Kuiper: Can it still meet its launching deadline?

[RedLineTrain]

SpaceX is taking care of it's customers at the expense of Starlink launches. Oneweb was a special case of not wanting to take advantage of a bad situation, but I'm not sure if they'll be too happy about sacrificing Starlink launch opportunities to loft Kupiers.

Falcon 9's capacity constraints don't seem to me to be very hard and fast.  SpaceX still has its doors wide open to new business.

https://twitter.com/thesheetztweetz/status/1636094419987251209

[meekGee]

Except for the 2026 date, I'd disagree. Terran-R, Neutron, and Antares/Firefly can all scale. All have better chance of F9-like low cost reusability than Vulcan. But they have first launch dates of 2026, 2024, and 2025, respectively. There's a decent chance one of them could actually beat New Glenn to first launch, given Blue Origin's history and pace.

Vulcan can kinda scale, at least compared to its competition, based on ULA experience base and brute force - at least initially.  But yeah, not long term.

As for the 2026 deadline, that's only half the problem Kuiper has.  The other is Starlink.  If Kuiper doesn't catch up, people won't be as generous as governments are in accepting a far-inferior second-best only because "we must have two providers".

Right now Starlink 2.0 might start launching before Kuiper demo sats do. Let that sink in.  Kuiper might go up against Starlink 3.0.

What will Kuiper do if Starlink service is half the price and twice the bandwidth?  This is not the "Microsoft vs Apple" dynamic. It is also not something Bezos can (or wants?) to bankroll.

[tssp_art]

Right now Starlink 2.0 might start launching before Kuiper demo sats do. Let that sink in.  Kuiper might go up against Starlink 3.0.

What will Kuiper do if Starlink service is half the price and twice the bandwidth?  This is not the "Microsoft vs Apple" dynamic. It is also not something Bezos can (or wants?) to bankroll.

That last part is actually quite complicated. Bezos is not the owner of Amazon - he's actually not even the largest shareholder (he holds about 10%). What that means is that the Amazon board and executive team have a fiduciary obligation to all the shareholders to maximize the value of their shares - and not to support the dreams or grudges held by their founder.

So yes, they could appeal to SpaceX for help and that might save the day - there is already some background noise about shareholder discontent that the multiple launch contracts did not include SpaceX. If they had used SpaceX  their prototype satellites would have been launched this past spring and they might be on their way to offering a credible service.

But even if SpaceX is willing, it will likely be with Falcon, because all available Starship flights will be used "experimentally" to develop and test tankers and depots, send Lunar Starship prototypes to the moon, and, of course, to launch Starlink (they have their own FCC deadlines to meet).

Or, Amazon could realize Kuiper is simply not doable with the disadvantages they have and (a) cancel it or (b) sell it to Bezos. Not sure which would be more entertaining.

[meekGee]

Right now Starlink 2.0 might start launching before Kuiper demo sats do. Let that sink in.  Kuiper might go up against Starlink 3.0.

What will Kuiper do if Starlink service is half the price and twice the bandwidth?  This is not the "Microsoft vs Apple" dynamic. It is also not something Bezos can (or wants?) to bankroll.

That last part is actually quite complicated. Bezos is not the owner of Amazon - he's actually not even the largest shareholder (he holds about 10%). What that means is that the Amazon board and executive team have a fiduciary obligation to all the shareholders to maximize the value of their shares - and not to support the dreams or grudges held by their founder.

So yes, they could appeal to SpaceX for help and that might save the day - there is already some background noise about shareholder discontent that the multiple launch contracts did not include SpaceX. If they had used SpaceX  their prototype satellites would have been launched this past spring and they might be on their way to offering a credible service.

But even if SpaceX is willing, it will likely be with Falcon, because all available Starship flights will be used "experimentally" to develop and test tankers and depots, send Lunar Starship prototypes to the moon, and, of course, to launch Starlink (they have their own FCC deadlines to meet).

Or, Amazon could realize Kuiper is simply not doable with the disadvantages they have and (a) cancel it or (b) sell it to Bezos. Not sure which would be more entertaining.

In theory yes, but then, how come the Kuiper test sats are on a future first flight of a BO powered rocket instead of an F9 transporter flight and already in orbit?

[DanClemmensen]

In theory yes, but then, how come the Kuiper test sats are on a futirw first flight of a BO powered rocket instead of an F9 transporter flight and already in orbit?

If launching the two test satellites is actually on the critical path for Kuiper, then they can launch on one of the nine Atlas V launchers that they contracted with ULA to provide.  Since they have not chosen to do this, I conclude that the test launch is not on the critical path. So what's the gating item for Kuiper? the satellites themselves? Ground infrastructure for the test? Software? Other?

[meekGee]

In theory yes, but then, how come the Kuiper test sats are on a futirw first flight of a BO powered rocket instead of an F9 transporter flight and already in orbit?

If launching the two test satellites is actually on the critical path for Kuiper, then they can launch on one of the nine Atlas V launchers that they contracted with ULA to provide.  Since they have not chosen to do this, I conclude that the test launch is not on the critical path. So what's the gating item for Kuiper? the satellites themselves? Ground infrastructure for the test? Software? Other?

That's a leap. I don't see how tests sats like this are NOT on the critical path.

It think that changing the assignment of one of only nine remaining Atlas rockets was just too much.

I think Kuiper are actually a lot closer to BO than some folks here think they are and are locked on Vulcan as opposed to F9.

Just judging by their actions.

[DanClemmensen]

In theory yes, but then, how come the Kuiper test sats are on a futirw first flight of a BO powered rocket instead of an F9 transporter flight and already in orbit?

If launching the two test satellites is actually on the critical path for Kuiper, then they can launch on one of the nine Atlas V launchers that they contracted with ULA to provide.  Since they have not chosen to do this, I conclude that the test launch is not on the critical path. So what's the gating item for Kuiper? the satellites themselves? Ground infrastructure for the test? Software? Other?

That's a leap. I don't see how tests sats like this are NOT on the critical path.

It think that changing the assignment of one of only nine remaining Atlas rockets was just too much.

I think Kuiper are actually a lot closer to BO than some folks here think they are and are locked on Vulcan as opposed to F9.

Just judging by their actions.

I think the test is on the critical path. I'm questioning whether or not the launch on Vulcan Centaur is on the critical path for the test. If the test cannot proceed until early next year for some other reason, then the Vulcan Centaur launch is not on the critical path.

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

[meekGee]

In theory yes, but then, how come the Kuiper test sats are on a futirw first flight of a BO powered rocket instead of an F9 transporter flight and already in orbit?

If launching the two test satellites is actually on the critical path for Kuiper, then they can launch on one of the nine Atlas V launchers that they contracted with ULA to provide.  Since they have not chosen to do this, I conclude that the test launch is not on the critical path. So what's the gating item for Kuiper? the satellites themselves? Ground infrastructure for the test? Software? Other?

That's a leap. I don't see how tests sats like this are NOT on the critical path.

It think that changing the assignment of one of only nine remaining Atlas rockets was just too much.

I think Kuiper are actually a lot closer to BO than some folks here think they are and are locked on Vulcan as opposed to F9.

Just judging by their actions.

I think the test is on the critical path. I'm questioning whether or not the launch on Vulcan Centaur is on the critical path for the test. If the test cannot proceed until early next year for some other reason, then the Vulcan Centaur launch is not on the critical path.

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

The satellites are ready for launch, no?  What else can hold up the test? Ground infrastructure?  I doubt that.
For all of the above reasons, Kuiper should have been pushing on those tests as if its life depended on them...

[DanClemmensen]

The satellites are ready for launch, no?  What else can hold up the test? Ground infrastructure?  I doubt that.
For all of the above reasons, Kuiper should have been pushing on those tests as if its life depended on them...

Ground infrastructure for a two-satellite test of an LEO constellation is a non-trivial proposition. To make a connection, you need a user terminal and a "teleport" in view of the satellites at the same time. The user terminal must be in the satellite's user beam and the teleport must be in the satellite's teleport beam. You use two satellites in the same orbit, spaced properly, so you can test handoffs, and (if implemented) intersatellite comms. But the orbit is almost certainly not synchronized in any way, so the satellites will not pass over the same locations every day and probably not even once a week. Therefore, you need multiple teleports and multiple user terminals to get any reasonable amount of testing done, and each test is only a few minutes long. If you are using polar orbits, you can cheat somewhat by testing at high latitudes, but it's still a major hassle.

We are in agreement: they should be testing already, so why aren't they?

[TrevorMonty]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

[Rebel44]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

/disclaimer: I work in telco - transmission (backbone) networks/

I seriously doubt that - the bandwidth Kuipers terminals are likely to provide (especially uplink) is likely to be only good enough to serve as a backup. Connections between data centers and big corporate customers are usually over optical fibers for good reasons - not for fun.

IMO, good encryption should also allow data to be reasonably secure (unless someone like the NSA is after you - in which case you are screwed anyway...) for commercial use purposes even on lines you don't physically control.

[joek]

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.

Depends on what you mean by "secure"?
1. Secure from snooping? No need. Existing transport security measures used by terrestial networks are more than sufficient.
2. Resilent-robust? Yes, some potential value add as a backup for existing terrestial networks.
This is not going to replace existing terrestial networks and ther Big Pipes as primary transport any time soon for those "AWS servers and their big customers". Neither will Starlink.

Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

Yes, just like any other ISP or backbone provider. But think you just contradicted yourself there. We talking about business or consumer market? Amazon is generally consumer focused; AWS business focused. So which is it?

This isn't a market Starlink can access without Amazon's approval.

Absolutely false. Consult Dr. Google for "Internet peering". It's been a thing for decades.[1] It would be stupendously stupid for AWS to disallow.


[1] edit That's why it's called the "Internet" (coined 1974), aka "Inter-net", aka "internetworking",  aka "a network of networks".

[DanClemmensen]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

/disclaimer: I work in telco - transmission (backbone) networks/

I seriously doubt that - the bandwidth Kuipers terminals are likely to provide (especially uplink) is likely to be only good enough to serve as a backup. Connections between data centers and big corporate customers are usually over optical fibers for good reasons - not for fun.

IMO, good encryption should also allow data to be reasonably secure (unless someone like the NSA is after you - in which case you are screwed anyway...) for commercial use purposes even on lines you don't physically control.

If all "customers" are AWS server farms, then all the links are in effect teleport links. Each satellite serves only a few links and each ground station/server farm has antennas for multiple satellites. Lots of bandwidth, all full duplex non-shared links. Probably still not as much as a terrestrial fiber, but many times as much as a Starlink customer. The huge advantage is lower latency. The RF and laser links operate at the speed of light in vacuum, which is about 300,000 km/s, and the links are straight lines. Fiber operates at the speed of light in fiber, which is about 200,000 km/s, and fiber cables are not straight at all, because they follow terrestrial rights-of-way and undersea routes that dodge around undersea topology and continents. AWD will move massive bulk data by fiber and certain premium data by satellite.

[TrevorMonty]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

/disclaimer: I work in telco - transmission (backbone) networks/

I seriously doubt that - the bandwidth Kuipers terminals are likely to provide (especially uplink) is likely to be only good enough to serve as a backup. Connections between data centers and big corporate customers are usually over optical fibers for good reasons - not for fun.

IMO, good encryption should also allow data to be reasonably secure (unless someone like the NSA is after you - in which case you are screwed anyway...) for commercial use purposes even on lines you don't physically control.

If all "customers" are AWS server farms, then all the links are in effect teleport links. Each satellite serves only a few links and each ground station/server farm has antennas for multiple satellites. Lots of bandwidth, all full duplex non-shared links. Probably still not as much as a terrestrial fiber, but many times as much as a Starlink customer. The huge advantage is lower latency. The RF and laser links operate at the speed of light in vacuum, which is about 300,000 km/s, and the links are straight lines. Fiber operates at the speed of light in fiber, which is about 200,000 km/s, and fiber cables are not straight at all, because they follow terrestrial rights-of-way and undersea routes that dodge around undersea topology and continents. AWD will move massive bulk data by fiber and certain premium data by satellite.

Terrestrial fibre paths typically have multiple exchanges which do add delays.

[TrevorMonty]

This isn't a market Starlink can access without Amazon's approval.

Absolutely false. Consult Dr. Google for "Internet peering". It's been a thing for decades.[1] It would be stupendously stupid for AWS to disallow.

[1] edit That's why it's called the "Internet" (coined 1974), aka "Inter-net", aka "internetworking",  aka "a network of networks".

AWS customers can use Starlink to access internet but if they want secure end to link using Starlink AWS has to connect  Starlink terminal to their server. Don't think there is any legal requirement for AWS to do this.

[joek]

AWS customers can use Starlink to access internet but if they want secure end to link using Starlink AWS has to connect  Starlink terminal to their server.

End-to-end security is done every day, all day, around the world, regardless of who is in the middle (other ISP's or backbone-transport providers). That's how you get secure transport from your pc > your ISP > any number of backbone providers > endpoint server without having any of those intermediaries having access to the content of your messages.

End-to-end means exactly that: it does not depend on the security of any intermediarias, such as Starlink terminals (the equivalent of your home router).

Don't think there is any legal requirement for AWS to do this.

No there is not. But AWS would be cutting off there nose to spite their face if they did otherwise. It's the Internet.

In any case, getting off topic. Suggest you do some research on what end-to-end security means and how it is implemented[1], the gfenesis of the Internet, and why there has been a stampede to connect to it among both business and individuals.


[1] Apologies, but you sound like a Bell-head from the 90's: "trust the network". Today (and for the last few decates) anyone in their right mind does not trust the network: we do end-to-end security. That is why, e.g., DoD uses public untrusted transport providers for much of their traffic.

[joek]

Terrestrial fibre paths typically have multiple exchanges which do add delays.

Whcih for most will not matter; it's going to be a difference of a few ms. If talking about high frequency trading, that's another matter and a niche market.

[DanClemmensen]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

/disclaimer: I work in telco - transmission (backbone) networks/

I seriously doubt that - the bandwidth Kuipers terminals are likely to provide (especially uplink) is likely to be only good enough to serve as a backup. Connections between data centers and big corporate customers are usually over optical fibers for good reasons - not for fun.

IMO, good encryption should also allow data to be reasonably secure (unless someone like the NSA is after you - in which case you are screwed anyway...) for commercial use purposes even on lines you don't physically control.

If all "customers" are AWS server farms, then all the links are in effect teleport links. Each satellite serves only a few links and each ground station/server farm has antennas for multiple satellites. Lots of bandwidth, all full duplex non-shared links. Probably still not as much as a terrestrial fiber, but many times as much as a Starlink customer. The huge advantage is lower latency. The RF and laser links operate at the speed of light in vacuum, which is about 300,000 km/s, and the links are straight lines. Fiber operates at the speed of light in fiber, which is about 200,000 km/s, and fiber cables are not straight at all, because they follow terrestrial rights-of-way and undersea routes that dodge around undersea topology and continents. AWD will move massive bulk data by fiber and certain premium data by satellite.

Terrestrial fibre paths typically have multiple exchanges which do add delays.

An LEO satellite system with inter-satellite links also has multiple "exchanges" that add delays, namely the satellites. Moderns routers (or any other type of packet forwarder), either satellite or terrestrial, adds very small switching delay. There may be queueing delay of course. To a first approximation I was GUESSING that the number of satellite hops will be about the same as the number of fiber hops.

[TrevorMonty]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

/disclaimer: I work in telco - transmission (backbone) networks/

I seriously doubt that - the bandwidth Kuipers terminals are likely to provide (especially uplink) is likely to be only good enough to serve as a backup. Connections between data centers and big corporate customers are usually over optical fibers for good reasons - not for fun.

IMO, good encryption should also allow data to be reasonably secure (unless someone like the NSA is after you - in which case you are screwed anyway...) for commercial use purposes even on lines you don't physically control.

If all "customers" are AWS server farms, then all the links are in effect teleport links. Each satellite serves only a few links and each ground station/server farm has antennas for multiple satellites. Lots of bandwidth, all full duplex non-shared links. Probably still not as much as a terrestrial fiber, but many times as much as a Starlink customer. The huge advantage is lower latency. The RF and laser links operate at the speed of light in vacuum, which is about 300,000 km/s, and the links are straight lines. Fiber operates at the speed of light in fiber, which is about 200,000 km/s, and fiber cables are not straight at all, because they follow terrestrial rights-of-way and undersea routes that dodge around undersea topology and continents. AWD will move massive bulk data by fiber and certain premium data by satellite.

Terrestrial fibre paths typically have multiple exchanges which do add delays.

An LEO satellite system with inter-satellite links also has multiple "exchanges" that add delays, namely the satellites. Moderns routers (or any other type of packet forwarder), either satellite or terrestrial, adds very small switching delay. There may be queueing delay of course. To a first approximation I was GUESSING that the number of satellite hops will be about the same as the number of fiber hops.

Latency from delays is very important in gaming where 10s ms can make difference at other end of spectrum a 1seconds latency doesn't matter to much when downloading large files or watching a video. With large files and videos datarate is more important.

[DanClemmensen]

The alternative theory seems to be that the test could occur now, but Kuiper is waiting on Vulcan Centaur for reasons that do not appear to make economic sense. They are losing possible customers to Starlink at an increasing rate every month.

Kuiper primary mission is to supply a secure datalink between AWS servers and their big customers. This isn't a market Starlink can access without Amazon's approval.
Any other business is a bonus. Starlink customers will change providers if they get a better deal or are frakked off with existing service.

/disclaimer: I work in telco - transmission (backbone) networks/

I seriously doubt that - the bandwidth Kuipers terminals are likely to provide (especially uplink) is likely to be only good enough to serve as a backup. Connections between data centers and big corporate customers are usually over optical fibers for good reasons - not for fun.

IMO, good encryption should also allow data to be reasonably secure (unless someone like the NSA is after you - in which case you are screwed anyway...) for commercial use purposes even on lines you don't physically control.

If all "customers" are AWS server farms, then all the links are in effect teleport links. Each satellite serves only a few links and each ground station/server farm has antennas for multiple satellites. Lots of bandwidth, all full duplex non-shared links. Probably still not as much as a terrestrial fiber, but many times as much as a Starlink customer. The huge advantage is lower latency. The RF and laser links operate at the speed of light in vacuum, which is about 300,000 km/s, and the links are straight lines. Fiber operates at the speed of light in fiber, which is about 200,000 km/s, and fiber cables are not straight at all, because they follow terrestrial rights-of-way and undersea routes that dodge around undersea topology and continents. AWD will move massive bulk data by fiber and certain premium data by satellite.

Terrestrial fibre paths typically have multiple exchanges which do add delays.

An LEO satellite system with inter-satellite links also has multiple "exchanges" that add delays, namely the satellites. Moderns routers (or any other type of packet forwarder), either satellite or terrestrial, adds very small switching delay. There may be queueing delay of course. To a first approximation I was GUESSING that the number of satellite hops will be about the same as the number of fiber hops.

Latency from delays is very important in gaming where 10s ms can make difference at other end of spectrum a 1seconds latency doesn't matter to much when downloading large files or watching a video. With large files and videos datarate is more important.

True, but the claim here is that Kuiper will be used for interconnect between AWS server farms, not from user to server. I was describing the difference between LEO and fiber delays. In this context, bulk transfer can use fiber, and the only truly delay-sensitive traffic that I know of is remote server synchronization traffic.

Some context: for RF and free-space laser, speed of light is 300,000 km/s and transit delay is one ms for each 300 km. Switching delays can be a low as one microsecond if using ASICs. Transmission delay is one microsecond to send a 10,000 bit packet at 10 Gbps. Queueing delays vary depending on load. Unless your net is grossly overloaded, transit delay will dominate.

This is dramatically different than it was when I got into this business in 1971, when nobody cared about transit delay unless you were using GEO satellites. Of course, we were using 2400 bps lines at the time, so transmission delay dominated.