Amazon Project Kuiper Broadband Constellation

[Newton_V]

<snip>
Interesting. I would have guessed 4 solids since 6 solids get only 11% more mass to LEO (27.2 tonnes vs. 24.6 tonnes) and I would have guessed 2 more solids raise costs by more than 11%. Either solids are cheaper than I thought or there's an important factor other than cost per kg.

Edit: 6 solids help more at higher inclinations for some reason, e.g. 25.8 tonnes for 6 solids to ISS, 21.6 tonnes for 4. That's probably why they're using 6.

The Centaur stage for the Kuiper missions is a lighter variant with less propellants. So the the 6 solid boosters have to compensate for the Centaur variant's shorter burn time getting to LEO.

The 4-solid variant with the shorter Centaur gets more mass to a LEO orbit than the 4-solid version with the 120k Centaur.
The 6-solid variant with the shorter Centaur gets more mass to a LEO orbit than the 6-solid version with the 120k Centaur.
It was the same with an Atlas V 401 to LEO.
For example:  Let's say you were putting 20k lbs into orbit Y, and had 5k lbs of PE (propellant excess) remaining, and total burn time was 800 seconds.  If you offloaded 2k lbs of propellant, the burn time to get same 20k mass to orbit Y would be about 40+ seconds shorter, but you would still have slightly over 5k lbs of PE remaining. --> It's less optimal to carry and then burn that extra mass unless your partial of payload mass to PE is > 1  (technically less than -1)
"Partial" explanation:  A partial of -1.2 means if your payload mass goes up 120 lbs, you only need to burn 100 more lbs of propellant to get to the same orbit.  +120/-100 = -1.2
Partials of loaded/tanked propellants are different.

[Rebel44]

Can you disclose what variant of Vulcan that Kuiper needs?

https://twitter.com/torybruno/status/1730971276876144760

Yes.  6 solids, long fairing, and the LEO optimized (small) version of Centaur V

Interesting. I would have guessed 4 solids since 6 solids get only 11% more mass to LEO (27.2 tonnes vs. 24.6 tonnes) and I would have guessed 2 more solids raise costs by more than 11%. Either solids are cheaper than I thought or there's an important factor other than cost per kg.

Edit: 6 solids help more at higher inclinations for some reason, e.g. 25.8 tonnes for 6 solids to ISS, 21.6 tonnes for 4. That's probably why they're using 6.

Part of the reason might be to minimize the (already very high) number of needed launches because I expect the launch rate to be seen as a big schedule risk.

[david1971]

What do we learn from the mid-2025 launch dates?  Why an 18 month delay between now and these launches?  Does it take that long to figure out a payload adapter for these satellites that will work on a Falcon 9?  Is Amazon that far away from being able to deliver their satellites in bulk?  Did SpaceX claim "we're going to give schedule priority for Star Link over new customers until then."?

Rephrasing my question, if schedule concerns are looming for Kuiper, why aren't these contracts for 2024, what's preventing a faster roll-out?

[AmigaClone]

What do we learn from the mid-2025 launch dates?  Why an 18 month delay between now and these launches?  Does it take that long to figure out a payload adapter for these satellites that will work on a Falcon 9?  Is Amazon that far away from being able to deliver their satellites in bulk?  Did SpaceX claim "we're going to give schedule priority for Star Link over new customers until then."?

Rephrasing my question, if schedule concerns are looming for Kuiper, why aren't these contracts for 2024, what's preventing a faster roll-out?

It could be something as simple as SpaceX having a standard contract that would see a primary payload being launched no earlier than 18 months after the contract is signed. If the client is willing to pay for it, then the time between the signing of the contract and the launch would be reduced.

[DanClemmensen]

What do we learn from the mid-2025 launch dates?  Why an 18 month delay between now and these launches?  Does it take that long to figure out a payload adapter for these satellites that will work on a Falcon 9?  Is Amazon that far away from being able to deliver their satellites in bulk?  Did SpaceX claim "we're going to give schedule priority for Star Link over new customers until then."?

Rephrasing my question, if schedule concerns are looming for Kuiper, why aren't these contracts for 2024, what's preventing a faster roll-out?

Other possibilities (pure speculation, no actual information):
  --Kuiper needs a few satellites in polar orbits and ULA will not be able to launch from VFSB soon enough for logistical reasons.
  --Their spreadsheets show a likely gap in the availabilties of their current LVs at that time.

[mandrewa]

  --Their spreadsheets show a likely gap in the availabilties of their current LVs at that time.

That is what I have been imagining.  Amazon has a model for satellite deployment.  It's based on some rather optimistic assumptions possibly. But even with those optimistic assumptions, the model is telling them they will not have enough satellites in orbit by the nominal deadline.

That's why they have ordered three launches of the Falcon 9.

It was the only choice if they take their model seriously.

[catdlr]

  --Their spreadsheets show a likely gap in the availabilties of their current LVs at that time.

That is what I have been imagining.  Amazon has a model for satellite deployment.  It's based on some rather optimistic assumptions possibly. But even with those optimistic assumptions, the model is telling them they will not have enough satellites in orbit by the nominal deadline.

That's why they have ordered three launches of the Falcon 9.

It was the only choice if they take their model seriously.

Also are they under some deadline to get their stats up and running to fulfill some FCC requirements?  I'm not sure, just asking.

[mandrewa]

  --Their spreadsheets show a likely gap in the availabilties of their current LVs at that time.

That is what I have been imagining.  Amazon has a model for satellite deployment.  It's based on some rather optimistic assumptions possibly. But even with those optimistic assumptions, the model is telling them they will not have enough satellites in orbit by the nominal deadline.

That's why they have ordered three launches of the Falcon 9.

It was the only choice if they take their model seriously.

Also are they under some deadline to get their stats up and running to fulfill some FCC requirements?  I'm not sure, just asking.

From Wikipedia:

"Under its granted FCC license, Amazon is required to launch and operate 50% of its satellites no later than July 30, 2026, and must launch and operate the remaining satellites no later than July 30, 2029.[5]"

Edit: I suspect that even if the FCC does grant Amazon an extension, part of that process will have been demonstration of a good faith effort to reach the goal in the first place.  So if their model says they have to do this, then they have to do it.

[DanClemmensen]

  --Their spreadsheets show a likely gap in the availabilties of their current LVs at that time.

That is what I have been imagining.  Amazon has a model for satellite deployment.  It's based on some rather optimistic assumptions possibly. But even with those optimistic assumptions, the model is telling them they will not have enough satellites in orbit by the nominal deadline.

That's why they have ordered three launches of the Falcon 9.

It was the only choice if they take their model seriously.

IMO it's still not a serious model. It appears to be a wildly optimistic "model" that assumes that all of the new LVs will meet their schedules for initial launch and will then instantly ramp up to their purely speculative launch rates.

Of course, it it were any LV except F9 in the last 40 years, I doubt they could get launches in this "short" (less than 18 month) timeframe. The only exception might have been Soyuz, but they are now embargoed. We are all pretending that SpaceX has infinite capacity to accomodate customers, but there has to be a limit somewhere, and someone (RUAG, I think, now called Beyond Gravity) must design and build an F9 variant of the Kuiper dispenser.

[Asteroza]

  --Their spreadsheets show a likely gap in the availabilties of their current LVs at that time.

That is what I have been imagining.  Amazon has a model for satellite deployment.  It's based on some rather optimistic assumptions possibly. But even with those optimistic assumptions, the model is telling them they will not have enough satellites in orbit by the nominal deadline.

That's why they have ordered three launches of the Falcon 9.

It was the only choice if they take their model seriously.

IMO it's still not a serious model. It appears to be a wildly optimistic "model" that assumes that all of the new LVs will meet their schedules for initial launch and will then instantly ramp up to their purely speculative launch rates.

Of course, it it were any LV except F9 in the last 40 years, I doubt they could get launches in this "short" (less than 18 month) timeframe. The only exception might have been Soyuz, but they are now embargoed. We are all pretending that SpaceX has infinite capacity to accomodate customers, but there has to be a limit somewhere, and someone (RUAG, I think, now called Beyond Gravity) must design and build an F9 variant of the Kuiper dispenser.

Dispensor shouldn't be an issue if it's EELV payload adapter based, just reuse the Vulcan one as a basis.

The more interesting fact of all of this is that their model shows the shortfall, which it didn't before. This means as a customer, they have definitively received new information that Vulcan, New Glenn, AND Ariane 6 will no hit their launch targets, in some combination, and sutffed that into the model to get this result. Your guess is as good as mine, but is this all 3 running late together, or is one in particular now running much later, enough to trigger the model needing additional lift from SpaceX in 2025?

[RedLineTrain]

So if their model says they have to do this, then they have to do it.

True, and this is a crazy way to run this business.  Musk is stealing as many marches he can while Amazon is playing footsie with a regulatory deadline.

[DanClemmensen]

IMO it's still not a serious model. It appears to be a wildly optimistic "model" that assumes that all of the new LVs will meet their schedules for initial launch and will then instantly ramp up to their purely speculative launch rates.

The more interesting fact of all of this is that their model shows the shortfall, which it didn't before. This means as a customer, they have definitively received new information that Vulcan, New Glenn, AND Ariane 6 will no hit their launch targets, in some combination, and sutffed that into the model to get this result. Your guess is as good as mine, but is this all 3 running late together, or is one in particular now running much later, enough to trigger the model needing additional lift from SpaceX in 2025?

If they were originally using a hyper-optimistic model assuming that no vendor would slip, then it only requires any one vendor to identify a slip to Amazon in order to trigger this order for F9s.  I think the obvious conjecture for spectators like us is that BO is slipping NG. This is consistent with the recent upheavals at BO and the presumed unofficial information path from BO back into Amazon. It's also pure and unsupported speculation.

[mn]

The deadline keeps being brought up as if it's in the constitution or something.

The question comes up and has been discussed in the past, up thread. It seems to me pretty clear (based on previous discussion, not from any outside info) that they can ask for an extension and will most likely get it, (assuming that by the time the deadline comes around they will already be launching regularly, so no matter how far behind they are, as long as they can show a clear path to meet a new date they will most likely get an extension). At least that is what I understand from previous rounds of discussion.

Time to market is perhaps a bigger issue for them. (Unless they are convinced they can undercut everyone no matter how late they are to the market, or that the market is bigger than can be served by the existing players)

[DanClemmensen]

Time to market is perhaps a bigger issue for them. (Unless they are convinced they can undercut everyone no matter how late they are to the market, or that the market is bigger than can be served by the existing players)

They may also have a captive market, namely the AWS servers. They can play games with integrating AWS service for customers who use their consumer-side dishes, and shift the pricing around so the satellite part is competitive with Starlink. Recall that their big network-side earthstations will be co-located with their AWS server farms, which presumably already have enormous connectivity to "the Internet" (i.e., to other tier-1 Internet providers). SpaceX did not already have a massive terrestrial Internet infrastructure.

It's still hard to see how they can catch up with Starlink for the typical customer, but we'll see.

[mn]

Time to market is perhaps a bigger issue for them. (Unless they are convinced they can undercut everyone no matter how late they are to the market, or that the market is bigger than can be served by the existing players)

They may also have a captive market, namely the AWS servers. They can play games with integrating AWS service for customers who use their consumer-side dishes, and shift the pricing around so the satellite part is competitive with Starlink. Recall that their big network-side earthstations will be co-located with their AWS server farms, which presumably already have enormous connectivity to "the Internet" (i.e., to other tier-1 Internet providers). SpaceX did not already have a massive terrestrial Internet infrastructure.

It's still hard to see how they can catch up with Starlink for the typical customer, but we'll see.

The AWS angle was discussed extensively upthread, I don't see a use case that will make sense, but time will tell. (I guess if they play shenanigans where you get a better price on A if you buy B even if that relationship is not supported by reality, then they can pretty much do whatever they want. They could include Kuiper with Amazon Prime if they want - good luck with that, all customers that are too rural to have good ground based broadband (and by definition are also costly to deliver to) will get Amazon Prime and Kuiper and they can lose money on both sides - and make it up in volume).

[Zed_Noir]

..... and someone (RUAG, I think, now called Beyond Gravity) must design and build an F9 variant of the Kuiper dispenser.

Maybe not, in the last Transporter launch SpaceX roll out a new dispenser system that seem to be able to accept a wide range of payloads. Amazon should seek out the cheapest dispenser vendor. If that is SpaceX, so be it.

[TrevorMonty]

It's still hard to see how they can catch up with Starlink for the typical customer, but we'll see.

Why do they need a bigger customer base than Starlink. AWS just needs a enough customers (internal or external) to make constellation profitable.

[DanClemmensen]

It's still hard to see how they can catch up with Starlink for the typical customer, but we'll see.

Why do they need a bigger customer base than Starlink. AWS just needs a enough customers (internal or external) to make constellation profitable.

True, but I think that with their high launch costs and late entry in the market, they will need higher prices than Starlink. They can charge higher prices if they have some service that Starlink cannot provide, but that service (e.g. faster access to AWS) must be valuable to enough customers for them to make a profit.

I had not really thought about their internal customers, but there are a whole lot of them, like every Amazon delivery truck.

[Robotbeat]

I actually think that SpaceX is getting a pretty good margin with Starlink subscription costs, so it isn’t obvious to me that Kuiper is going to have to charge more. I think in the near term, they may have to charge significantly LESS to get traction. For very high margin customers like marine and aviation, there’s probably a lot more room for kuiper to compete

[DanClemmensen]

I actually think that SpaceX is getting a pretty good margin with Starlink subscription costs, so it isn’t obvious to me that Kuiper is going to have to charge more. I think in the near term, they may have to charge significantly LESS to get traction. For very high margin customers like marine and aviation, there’s probably a lot more room for kuiper to compete

What keeps Starlink from lowering its price in this scenario? Classic: a de facto monopoly sees competition and lowers it price. It happens in fixed comms a lot, which is why monopolists fight so hard to legislate enforcement of the monopoly.