

So if Starship actually is slightly cheaper say $35M a launch then Space will have saved at least $1B over the launch of 100 starships full of Starlink sats. But lets say the actual F9 launch costs for Starlink is the $30M value ($500K/sat). A Starship launch cost of $60M would save SpaceX $3B over the launch of 100 Starships with a total of 18,000 sats.

As more Starshiops are launched the price per launch of Starship will go down. Some of the Starlink savings will be the payload increases such that instead of 180 sats per launch it is 240 sats per launch. A 30% decrease in cost of launch per sat without decreasing the specific cost per launch just increase the performance (tonnage) per launch.

[Ludus]

IMHO in the near to medium term how Starship launches Starlink will depend on how orbital re-entry (for Starship) and reuse (for both Starship and Super Heavy) go.  If they manage to fully recover the launch stack relatively early on and are able to reuse them, it makes sense to launch them more frequently and not-have-to-be full.  If they are regularly losing Starships (or worse, or Super Heavys), they'll want to pack them to the gills rather than build a bunch more expendable boosters/ships....

Good point.

It’s worth noting that a high Starship production rate is also part of the plan in advance.

This plan goes way beyond any conceivable demand from outside customers. It’s the opposite situation from STS and allows rapid iterative improvement. It doesn’t matter so much if they lose some as long as no humans are involved and there’s a new generation that fixes the problem coming along in a month or two anyway. Not that they are being careless but they’re designing a system that isn’t going to shut down for a year and be obsessed about what went wrong if there is a failure. There’s no assumption any generation of Starship won’t be quickly scrapped.

Other customers can piggyback on this process and get better freight rates into orbit as long as they can accept the rules. Effectively SpaceX will soon end up launching 90%+ by mass of everything humans put in space and the entire global launch market will easily fit into the extra volume/mass available on scheduled Starship-Starlink launches. 

Many customers obviously will reject this option for various reasons but it’s still how the numbers will come out.

Robotbeat’s hundreds of scheduled Starship-Starlink launches that don’t push the limits and can easily accommodate other payloads (as they’ve been trying on F9) fits with this.

[TheRadicalModerate]

Boatloads of money.

SpaceX is spending on the launch of each (just for the launch) Starlink sat from an estimated low of $250K to a high of $500K. For a Starship launch that carries just 180 Starlink sats that is a value of  $45M to $90M. 100 starship launches of Starlink at current expenditure amount for Starlink deployment is a value of $4.5B to $9B.

Is that enough for a boatload of money?

So if Starship actually is slightly cheaper say $35M a launch then Space will have saved at least $1B over the launch of 100 starships full of Starlink sats. But lets say the actual F9 launch costs for Starlink is the $30M value ($500K/sat). A Starship launch cost of $60M would save SpaceX $3B over the launch of 100 Starships with a total of 18,000 sats.

As more Starshiops are launched the price per launch of Starship will go down. Some of the Starlink savings will be the payload increases such that instead of 180 sats per launch it is 240 sats per launch. A 30% decrease in cost of launch per sat without decreasing the specific cost per launch just increase the performance (tonnage) per launch.

The boatload of money for the actual Starlinks is:

a) Invariant, and...
b) Attached to an actual revenue stream, which is enabled by launching it.

Launching more Starships than necessary has a value, in that getting the system marginally more reliable reduces future costs, which improves the marginal return on future launches.  But the IRR computation for Starlink is easy, and the computation for whether, say, doubling the number of launches has an actual return is... not.

[groundbound]

Boatloads of money.

SpaceX is spending on the launch of each (just for the launch) Starlink sat from an estimated low of $250K to a high of $500K. For a Starship launch that carries just 180 Starlink sats that is a value of  $45M to $90M. 100 starship launches of Starlink at current expenditure amount for Starlink deployment is a value of $4.5B to $9B.

Is that enough for a boatload of money?

So if Starship actually is slightly cheaper say $35M a launch then Space will have saved at least $1B over the launch of 100 starships full of Starlink sats. But lets say the actual F9 launch costs for Starlink is the $30M value ($500K/sat). A Starship launch cost of $60M would save SpaceX $3B over the launch of 100 Starships with a total of 18,000 sats.

As more Starshiops are launched the price per launch of Starship will go down. Some of the Starlink savings will be the payload increases such that instead of 180 sats per launch it is 240 sats per launch. A 30% decrease in cost of launch per sat without decreasing the specific cost per launch just increase the performance (tonnage) per launch.

The boatload of money for the actual Starlinks is:

a) Invariant, and...
b) Attached to an actual revenue stream, which is enabled by launching it.

Launching more Starships than necessary has a value, in that getting the system marginally more reliable reduces future costs, which improves the marginal return on future launches.  But the IRR computation for Starlink is easy, and the computation for whether, say, doubling the number of launches has an actual return is... not.

You missed a major term in the equation, without which any understanding is compromised.

One of Musk's major superpowers, which he has so far used sparingly with SpaceX/Starlink, is to obtain access to stupid quantities of capital at stupidly favorable terms.

Look at what has been done with a certain auto/energy company over the years. I've always suspected that certain nuances of goals/plans/products/announcements are chosen to maximize this, both there and at SpaceX.

And getting a decade's use of $10 billion starting a year from now without giving up anything important would change the financial picture you lay out substantially, without really invalidating anything you said.

[TheRadicalModerate]

Boatloads of money.

SpaceX is spending on the launch of each (just for the launch) Starlink sat from an estimated low of $250K to a high of $500K. For a Starship launch that carries just 180 Starlink sats that is a value of  $45M to $90M. 100 starship launches of Starlink at current expenditure amount for Starlink deployment is a value of $4.5B to $9B.

Is that enough for a boatload of money?

So if Starship actually is slightly cheaper say $35M a launch then Space will have saved at least $1B over the launch of 100 starships full of Starlink sats. But lets say the actual F9 launch costs for Starlink is the $30M value ($500K/sat). A Starship launch cost of $60M would save SpaceX $3B over the launch of 100 Starships with a total of 18,000 sats.

As more Starshiops are launched the price per launch of Starship will go down. Some of the Starlink savings will be the payload increases such that instead of 180 sats per launch it is 240 sats per launch. A 30% decrease in cost of launch per sat without decreasing the specific cost per launch just increase the performance (tonnage) per launch.

The boatload of money for the actual Starlinks is:

a) Invariant, and...
b) Attached to an actual revenue stream, which is enabled by launching it.

Launching more Starships than necessary has a value, in that getting the system marginally more reliable reduces future costs, which improves the marginal return on future launches.  But the IRR computation for Starlink is easy, and the computation for whether, say, doubling the number of launches has an actual return is... not.

You missed a major term in the equation, without which any understanding is compromised.

One of Musk's major superpowers, which he has so far used sparingly with SpaceX/Starlink, is to obtain access to stupid quantities of capital at stupidly favorable terms.

Look at what has been done with a certain auto/energy company over the years. I've always suspected that certain nuances of goals/plans/products/announcements are chosen to maximize this, both there and at SpaceX.

And getting a decade's use of $10 billion starting a year from now without giving up anything important would change the financial picture you lay out substantially, without really invalidating anything you said.

I'd guess that a big chunk of those stupid amounts of capital were attracted because Starlink could turn out to be a license to print money at high scale, in a very short timeframe.  Starship is absolutely a decent chunk of that biz case, because without it you don't get the scale you need.

But then you have to ask yourself if the bulk of the investors really care if a small reduction in the marginal cost of a Starship launch 5 or 6 years from now, instead of maybe 8 or 9 years if the testing program is more frugal, is worth tossing $0.5B-$1.0B into the test program, when the piece of tech that allows Starlink to scale (i.e., pretty cheap Starship launches) already works.

The thing on the bubble here is whether the investors actually believe that there's big money to be made in crewed Starship missions in the next 5-6 years.  And we're talking big money--money that would be of the same order as Starlink.  I just don't see that happening.  Sure, some of the investors probably believe that Starship grows up to be a solar-system-spanning monopoly that generates trillions of dollars of revenue, and they're willing to invest for the very long term.  But, as you say, SpaceX is burning through stupid amounts of capital right now.  Better to take that extra $0.5B-$1.0B and make more Starlinks with it, so that when the time comes to make the thousands of Starships and launch tens of thousands of people to the Moon, Mars, and beyond, the cash flow situation isn't even a consideration.

[Ludus]

But then you have to ask yourself if the bulk of the investors really care if a small reduction in the marginal cost of a Starship launch 5 or 6 years from now, instead of maybe 8 or 9 years if the testing program is more frugal, is worth tossing $0.5B-$1.0B into the test program, when the piece of tech that allows Starlink to scale (i.e., pretty cheap Starship launches) already works.

The thing on the bubble here is whether the investors actually believe that there's big money to be made in crewed Starship missions in the next 5-6 years.  And we're talking big money--money that would be of the same order as Starlink.  I just don't see that happening.  Sure, some of the investors probably believe that Starship grows up to be a solar-system-spanning monopoly that generates trillions of dollars of revenue, and they're willing to invest for the very long term.  But, as you say, SpaceX is burning through stupid amounts of capital right now.  Better to take that extra $0.5B-$1.0B and make more Starlinks with it, so that when the time comes to make the thousands of Starships and launch tens of thousands of people to the Moon, Mars, and beyond, the cash flow situation isn't even a consideration.

The bulk of the Investors accept that Elon Musk’s SpaceX is an integral thing, it can’t be parsed into radical vision vs frugal conventional investment whenever they want. It achieves the breakthrough positions that it has, precisely because it doesn’t do the sort of business calculation that Boeing or ULA does. There was no business case for developing reusable rockets or as soon as they’re paying off moving on to a crazy giant Mars Rocket. It finds itself in unusually advantageous positions because it doesn’t explore the landscape of possibility in the usual way. If Starlink is profitable enough to seriously help pay for Mars settlement, nobody will second guess how Musk decides to allocate capital.

[oldAtlas_Eguy]

Even if Starlink is barely profitable. It will fund the launch of dozens of Starships a year.

Starship will likely progress rapidly to multiple reuse of SH and SS.

SH reuse is likely after about 6 moths to a year of orbital flights. (2022) After that the number of flights achieved per SH will rapidly climb to as many as a value of 4 in one year.

Reuse for SS may take 2 years to achieve. But the flight numbers after that likely will rapidly increase such  that by 2024 with 8 flights per SH and 4 for SS. This could represent a per launch cost as low as $25M.

[TheRadicalModerate]

Even if Starlink is barely profitable. It will fund the launch of dozens of Starships a year.

Starship will likely progress rapidly to multiple reuse of SH and SS.

SH reuse is likely after about 6 moths to a year of orbital flights. (2022) After that the number of flights achieved per SH will rapidly climb to as many as a value of 4 in one year.

Reuse for SS may take 2 years to achieve. But the flight numbers after that likely will rapidly increase such  that by 2024 with 8 flights per SH and 4 for SS. This could represent a per launch cost as low as $25M.

Which has more bang for the buck:  using Starlink profits to fund relatively more Starship launches, or using them to fund building more Starlinks?  Starlink is exactly the kind of business where you'd expect massive returns on scale.  Launching stuff is not.

[Lar]

Even if Starlink is barely profitable. It will fund the launch of dozens of Starships a year.

Starship will likely progress rapidly to multiple reuse of SH and SS.

SH reuse is likely after about 6 moths to a year of orbital flights. (2022) After that the number of flights achieved per SH will rapidly climb to as many as a value of 4 in one year.

Reuse for SS may take 2 years to achieve. But the flight numbers after that likely will rapidly increase such  that by 2024 with 8 flights per SH and 4 for SS. This could represent a per launch cost as low as $25M.

Which has more bang for the buck:  using Starlink profits to fund relatively more Starship launches, or using them to fund building more Starlinks?  Starlink is exactly the kind of business where you'd expect massive returns on scale.  Launching stuff is not.

That's not really an either/or.

[oldAtlas_Eguy]

Even if Starlink is barely profitable. It will fund the launch of dozens of Starships a year.

Starship will likely progress rapidly to multiple reuse of SH and SS.

SH reuse is likely after about 6 moths to a year of orbital flights. (2022) After that the number of flights achieved per SH will rapidly climb to as many as a value of 4 in one year.

Reuse for SS may take 2 years to achieve. But the flight numbers after that likely will rapidly increase such  that by 2024 with 8 flights per SH and 4 for SS. This could represent a per launch cost as low as $25M.

Which has more bang for the buck:  using Starlink profits to fund relatively more Starship launches, or using them to fund building more Starlinks?  Starlink is exactly the kind of business where you'd expect massive returns on scale.  Launching stuff is not.

That's not really an either/or.

Without Starship there will not be a significant expansion of Starlink because of the logistics of launch. In order to match Starship launches in a year (12) it would take 48 F9 launches in a year ~1 a week. At the moment the F9 US manufacturing tops out at <50/year and with at least 12 non Starlink launches also on F9 the best number of launches per year of Starlink on F9 would be 36.

Without Starship rapid expansion of Starlink will not occur.

Also with an average for Starship launches of 100 launches over a period of 4 years of $50M/launch with the cost at the end of the period being ~$25M/launch. The cost per sat for launch of a Starlink sat over that period by Starship. Equivalent to 400 F9 launches. Would be =< the best estimate/information from SpaceX/Musk of F9 launch costs of $15m/launch of an F9 (which may be glossing over the some the total system costs).

NOTE: 100 F9 launches at this time is logistically not possible. Would require the doubling of the manufacturing space and tooling for F9 manufacturing and M1D at Hawthorne as well as doubling or tripling the testing facilities at McGregor for M1D and F9 testing to accomplish that launch rate. Not to mention launching once a week from each of the 2 Cape pads as well as 12 or more times per year from VAFB to support at total launch rate of somewhere around 110 to 120 in a year.

NOTE: Whereas the average launch of 25 Starships in a year of ~1 per month from a probable each of 2 pads (LC 39A and Boca Chica) is easily achieved. Where 18 SS are manufacture per year and 6 SH are manufactured per year. That is currently Boca Chica's build rate capacity not requiring additional equipment/tooling other than that currently being installed.

The money will be spent to put the satellites into orbit no mater what the LV. But revenue cannot be made on satellites that are sitting on the ground.

[TheRadicalModerate]

Without Starship there will not be a significant expansion of Starlink because of the logistics of launch. In order to match Starship launches in a year (12) it would take 48 F9 launches in a year ~1 a week. At the moment the F9 US manufacturing tops out at <50/year and with at least 12 non Starlink launches also on F9 the best number of launches per year of Starlink on F9 would be 36.

Without Starship rapid expansion of Starlink will not occur.

Also with an average for Starship launches of 100 launches over a period of 4 years of $50M/launch with the cost at the end of the period being ~$25M/launch. The cost per sat for launch of a Starlink sat over that period by Starship. Equivalent to 400 F9 launches. Would be =< the best estimate/information from SpaceX/Musk of F9 launch costs of $15m/launch of an F9 (which may be glossing over the some the total system costs).

NOTE: 100 F9 launches at this time is logistically not possible. Would require the doubling of the manufacturing space and tooling for F9 manufacturing and M1D at Hawthorne as well as doubling or tripling the testing facilities at McGregor for M1D and F9 testing to accomplish that launch rate. Not to mention launching once a week from each of the 2 Cape pads as well as 12 or more times per year from VAFB to support at total launch rate of somewhere around 110 to 120 in a year.

NOTE: Whereas the average launch of 25 Starships in a year of ~1 per month from a probable each of 2 pads (LC 39A and Boca Chica) is easily achieved. Where 18 SS are manufacture per year and 6 SH are manufactured per year. That is currently Boca Chica's build rate capacity not requiring additional equipment/tooling other than that currently being installed.

The money will be spent to put the satellites into orbit no mater what the LV. But revenue cannot be made on satellites that are sitting on the ground.

I agree with everything you have in here.  Starship is absolutely necessary to get Starlink deployed at a scale large enough that the returns on scale kick it.  And it obviously does no good to launch fewer Starships than are necessary to put all the Starlinks that the factory produces into orbit ASAP.

The question is whether it makes sense to launch more Starships than are necessary to service Starlink output (plus the small number of whatever other customers come along), simply to drive Starship reuse up and launch costs down.  Robotbeat was implicitly arguing for this approach here, and it doesn't make any sense to me.

If you were going to halve the cost of launch in only a few launches by doing this, then halving the number of Starlinks to double the tempo might make sense.  Likewise, if there's some terrible burning need to actually send people through launch and EDL on a Starship before 2024, it might make sense.  But neither of these things are going to happen.

Starship simply doesn't need to get that cheap that fast.  You can get more than adequate cost reductions simply by launching the 70-80 Starships you'll need to get the proposed Starlink constellation deployed over the next 4-5 years.