Amazon Project Kuiper Broadband Constellation

[RedLineTrain]

Thank goodness entrepreneurs don't have your defeatist attitude, other wise all our product and services suppliers would have monopoly. Customers switch suppliers regularly for varies reasons g poor service, price,  features. I've switched broadband providers of broadband, power,  banks and insurance companies on a few occasions for all of those reasons.

I think it's reasonable to expect that entrepreneurs have a credible business plan.  Kuiper's plan might have made sense two years ago, but that's a lifetime ago and now needs to be evaluated afresh.

[DanClemmensen]

Thank goodness entrepreneurs don't have your defeatist attitude, other wise all our product and services suppliers would have monopoly. Customers switch suppliers regularly for varies reasons g poor service, price,  features. I've switched broadband providers of broadband, power,  banks and insurance companies on a few occasions for all of those reasons.

I think it's reasonable to expect that entrepreneurs have a credible business plan.  Kuiper's plan might have made sense two years ago, but that's a lifetime ago and now needs to be evaluated afresh.

The following is all personal opinion based solely on information in this forum.

I think their plan was marginal two years ago, but they might have succeeded if everything worked perfectly. However, all of their Launch providers slipped badly. They should have shifted the test satellites to an alternate existing LV (Atlas V or F9) instantly instead of switching to Vulcan, losing another year, and then switching to Atlas V anyway. Had they done so they would have started launching on Atlas V in June of 2023 or before, and would by now have more than 200 satellites, which is enough to build out and thoroughly test the ground infrastructure, and they should have started launching on F9 as soon as it could be arranged once it was clear that Vulcan Centaur would not be able to ramp quickly.

As of now, I see no way they can expect to meet the June 2026 deadline for the 1600 satellites, so they will need to get an extension. Actually, there is one way: start launching on F9 right now. They need about 58 satellites/month for 28 months. The problem of course is they probably cannot do this because they have not tested their deployer, which they would have been able to do if they had started launching on Atlas V.

[Tywin]

I see a lot of errors of appreciation in this debate...

First Starlink, was NOT the first mega constellation, it was Oneweb, and the idea was very old....

Second, Starlink HAS competition, all GEO Operators have satellite internet, and they still have a huge market share in this sector...

Third, Kuiper does not need to be the market leader at first, only the second constellation in the West Hemisphere, Mega, and Amazon, can and usually will do, phone connection too...like messaging, and SOS....

Fourth, everyone talking about the huge revenew a "little" Amazon's profits .... ok, how much profit do they have to HAVE
 RIGH NOW, Starlink...?


Because so far, SpaceX survive with new capital every year ...

Fifth, it is impossible to predict a new market, so many early adopters disappear in the mature market ...

We will see ...

[Vultur]

The reason for Kuiper, I think, is that Amazon will have control of it, unlike Starlink or OneWeb.

Whether that is a good enough reason is entirely another question...

[tssp_art]

Well, Kuiper could make sense if Amazon had a way of overcoming the SpaceX advantage of being able to launch at low cost relative to all other constellation providers. And the advantage is not just for the initial deployment but in the 10-20% renewal required every year to replace the limited lifespan satellites.

So, what if Amazon plans to buy Blue Origin. They would focus it primarily. on Kuiper but willing to sell any excess capacity to others. The cadence established by Kuiper launches would make thier New Glenn offering a reliable and (relatively) low cost option. I know the talk today is of Blue Origin buying ULA but maybe that's not what Amazon has in mind.

This is similar to the model that created AWS. Initially, Amazon needed a "cloud" of servers to accommodate the unprecedented computing needs of of Amazon retail during the Christmas rush. The infrastructure was flexible and designed to easily support a large heterogenous group of users with dynamically allocated virtual servers. But rather than letting it side idle for 9 months a year Amazon made it available to outside users as a service. This rapidly grew into the AWS part of Amazon that actually generated profit - something that has still eluded the consumer facing business.

So, Amazon buys Blue Origin, reshapes it into a highly efficient launch provider for the main purpose of launching and maintaining Kuiper (not unlike SpaceX's relationship with Starlink). That would give Kuiper a low cost launch provider and maybe a fighting chance against Starlink's monstrous head start. Lots of "if"s and "maybe"s in that speculation but it seems like it might align the incentives correctly.

[Vultur]

That might work - if Blue Origin under Amazon's control could become competitive with SpaceX in launch. That frankly seems unlikely.

And that's not a slam against Blue Origin. It's not really clear that anyone can compete with SpaceX in launch. China, maybe, but it's not really fair to compare one of the world's largest nations to a single company - China's launches come from a number of different companies/agencies. (And last year, SpaceX still launched more.)

Now, there's nothing magical about SpaceX. There's no fundamental reason why another company, perhaps Blue Origin/Amazon, couldn't do equally well - maybe better, given that they could learn from SpaceX and skip some of the testing SpaceX needed to figure out, e.g., how to land boosters. But in practice, nobody seems to be getting there. I think SpaceX benefited from a bit of a "perfect storm" combination of right people/right time/fresh-start organizational structure, and has used that to get a lead that would now be incredibly hard to overcome.

--

But another way to look at it is kind of analogous to why nations want their own space access, etc., rather than just buying rides from SpaceX. A company as big as Amazon is effectively operating on that scale: Amazon's revenues are comparable to the GDP of a mid-size nation.

Amazon isn't going to want to be totally dependent on SpaceX, especially given the perception (not going to argue if it's valid or not, only that the perception exists) that Elon Musk is somewhat unpredictable. If Amazon has the money to spend, then having their own constellation as a hedge against that is understandable. Perhaps not financially sensible - maybe it will work out badly for them in the long run - but understandable.

It wasn't that long ago, really, that Starlink was "a crazy idea no one took seriously". People are adapting to it suddenly being huge.

[Tywin]

I don't see any technologies advantange that the family Falcon has that can not be replicate...

Some people here talk like the Ford, modele T, of the rocket...guess what?

We drive many cars, that are not Ford...

[Tywin]

Where is the technologies advantage of Starlinks, laser satellite?

Kuiper will have too...

You need, Money, money and more money, Napoleon said...

Guess what?

Amazon HAVE a lot...

[RedLineTrain]

The following is all personal opinion based solely on information in this forum.

I think their plan was marginal two years ago, but they might have succeeded if everything worked perfectly. However, all of their Launch providers slipped badly. They should have shifted the test satellites to an alternate existing LV (Atlas V or F9) instantly instead of switching to Vulcan, losing another year, and then switching to Atlas V anyway. Had they done so they would have started launching on Atlas V in June of 2023 or before, and would by now have more than 200 satellites, which is enough to build out and thoroughly test the ground infrastructure, and they should have started launching on F9 as soon as it could be arranged once it was clear that Vulcan Centaur would not be able to ramp quickly.

As of now, I see no way they can expect to meet the June 2026 deadline for the 1600 satellites, so they will need to get an extension. Actually, there is one way: start launching on F9 right now. They need about 58 satellites/month for 28 months. The problem of course is they probably cannot do this because they have not tested their deployer, which they would have been able to do if they had started launching on Atlas V.

It seems likely that they aren't launching now because they don't have the satellites to launch now.  It's a coordination error.

[DanClemmensen]

The following is all personal opinion based solely on information in this forum.

I think their plan was marginal two years ago, but they might have succeeded if everything worked perfectly. However, all of their Launch providers slipped badly. They should have shifted the test satellites to an alternate existing LV (Atlas V or F9) instantly instead of switching to Vulcan, losing another year, and then switching to Atlas V anyway. Had they done so they would have started launching on Atlas V in June of 2023 or before, and would by now have more than 200 satellites, which is enough to build out and thoroughly test the ground infrastructure, and they should have started launching on F9 as soon as it could be arranged once it was clear that Vulcan Centaur would not be able to ramp quickly.

As of now, I see no way they can expect to meet the June 2026 deadline for the 1600 satellites, so they will need to get an extension. Actually, there is one way: start launching on F9 right now. They need about 58 satellites/month for 28 months. The problem of course is they probably cannot do this because they have not tested their deployer, which they would have been able to do if they had started launching on Atlas V.

It seems likely that they aren't launching now because they don't have the satellites to launch now.  It's a coordination error.

I speculate that it's not a coordination error. The plan was likely well coordinated, but a critical-path item slipped.
I was assuming that the factory was mostly ready to go and was waiting for the results of the 2-satellite test and any resulting design modifications before starting actual serial production. Thus, the test launch is on the critical path. The project plan would have included a nominal fixed time between test launch and start of production. If they were not ready to go like this, then project Kuiper is in even worse shape than my analysis. It looked to me back then that they were assuming this would take about nine months between test launch and first production launch, so the factory should have been in the almost-ready state no later than about October 2023.

[RedLineTrain]

Now, there's nothing magical about SpaceX. There's no fundamental reason why another company, perhaps Blue Origin/Amazon, couldn't do equally well - maybe better, given that they could learn from SpaceX and skip some of the testing SpaceX needed to figure out, e.g., how to land boosters. But in practice, nobody seems to be getting there. I think SpaceX benefited from a bit of a "perfect storm" combination of right people/right time/fresh-start organizational structure, and has used that to get a lead that would now be incredibly hard to overcome.

Megaconstellations are a complex coordination problem among four dissimilar businesses and tech trees (launch, satellites, consumer electronics, and the network/ground segment) and tens of billions of dollars of capital over time.  Some of those businesses tend toward winner-takes-all or winner-takes-most economics.  There is also a regulatory overlay.

SpaceX has chosen an integrated solution in order to facilitate this coordination and to minimize costs.  As far as I can tell, much of this coordination is done in Musk's head.  Amazon has chosen a solution that is integrated except for launch (and maybe consumer electronics).  Through AWS, they also get the network portion for no additional cost.  The Chinese government has chosen a solution that is integrated at the national level.  Oneweb, Telesat, and Iris^2 are not ambitious enough to matter.

So far, it appears that Amazon is struggling with this coordination.  Kuiper is coming in 5x-10x higher cost than Starlink and is roughly 5 years behind.  While there is nothing fundamental that Amazon cannot do, it must do it.  It must execute at a higher level than Starlink in order to thrive.  Starlink's lead is only getting larger.

[RedLineTrain]

I speculate that it's not a coordination error. The plan was likely well coordinated, but a critical-path item slipped.
I was assuming that the factory was mostly ready to go and was waiting for the results of the 2-satellite test and any resulting design modifications before starting actual serial production. Thus, the test launch is on the critical path. The project plan would have included a nominal fixed time between test launch and start of production. If they were not ready to go like this, then project Kuiper is in even worse shape than my analysis. It looked to me back then that they were assuming this would take about nine months between test launch and first production launch, so the factory should have been in the almost-ready state no later than about October 2023.

This sounds like a coordination error, or at least a mismatch that was not adequately mitigated.  Agency is real.  Management must manage!

There have been at least two big turning points for Starlink that were solved by agency.  One was Musk's June 2018 firing Starlink management and redesigning the satellites with Mark Juncosa and team.  The other was the introduction of the v2 mini and the Falcon 9 launch ramp, since Starship is a few years later than wished.

[meekGee]

Where is the technologies advantage of Starlinks, laser satellite?

Kuiper will have too...

You need, Money, money and more money, Napoleon said...

Guess what?

Amazon HAVE a lot...

You need money, nut money is not the only thing.  Necessary, but not sufficient.

Two examples above: Amazon phone, and your own example, MS phone.

So what else does a second comer like Amazon need?  Comparable technologies.

They will sure have laser links, but what they're missing is cheap launch.

Right now they're using Atlas against Falcon.  In a year or so, they'll use NG against Starship.

It's like trying to set up competition with FedEx, but without access to jet engine technology.

Also the matter of timing: Starlink is 5 years ahead, and this is not the car market.. They have over a million costumers, revenue is around $2B, and turning a profit now.  This is before Starship, V2, and direct to phone service.

Those are insane numbers.  Amazon having money is not enough to make Kuiper succeed in competing against it.

[RedLineTrain]

Also the matter of timing: Starlink is 5 years ahead, and this is not the car market.. They have over a million costumers, revenue is around $2B, and turning a profit now.  This is before Starship, V2, and direct to phone service.

Rather, Payload estimates that Starlink 2023 revenues were $4.2 billion and projects that 2024 revenues will be $6.8 billion.

https://payloadspace.com/estimating-spacexs-2023-revenue/
https://payloadspace.com/predicting-spacexs-2024-revenue/

[meekGee]

Also the matter of timing: Starlink is 5 years ahead, and this is not the car market.. They have over a million costumers, revenue is around $2B, and turning a profit now.  This is before Starship, V2, and direct to phone service.

Rather, Payload estimates that Starlink 2023 revenues were $4.2 billion and projects that 2024 revenues will be $6.8 billion.

https://payloadspace.com/estimating-spacexs-2023-revenue/
https://payloadspace.com/predicting-spacexs-2024-revenue/

Oh wow, I blinked for a year.

[meekGee]

Also the matter of timing: Starlink is 5 years ahead, and this is not the car market.. They have over a million costumers, revenue is around $2B, and turning a profit now.  This is before Starship, V2, and direct to phone service.

Rather, Payload estimates that Starlink 2023 revenues were $4.2 billion and projects that 2024 revenues will be $6.8 billion.

https://payloadspace.com/estimating-spacexs-2023-revenue/
https://payloadspace.com/predicting-spacexs-2024-revenue/

Oh wow, I blinked for a year.

Keep in mind that Amazon, the entire entirety of it, is some $570B. For Starlink to get to 1% of that after 5 years, that's saying something.

It also speaks to the required investment.

Starlink launched, what, 80 times last year?
Probably at least another 100 times this year?

How is Kuiper going to match that?  Not with a handful of Atlas or Vulcan launches.  Even NG can't compete, and by the time NG flies, Starship will too.

I just can't aee a path for Kuiper to compete.

[Vultur]

I don't see any technologies advantange that the family Falcon has that can not be replicate...

I would argue that SpaceX's big advantage is not technological, but organizational. Other companies could use/are using/are working toward the same basic technological solutions - propulsive landing, landing on barges, methalox, etc.

SpaceX's largest advantages are IMO cost efficiency (compare the cost of developing F1 & then F9 to what Blue Origin will have spent before NG flies -- or even what Blue spent up to last month, if we charitably count the use of BE-4 on Vulcan as BO's "first orbital") and speed of iteration.

For Kuiper to compete, they need that speed and efficiency, I think. That would mean launching primarily on F9 (if that's possible for them - sure, they can buy some F9 launches, but enough? And even then, they wouldn't be getting SpaceX's internal pricing, so Starlink still has an advantage) or finding another launcher that can compete on that - which does not currently exist. Maybe NG will, someday, but BO's corporate history does not suggest it. Does Vulcan even aspire to F9-like launch rates? (Is it even possible, given the competition for BE-4 with NG?)

Maybe RL's Neutron, eventually? or Relativity's Terran R? But Kuiper's deadlines are too soon for those.

[matthewkantar]

Warning: May contain sports analogy.

SpaceX fired the quarterback of its satellite team. Kuiper has to win the Super Bowl with a castoff QB they picked up. The castoff knows a lot about their old team, has been down the road, but are they realistically going to win it all?

[Zed_Noir]

<snip>
Starlink launched, what, 80 times last year?
Probably at least another 100 times this year?

How is Kuiper going to match that?  Not with a handful of Atlas or Vulcan launches.  Even NG can't compete, and by the time NG flies, Starship will too.

I just can't aeesee a path for Kuiper to compete.

Ye of little imagination. One of the possible path forward for Kuiper is to book many flights on customized Kuiper Deployment Starships to deployed their LEO constellation.

SpaceX as a Spacing Guild (Western commercial launch provider with almost monopoly status by tonnage lofted to orbit) can not turn down a paying customer, if said customer is willing to fund the development and production of a Starship variant.

It is a win-win for almost everybody. Amazon able to deployed their constellation more or less on schedule, while avoiding further shareholder lawsuits. SpaceX gets bundles of cash and lobbying help. The consumer will likely be offer better internet access deals. Other LEO satcom constellations could also deployed and replenished their constellation with Starship. The only loser will be some bald guy who make bad presumptions and bad program implementation choices, have to suffer the indignity of being "rescue" by his arch rival.

[DanClemmensen]

<snip>
Starlink launched, what, 80 times last year?
Probably at least another 100 times this year?

How is Kuiper going to match that?  Not with a handful of Atlas or Vulcan launches.  Even NG can't compete, and by the time NG flies, Starship will too.

I just can't aeesee a path for Kuiper to compete.

Ye of little imagination. One of the possible path forward for Kuiper is to book many flights on customized Kuiper Deployment Starships to deployed their LEO constellation.

SpaceX as a Spacing Guild (Western commercial launch provider with almost monopoly status by tonnage lofted to orbit) can not turn down a paying customer, if said customer is willing to fund the development and production of a Starship variant.

It is a win-win for almost everybody. Amazon able to deployed their constellation more or less on schedule, while avoiding further shareholder lawsuits. SpaceX gets bundles of cash and lobbying help. The consumer will likely be offer better internet access deals. Other LEO satcom constellations could also deployed and replenished their constellation with Starship. The only loser will be some bald guy who make bad presumptions and bad program implementation choices, have to suffer the indignity of being "rescue" by his arch rival.

A custom Starship seems like a lot of extra work. Simpler to just purchase Starlink V.2 Satellites from SpaceX and use them as Kuiper satellites instead of building their own.