Amazon Project Kuiper Broadband Constellation

[ncb1397]

Bezos had more money than god when Musk predicted that "Unicorns will dance in the flame trench if Bezos has a man-rated rocket within 5 years".  That date passed quietly last year, and NG is still more than a year away. 

Musk didn't meet the 5 year time frame either. But, no, Bezos's financial resources when that was said wasn't comparable to Amazon's financial resources now. Amazon's annual earnings now are similar to his net worth back then. But that is beside the point. The point was if competitors think buying rockets from one company or another will hurt their competitive position by breeding competitors (or healthier competitors). For Amazon, a $1 billion order from OneWeb makes very little difference. That is revenue of about a day. For SpaceX, it is more like 6 months.

Short lifespan is a feature, not a bug. It means upgrading capabilities happens quickly in a routine manner. The SV way is MVP and then iterate like craazy.

No, you can upgrade in a responsible matter either way. And if your projections for the future capabilities of satellites doesn't come to pass in the <3 year time frame, you can choose not to replace at that time and pass the savings onto your customers (to the detriment of your competitors). Constellation operators that have satellites that last 2-3x as long will have a significant advantage. They can even keep multiple generations of satellites operational at the same time increasing the total satellite capacity available to them. The one that wins the constellation wars may be the company that just builds a global constellation, and then takes a capex holiday where their satellite constructions costs go to zero, their launch costs go to zero. This is how Iridium was able to afford Iridium Next, if they were forced to do that every 3 years, they would have gone bankrupt (again).

[Ronsmytheiii]

The constellation business will become a race. SpaceX has a much more capable launcher (either now, or in 4 years).  This allows them to push a more capable constellation, with lower costs.

OneWeb and Amazon can't compete with that, except for making sure they can capitalize on a potential mistake by SpaceX.

Not true, Amazon only cares about expanding its services (ie Amazon Prime, AWS, it’s regular marketplace ect) The constellation is a means to an end, and is more an infrastructure cost to Amazon itself. Vastly expanding Amazon’s consumer reach is well worth the cost, as its cloud storage business is much more lucrative (and in that case, competitors are Google and Microsoft, not SpaceX or OmeWeb.)

[noogie]

I wonder if Amazon can close the business case for this satellite constellation purely on their own internal demand from AWS, their delivery fleet and future autonomous endeavours (drones, etc)
If the main aim is for internal use, then it makes it a pretty audacious vertical integration move.
Will this mean Google is more aggressive at using Starlink (as they are an investor in SpaceX), will other cloud providers also feel the need to team up with another LEO satellite constellation?

[oldAtlas_Eguy]

Hmmm
Cloud services seeking global low latency private networks.

[TrevorMonty]

Well Amazon/Blue Origin is getting a late start.  We have Alexa's Cubes, Fire TV, etc.  Alexa's search engine is very poor compared to Google.  If Starlink uses Google, it will be no contest unless Amazon Alexa greatly improves.

You are good at highlightly their not so successfu products but overlook the most successful one.
Amazon Web Services was first of its type  and dominates market at 32%, twice its nearest competitor. With revenue of $25b in 2018, it accounts for large chunk of Amazon group profit. AWS can afford to fund this constellation on its own and may well be its biggest customer given how much a year they must spend on internet data between their data centres.

[TrevorMonty]

Better internet security is also big plus with these constellations. Currently data between AWS servers and customers goes via internet that takes which ever route internet switches decide. Not great way to guarantee data security. With constellation all data between servers and big customers can go via secure satellite constellation, with bonus of low latency.
This especially important with government agencies, big corporates and financial institutions.

[ncb1397]

Better internet security is also big plus with these constellations. Currently data between AWS servers and customers goes via internet that takes which ever route internet switches decide. Not great way to guarantee data security. With constellation all data between servers and big customers can go via secure satellite constellation, with bonus of low latency.
This especially important with government agencies, big corporates and financial institutions.

Why not just encrypt the AWS data with the same protocol you use to encrypt the wireless signals between the satellites and the ground?

[Robotbeat]

...
That said, I also wonder how this is going to impact the desire of commsat companies to launch on Blue Origin in the future. Now one of the big things going for them relative to SpaceX (that launching on them wasn't funding a competitor) is no longer true.

~Jon

I suspect eventually people will stop caring. There'll be Starlink, OneWeb, Telesat, and Kuiper up there, and people will just contract with whoever's cheapest and responsive (while being sure to keep your options option by booking a few on other launch providers.

I say, the more the merrier! The killer app for RLVs is these megaconstellations. We need at least two healthy RLVs, so more of these birds is good. Glad they're shooting for 600ish km altitude instead of 1000ish km, though. If something really bad happens at least in a few decades the debris will be basically gone. Ideally, it'd be even lower so that debris enters ~immediately.

[Ultrafamicom]

Any of these constellations will be requiring a constant launch cadence.  1/week or more.

Really? 250 kg satellites, 100,000 kg per launch, 400 satellites per launch. 10 launches for deployment. They get replaced every 2.5 months?

edit: even 500 kg satellites, 45,000 kg per launch, 90 satellites per launch, 3236 satellites would be 36 launches or replacing the constellation nearly twice a year at 1 per week.

Maybe I'm behind the times but I remember 45,000 kg per launch, so yeah, 90 at most.  Less the launch mount, which is not negligible.  You're also constrained by orbital planes, so OOM, you're delivering one orbital plane per launch, give or take.

For smaller constellations (like the proposed Amazon one), 20-30 planes?
But for Starlink, 3x as many?  (I don't have the official count)

If lifetime is 3 years, you have to replace 33% of the constellation every year, after a 3-year build-up.
Or, you try to accelerate the build-up by launching even more often.

That's where launch capacity and launch cost will really matter.

Bezos is doing all he can to make sure that if SpaceX slips, he'll be ready to take advantage of it.  But if SpaceX doesn't slip, I think he's going to have a really hard time keeping up.

No.

I think it had been mentioned many times here, that because of precession effect  caused by equatorial bulge, satellite can drift between different planes easily. Even iridium doesn't launch one plane a time.

Iridium didn't have much choice, not enough satellites per plane.  Changing planes using precession takes time, and these mega-constellation satellites don't have much lifetime to waste.

I don't know for sure, but it's a safe guess that the preferred method will still be one plane per launch, and any orbital spares similarly in-plane.

The down side will be that you don't get partial coverage when you're partially deployed, so I can see perhaps a compromise where two half-planes are deployed with one launch. Maybe.  Unless the number of satellites is large enough that it takes two launches to get one plane populated anyway, and then the problem is solved.

Mega constellations don't have choice as well, most satellites will be launched to orbits lower than working orbits to avoid creating junk in case of engine malfunction. In any case weeks of orbit raising before operation looks like unavoidable.

[meekGee]

Any of these constellations will be requiring a constant launch cadence.  1/week or more.

Really? 250 kg satellites, 100,000 kg per launch, 400 satellites per launch. 10 launches for deployment. They get replaced every 2.5 months?

edit: even 500 kg satellites, 45,000 kg per launch, 90 satellites per launch, 3236 satellites would be 36 launches or replacing the constellation nearly twice a year at 1 per week.

Maybe I'm behind the times but I remember 45,000 kg per launch, so yeah, 90 at most.  Less the launch mount, which is not negligible.  You're also constrained by orbital planes, so OOM, you're delivering one orbital plane per launch, give or take.

For smaller constellations (like the proposed Amazon one), 20-30 planes?
But for Starlink, 3x as many?  (I don't have the official count)

If lifetime is 3 years, you have to replace 33% of the constellation every year, after a 3-year build-up.
Or, you try to accelerate the build-up by launching even more often.

That's where launch capacity and launch cost will really matter.

Bezos is doing all he can to make sure that if SpaceX slips, he'll be ready to take advantage of it.  But if SpaceX doesn't slip, I think he's going to have a really hard time keeping up.

No.

I think it had been mentioned many times here, that because of precession effect  caused by equatorial bulge, satellite can drift between different planes easily. Even iridium doesn't launch one plane a time.

Iridium didn't have much choice, not enough satellites per plane.  Changing planes using precession takes time, and these mega-constellation satellites don't have much lifetime to waste.

I don't know for sure, but it's a safe guess that the preferred method will still be one plane per launch, and any orbital spares similarly in-plane.

The down side will be that you don't get partial coverage when you're partially deployed, so I can see perhaps a compromise where two half-planes are deployed with one launch. Maybe.  Unless the number of satellites is large enough that it takes two launches to get one plane populated anyway, and then the problem is solved.

Mega constellations don't have choice as well, most satellites will be launched to orbits lower than working orbits to avoid creating junk in case of engine malfunction. In any case weeks of orbit raising before operation looks like unavoidable.

The final orbits are pretty short lived, so the injection orbits will be pretty close to them IMO

[Ultrafamicom]

Any of these constellations will be requiring a constant launch cadence.  1/week or more.

Really? 250 kg satellites, 100,000 kg per launch, 400 satellites per launch. 10 launches for deployment. They get replaced every 2.5 months?

edit: even 500 kg satellites, 45,000 kg per launch, 90 satellites per launch, 3236 satellites would be 36 launches or replacing the constellation nearly twice a year at 1 per week.

Maybe I'm behind the times but I remember 45,000 kg per launch, so yeah, 90 at most.  Less the launch mount, which is not negligible.  You're also constrained by orbital planes, so OOM, you're delivering one orbital plane per launch, give or take.

For smaller constellations (like the proposed Amazon one), 20-30 planes?
But for Starlink, 3x as many?  (I don't have the official count)

If lifetime is 3 years, you have to replace 33% of the constellation every year, after a 3-year build-up.
Or, you try to accelerate the build-up by launching even more often.

That's where launch capacity and launch cost will really matter.

Bezos is doing all he can to make sure that if SpaceX slips, he'll be ready to take advantage of it.  But if SpaceX doesn't slip, I think he's going to have a really hard time keeping up.

No.

I think it had been mentioned many times here, that because of precession effect  caused by equatorial bulge, satellite can drift between different planes easily. Even iridium doesn't launch one plane a time.

Iridium didn't have much choice, not enough satellites per plane.  Changing planes using precession takes time, and these mega-constellation satellites don't have much lifetime to waste.

I don't know for sure, but it's a safe guess that the preferred method will still be one plane per launch, and any orbital spares similarly in-plane.

The down side will be that you don't get partial coverage when you're partially deployed, so I can see perhaps a compromise where two half-planes are deployed with one launch. Maybe.  Unless the number of satellites is large enough that it takes two launches to get one plane populated anyway, and then the problem is solved.

Mega constellations don't have choice as well, most satellites will be launched to orbits lower than working orbits to avoid creating junk in case of engine malfunction. In any case weeks of orbit raising before operation looks like unavoidable.

The final orbits are pretty short lived, so the injection orbits will be pretty close to them IMO

As far as I understand the main factor of choosing orbit is cover area rather than life, for example Oneweb has orbits siginificantly higher than Iridium.

[Ludus]

Is there any indication that the Kuiper Constellation has the ambition Starlink has for laser intersatellite links and becoming a new lower latency internet backbone? Or does it seem to be another One Web?

[meekGee]

Is there any indication that the Kuiper Constellation has the ambition Starlink has for laser intersatellite links and becoming a new lower latency internet backbone? Or does it seem to be another One Web?

IMO given Amazon's AWS business this is both expected and almost a foregone conclusion.

[Oli]

Is there any indication that the Kuiper Constellation has the ambition Starlink has for laser intersatellite links and becoming a new lower latency internet backbone? Or does it seem to be another One Web?

From the article:

“Project Kuiper is a new initiative to launch a constellation of low Earth orbit satellites that will provide low-latency, high-speed broadband connectivity to unserved and underserved communities around the world,” an Amazon spokesperson said in an emailed statement. “This is a long-term project that envisions serving tens of millions of people who lack basic access to broadband internet. We look forward to partnering on this initiative with companies that share this common vision.”

It's another OneWeb.

IMO given Amazon's AWS business this is both expected and almost a foregone conclusion.

I don't see why Amazon's AWS should have a preference for a particular infrastructure.

[meekGee]

IMO given Amazon's AWS business this is both expected and almost a foregone conclusion.

I don't see why Amazon's AWS should have a preference for a particular infrastructure.

You don't see why a data center and cloud hosting company would care about a wholly-controlled fast backbone?

[Oli]

IMO given Amazon's AWS business this is both expected and almost a foregone conclusion.

I don't see why Amazon's AWS should have a preference for a particular infrastructure.

You don't see why a data center and cloud hosting company would care about a wholly-controlled fast backbone?

Depends. If the margins in the infrastructure business are high, higher than what they can achieve by expanding their existing business, they might enter the market. I don't see why they have to control it.

[meekGee]

IMO given Amazon's AWS business this is both expected and almost a foregone conclusion.

I don't see why Amazon's AWS should have a preference for a particular infrastructure.

You don't see why a data center and cloud hosting company would care about a wholly-controlled fast backbone?

Depends. If the margins in the infrastructure business are high, higher than what they can achieve by expanding their existing business, they might enter the market. I don't see why they have to control it.

The question was whether they'd care which architecture the constellation that they've announced they want to build will use.

Once they're in, ISTM that of course they'll want a sat-to-sat backbone.

[Oli]

Once they're in, ISTM that of course they'll want a sat-to-sat backbone.

Their official statement sounds like OneWeb. I also don't think intersatellite links automatically imply backbone for the entire internet. In any case, it's pointless to speculate.

[Darkseraph]

Is there any indication that the Kuiper Constellation has the ambition Starlink has for laser intersatellite links and becoming a new lower latency internet backbone? Or does it seem to be another One Web?

Yes. On their notional roadmap, there is a node for "High Bandwidth Optical Communications". That does not conclusively prove they intent laser links in space between sats but it does indicate this is a direction they may go. And it fits very well with their other business, data centers.

[Oli]

Yes. On their notional roadmap...

 

That's BO's roadmap. Why is this thread in the BO section anyway?