Starlink : Markets and Marketing

[r8ix]

A lot more latency with GEO, so what advantages does GEO offer? They don’t/won’t have advantage of bandwidth, latency, or price, if/when Starlink is functional.

[Eka]

A lot more latency with GEO, so what advantages does GEO offer? They don’t/won’t have advantage of bandwidth, latency, or price, if/when Starlink is functional.

Simpler user terminals. A single transceiver with a single antenna at the focus of a parabola works nicely. No need for complex phase shifting electronics to drive many antennas. The interesting thing is once the phase shifting electronics is incorporated into a ASIC, the phased array antenna is easier to make. Yeah, I know sounds crazy, but it is a PCB and can be made using standard circuit board making equipment. User installation is also much easier.

[OTV Booster]

A lot more latency with GEO, so what advantages does GEO offer? They don’t/won’t have advantage of bandwidth, latency, or price, if/when Starlink is functional.

Streaming Videeo doesn’t care about latency. It’s inherently high bandwidth. Broadcast TV is dying and streaming video will pick up this slack. Cable is heading this way too.

SV delivery is sat, cable and a bit of DSL. Ignoring marketing ignorance brought on by last mile monopoly, SV doesn’t care and neither does the customer. If Pigeon Packet can give the bandwidth, it’s a go.

Looking at costs, and for the moment  ignoring SL’s obvious cost advantage in launch, a constellation of 41k sats at $150k (guesstimate) each will cost $6B. Three GEO megasats will be much less for only three launches and 2-3x the lifetime. With future launch costs showing every promise of dropping dramatically the equation morphs. Sats no longer need to be so robust... yada yada. And only three sats to manage.

Then there is the cost of the user equipment. SL looks to be inherently more expensive. Of course volume impacts this but that is as much a marketing issue as technical.

I can’t speak to bandwidth but I doubt GEO sats are anyway near their practical limit and I do not have a clear understanding of the impact of open ocean on SL’s theoretical throughput.

Again, from a strictly technical POV, SL offers low latency and maybe higher bandwidth. System costs, again in absolute numbers and ignoring the size of the customer base, looks to be higher for SL.

Where SL shows it’s promise is what this thread is discussing - markets and marketing.

Looking at the backbone, where latency is important, SL will hammer the competition. Gaming, VOIP and maybe most important, DoD. Where latency is less important, the growing streaming video market being central, it’s up to SX to be competitive.

On the last mile, latency is important. This gives SL a strong competitive advantage. Maybe more important to the last mile is SL’s hopeful willingness to not embrace embrace the sphincter culture so prevalent in internet delivery.

ISTM SL has some natural market advantages and some disadvantages. From a strictly technical POV, SL and GEO are complementary for the backbone. For the last mile, if SL plays nice and lets the other guy be the sphincter they have been in the past, we will have a fully funded mars settlement.

For the last mile technology is important but customer service is really important.

Phil

[TrevorMonty]

There are companies now using smallsats for GEO broadband, with lot lower setup cost. These are targetting specfic markets and locations eg Alaska.

https://spacenews.com/geo-smallsats-emerging-as-a-best-of-both-worlds-approach/

These GEO smallsats will become more capable and cheaper as likes of Made In Space enable them to have large solar arrays and antennas.

[tbellman]

Streaming Videeo doesn’t care about latency. It’s inherently high bandwidth. Broadcast TV is dying and streaming video will pick up this slack. Cable is heading this way too.

SV delivery is sat, cable and a bit of DSL. Ignoring marketing ignorance brought on by last mile monopoly, SV doesn’t care and neither does the customer. If Pigeon Packet can give the bandwidth, it’s a go.

Not quite true.  In practice, most delivery of streaming video use the TCP protocol (more specifically, HTTP or HTTPS, but they in turn use TCP), and TCP itself is sensitive to round-trip latency (often called RTT, round-trip time).  It is rather difficult to get high bandwidth through TCP when you have high RTT.  And if there is any contention (with resulting packet drops) along the path, high RTT tends to kill your usable bandwidth quite brutally.

(There is a new protocol called QUIC being developed and deployed, mainly by Google, to use instead of TCP.  It might be that it works better for high-RTT connections; I haven't read up on it enough.)

[launchwatcher]

Not quite true.  In practice, most delivery of streaming video use the TCP protocol (more specifically, HTTP or HTTPS, but they in turn use TCP), and TCP itself is sensitive to round-trip latency (often called RTT, round-trip time).  It is rather difficult to get high bandwidth through TCP when you have high RTT.  And if there is any contention (with resulting packet drops) along the path, high RTT tends to kill your usable bandwidth quite brutally.

Check out the BBR congestion control algorithms developed for TCP:

https://datatracker.ietf.org/meeting/104/materials/slides-104-iccrg-an-update-on-bbr-00

(There is a new protocol called QUIC being developed and deployed, mainly by Google, to use instead of TCP.  It might be that it works better for high-RTT connections; I haven't read up on it enough.)

IMHO QUIC's main evolutionary advantage over TCP is that it's easier to deploy algorithmic improvements to the clients (browser patch vs OS patch) but the same advanced congestion control and pacing algorithms can be used with both protocols.   (Google is using BBR with both protocols).

[indaco1]

..In practice, most delivery of streaming video use the TCP protocol ..

Most delivery of streaming use CDNs, ie the content is delivered to a certain area just once (perhaps using a proprietary protocol and bypassing network backbone at all) then it's cached in a local proxy server.

GEO satellites utilization factor could be much better because they are always active unlike LEO satellites that are on unpopulated areas most of time.  On the other hand launch to GEO is much more expensive, so I don't know if they will make sense even for uses not sensitive to latency.

[wannamoonbase]

Would it make sense, do develop buzz, that during partial coverage to have customers buy an antenna to use, but not charge for service?

There are enough nerds like us out here that would take partial coverage and play with the systems and post results and videos.

Occurred to me that a conventional company might not do this.  But that Musk might want to do something like this for free media as well as to start standing up the service.

Also, when the go live date came, you’d have some people you could start billing right away.

[RedLineTrain]

Would it make sense, do develop buzz, that during partial coverage to have customers buy an antenna to use, but not charge for service?

Don't think they need buzz, as the word-of-mouth on this is already incredible.  The potential retail customers for this are very motivated.

https://www.inverse.com/innovation/elon-musk-starlink-internet

In the past, Musk has given referral bonuses to encourage word-of-mouth, but I don't think he has ever not charged for service.  And his businesses have never advertised that I know of.

[tbellman]

..In practice, most delivery of streaming video use the TCP protocol ..

Most delivery of streaming use CDNs, ie the content is delivered to a certain area just once (perhaps using a proprietary protocol and bypassing network backbone at all) then it's cached in a local proxy server.

I'm well aware, but for GEO satellite links, that only helps if the cache is inbetween the satellite link and the user.  So it might be applicable to places like remote islands with no sub-sea fiber cables, several thousand users (so it is worth it for the CDN to deploy the cache) and low-latency connections within that area.  If the GEO satellite link goes to the end-user, the cache will still be half a second RTT away, and HTTP(S) over TCP will be the usual protocol.

(Also, local CDN caches, at least ones hosted at ISPs, are starting to get less popular, due to more and more traffic going over HTTPS instead of HTTP, and the domain-owners not being keen on putting their host certificates in places they don't control.)

[FutureSpaceTourist]

Fighting talk ...

https://twitter.com/thesheetztweetz/status/1237013140191301632

SpaceX VP Jonathan Hofeller, on Starlink satellite network pricing: Whatever OneWeb says, our price is less.

#SATShow

[ZachF]

Another thing that needs to be factored in when cost comparing large GEO birds and Starlink, is that large GEO birds are big capital projects, and debt servicing is going to impact the $ per GB figure potentially by a lot.

[RocketGoBoom]

Fighting talk ...

https://twitter.com/thesheetztweetz/status/1237013140191301632

SpaceX VP Jonathan Hofeller, on Starlink satellite network pricing: Whatever OneWeb says, our price is less.

#SATShow

Hofeller is confirming that our math on costs is probably accurate. SpaceX has a huge cost ($/Gbps) advantage over OneWeb.

[Asteroza]

We also discussed EchoStar / HughesNet and their Jupiter 3 satellite. The cost metrics for EchoStar are about 40% more expensive than Viasat's. I agreed with all of his numbers. Also, they are only doing one satellite with 500 Gbps compared to Viasat doing three satellites with 1,000 Gbps each. EchoStar is basically screwed and won't be financially competitive with Viasat or Starlink.

The CEO also made the point that Viasat has 1/3 of their business with the US govt, much of which is custom designed solutions with long term contracts. So Viasat has options for their capacity that EchoStar does not have. EchoStar is mostly a pure play on the satellite internet market and EchoStar is in the worst position competitively. EchoStar also has the most North American customers for Starlink to poach, about 2x more than Viasat.

There's an announcement that HughesNet is partnering with OneWeb now, so it  looks like they are trying to buy their way back into competitiveness. Will the conglomerate of OneWeb/Hugesnet(Echostar) put the screws on Viasat?

[ninjaneer]

We also discussed EchoStar / HughesNet and their Jupiter 3 satellite. The cost metrics for EchoStar are about 40% more expensive than Viasat's. I agreed with all of his numbers. Also, they are only doing one satellite with 500 Gbps compared to Viasat doing three satellites with 1,000 Gbps each. EchoStar is basically screwed and won't be financially competitive with Viasat or Starlink.

The CEO also made the point that Viasat has 1/3 of their business with the US govt, much of which is custom designed solutions with long term contracts. So Viasat has options for their capacity that EchoStar does not have. EchoStar is mostly a pure play on the satellite internet market and EchoStar is in the worst position competitively. EchoStar also has the most North American customers for Starlink to poach, about 2x more than Viasat.

There's an announcement that HughesNet is partnering with OneWeb now, so it  looks like they are trying to buy their way back into competitiveness. Will the conglomerate of OneWeb/Hugesnet(Echostar) put the screws on Viasat?

That partnership has been out since 2015:
https://www.hughes.com/resources/press-releases/hughes-signs-190m-contract-oneweb-production-ground-network-system-global

It's hard to say what will happen there; both HughesNet and ViaSat have a filthy reputation in the consumer market, so they both may be gutted out of well-earned spite unless they start offering plans at the price of cable internet.

[RocketGoBoom]

Looking at costs, and for the moment  ignoring SL’s obvious cost advantage in launch, a constellation of 41k sats at $150k (guesstimate) each will cost $6B. Three GEO megasats will be much less for only three launches and 2-3x the lifetime. With future launch costs showing every promise of dropping dramatically the equation morphs. Sats no longer need to be so robust... yada yada. And only three sats to manage.

That is not really a fair comparison.

Three Viasat 3 satellites are going to provide 3 Tbps of nameplate capacity for a build plus launch cost of $2.1 billion ($700 million each).

Starlink's 42,000 satellite plan will provide 840 Tbps of nameplate capacity for a build plus launch cost dramatically lower per Tbps.

Then it gets complex with utilization percentages while over the oceans and during night hours, etc. But on a cost to build plus launch capacity into orbit, Starlink is launching capacity into orbit at about 1/10 the cost of Viasat. And Viasat looks like the best cost numbers of the GEO satellite operators.

[RocketGoBoom]

We also discussed EchoStar / HughesNet and their Jupiter 3 satellite. The cost metrics for EchoStar are about 40% more expensive than Viasat's. I agreed with all of his numbers. Also, they are only doing one satellite with 500 Gbps compared to Viasat doing three satellites with 1,000 Gbps each. EchoStar is basically screwed and won't be financially competitive with Viasat or Starlink.

The CEO also made the point that Viasat has 1/3 of their business with the US govt, much of which is custom designed solutions with long term contracts. So Viasat has options for their capacity that EchoStar does not have. EchoStar is mostly a pure play on the satellite internet market and EchoStar is in the worst position competitively. EchoStar also has the most North American customers for Starlink to poach, about 2x more than Viasat.

There's an announcement that HughesNet is partnering with OneWeb now, so it  looks like they are trying to buy their way back into competitiveness. Will the conglomerate of OneWeb/Hugesnet(Echostar) put the screws on Viasat?

HughesNet is reselling Oneweb servicce. That means HughesNet is now the middleman between Oneweb and the end customer. So both Oneweb and Hughesnet are becoming lower margin companies.

And OneWeb has significantly higher costs for capacity than Starlink does.

Frankly, that looks like a disaster to me. Neither company is going to make much money. Starlink can offer plans below OneWeb's cost. And Starlink will have WAY more capacity available than OneWeb.

EchoStar HughesNet is sort of screwed no matter what. In GEO, HughesNet is only building one satellite (Jupiter 3) with 500 Gbps of capacity, while their main GEO competitor Viasat is building 3 satellites with a total of 3,000 Gbps of capacity. Now HughesNet is linking up with a reseller deal to offer high cost OneWeb services also? HughesNet is desperate and doesn't really have a chance.

[envy887]

Looking at costs, and for the moment  ignoring SL’s obvious cost advantage in launch, a constellation of 41k sats at $150k (guesstimate) each will cost $6B. Three GEO megasats will be much less for only three launches and 2-3x the lifetime. With future launch costs showing every promise of dropping dramatically the equation morphs. Sats no longer need to be so robust... yada yada. And only three sats to manage.

That is not really a fair comparison.

Three Viasat 3 satellites are going to provide 3 Tbps of nameplate capacity for a build plus launch cost of $2.1 billion ($700 million each).

Starlink's 42,000 satellite plan will provide 840 Tbps of nameplate capacity for a build plus launch cost dramatically lower per Tbps.

Then it gets complex with utilization percentages while over the oceans and during night hours, etc. But on a cost to build plus launch capacity into orbit, Starlink is launching capacity into orbit at about 1/10 the cost of Viasat. And Viasat looks like the best cost numbers of the GEO satellite operators.

This. SpaceX is getting 20 Gbps per satellite and 60 satellites per launch, or 1.2 Tbps per launch. Viasat is getting 1 Tbps per launch. And SpaceX is getting launches for internal cost, probably less then $25M each. Viasat is paying 3 or 4 times that per launch.

So launch costs, once normalized for Viasat's 2-3x longer life, are still slightly in SpaceX's favor. And satellite costs are probably 20x or more in SpaceX's favor.

[thirtyone]

Airbus is manufacturing Oneweb satellites and is a major investor in Oneweb, Ariane Group is a JV of Airbus and Safran, and yet he doesn't see to have a problem launching on Ariane rockets...

Definitely a good point. I think this is at least a bit of a psychological thing at this point (in an ideal world, CEOs are rational, but...) and Starlink's performance probably more existentially threatens the market than OneWeb. And because there is that extra degree of separation, and launch profits don't immediately go into R&D for your competitor (unless you invest more into Starlink, etc.)

Some new info since those posts:

From https://spacenews.com/satellite-operators-hint-at-fear-of-spacex-blue-origin-becoming-competitors:

“It’s a consideration,” [Eutelsat Deputy CEO Michel] Azibert said of the risk that launch providers could become competitors. Eutelsat, which operates around 40 geostationary satellites, has launched twice with SpaceX and reserved a launch slot on Blue Origin’s future New Glenn rocket, scheduled for first flight in 2021.

“If in the end we have completely vertically integrated companies competing with us, it might affect, a little bit, the relationship,” Azibert said.

“We’ve had a very close relationship with SpaceX,” [SES CEO Steve] Collar said. “We were their first commercial launch [and] their first flight-proven booster launch. But at some point you’ve got to think about what’s the right thing for our business.”

“At this point it’s not an issue, but could it become an issue in the future? Yeah, it could,” he added.

Definitely a few other customers starting to think about launching on competitors' rockets, including New Glenn. FWIW the Hughes representative basically said they'd be smart about it and if the launch is cheapest with SpaceX, they'll launch with them.

Of course Elon rather explicitly mentioned that he's not even thinking about selling off Starlink right now, with really what should be a very hopeful statement for GEO sat operators - every LEO constellation has gone bankrupt, and his only goal at the moment is to not go bankrupt. I think that's a difference in perspective between him and some of more fiscally-minded heads at SpaceX. He's a lot more focused on technological goals, so he'd rather not spend too much time thinking about company organization after the difficult technological barriers are overcome, while others also want to start planning in advance for how to squeeze more money out to fund the longer term goals at the company (i.e. colony on Mars).

[RocketGoBoom]

Of course Elon rather explicitly mentioned that he's not even thinking about selling off Starlink right now, with really what should be a very hopeful statement for GEO sat operators - every LEO constellation has gone bankrupt, and his only goal at the moment is to not go bankrupt.

The GEO sat operators would be crazy not to be concerned about SpaceX and OneWeb. New competition means that some portion of their existing customer base is likely to leave them for either SpaceX or OneWeb. Many of the legacy GEO satellite operators are barely surviving financially right now. Losing 20% of your customer base, and still having all of the same high fixed costs, would be deadly to most businesses.

Go take a look at the stock prices of Intelsat, Eutalsat, SES, Viasat, EchoStar. None of them are looking good. All of them are either losing money or at best barely profitable. And the competition has not yet even started stealing their customers.

What are their profitability numbers going to look like if SpaceX takes 10% of EchoStar HughesNet customers? 20% of customers? 50%?

Just as an example, EchoStar lost $50 million in the past year and projected to lose about $50 million next year on $2 billion in revenue ... before competition starts from SpaceX. What do you think those numbers are likely to be AFTER competition from Starlink begins? What do you think is likely to happen to EchoStar's stock price when Wall Street realizes it?

This is going to get ugly. Not every legacy satellite operator is going to be around in 3 years. Some of them are likely going to go through bankruptcy reorganization.

Their survival depends on Starlink failing completely and not entering their satellite broadband market at all. If Starlink functions at all, whether it makes a lot of money or a little money, it is going to hurt the existing players tremendously. The existing players (Viasat, EchoStar, Telesat, Eutelsat, SES, etc) all need Starlink to fail.