SpaceX and OneWeb internet satellite comparison

[nadreck]

Total OneWeb cost is

Satellite Build: ~$500M (guess)
Launcher One: 39 x <$10M   = ~$300M
Soyuz           : 21 x <$100M = ~$1800M
Ground segment : ~ $100M (guess)

Total: ~ $2.7B

So current funding of $500M is only ~20% of the total.

I would note that Iridium managed to use a lot of debt financing.  I think floating ~$2B in bonds is possible.

Iridium went bankrupt.

And Motorola had guaranteed a portion of its debt ($750M) plus had put $400M in equity into the company. BUT Motorola built them a lot of satellites and in total had a $6.6B contract with Iridium, now how much of that was not paid for in the end I don't know. However it is arguable that Motorola still benefited from the total business even if they "lost" $1.15B in investment and debt guarantees.

Like a lot of fiber networks being built at the same time, the people who came along after and picked up the pieces (at < $0.10 on the dollar) managed to make a business of it. However they never had to pay the cost to build it out in the first place.

[AncientU]

This is so fun!
Another space race.

[nadreck]

This is so fun!
Another space race.

So true and with exciting deadlines to provide immediacy and drama (ex. Wyler's license expiring).

[billh]

One Web cannot serve the equator region.

OneWeb can serve the equator region, Intelsat sats may just provide another option for customers.

Then how does using Intelsat make any sense?

As I recall, they plan to use spectrum that Intelsat's geostationary (equatorial) satellites already use. They must design their system not to interfere.

[guckyfan]

As I recall, they plan to use spectrum that Intelsat's geostationary (equatorial) satellites already use. They must design their system not to interfere.

True, well I don't know it is Intelsat, it may be different providers in different locations. However these licenses require non interference with signals from geostationary satellites. They can in principle still serve all regions of the world. It requires strict control of where and in which direction they transmit with what power. So while equatorial they cannot serve directly down but in an angle while serving equatorial regions with satellites off equatorial positions. They just need to make sure they dont beam down parallel to geostationary but in an angle.

Intelsat providing service for OneWeb customers makes however sense only if they cannot serve equatorial regions for whatever reason IMO.

[baldusi]

IIRC, Galileo Soyuz was €56/flt. But that's a Soyuz-STB/Fregat-MT. The 2.1B is 38% more expensive than the 2.1A. I don't know if they need a Fregat or not. That alone should be around 6M or so. So if they are going to use a single Soyuz-2.1A they could easily get 35M/flt. If it is a Soyuz-2.1A/Fregat that should be 40M. May be Baikonour could be 5M to 8M cheaper.

[TrevorMonty]

IIRC, Galileo Soyuz was €56/flt. But that's a Soyuz-STB/Fregat-MT. The 2.1B is 38% more expensive than the 2.1A. I don't know if they need a Fregat or not. That alone should be around 6M or so. So if they are going to use a single Soyuz-2.1A they could easily get 35M/flt. If it is a Soyuz-2.1A/Fregat that should be 40M. May be Baikonour could be 5M to 8M cheaper.

Based on your figures they should be able to build and deploy each satellite for <$2M each, add in the ground stations and the whole constellation of 900 could cost around $2B. They only need a $500m- $1B a year of revenue to be viable.  10M customers paying $50 a month is $600m a year. Shouldn't be a problem finding 10M customers when they can access 7B.

[dkovacic]

In the article http://www.themoscowtimes.com/article.php?id=505825 there is a quote of 400 million USD contract value for 16 soyuz launchers

[dkovacic]

In the article http://www.themoscowtimes.com/article.php?id=505825 there is a quote of 400 million USD contract value for 16 soyuz launchers

...which gives a price of 25 million per Soyuz. This seems quite low as Galileo launches were quoted at 56 million EUR.

[MikeAtkinson]

Lots more OneWeb info at https://twitter.com/pbdes

Also http://spacenews.com/launch-options-were-key-to-arianespaces-oneweb-win/
and http://spacenews.com/news-analysis-onewebs-big-announcement-should-quiet-doubters/

I haven't time to pull out all the facts from these.

[randomly]

The first Iridium sats were ~1500kg, and Globalstar is ~500kg. "beer fridge" sized sounds more like the 50-100kg size that Skybox has been doing.

~Jon

The first Iridium satellites were 1500 LBS - 700kg
Iridium Next satellites are ~800kg

This is roughly 1/10 the size of the big GEO sats. I wouldn't be surprised if Spacex satellites are on the order of 500kg.

[randomly]

After reading through this thread there are a few comments and clarifications I want to make.

There are three variables you can use to increase data bandwidth.

1) Use more spectrum
2) Use more complex modulation schemes. The complexity is limited by your signal to noise ratio. The better your signal to noise ratio, the more bits you can pack into each symbol. To increase signal to noise you can use more transmit power and/or a larger antenna. (and/or reduce the distance between xmit and rcvr. which is what operating in LEO does for you compared to GEO)
3) Spatial diversity - this means using multiple beams (in the same spectrum) each with it's own data stream. Multipath is commonly used to increase data rates in WIFI. It requires multiple antennas.

You can't use multipath from a single satellite because you cannot separate the antennas far enough to discriminate the signals, and above about 10Ghz propagation is direct line of sight only, there is no useable multipath signals.

A key aspect of the Spacex system that people seem to be missing is that the ground based antenna is a phased array. This is more complicated and costly, but it has one huge advantage. It can simultaneously maintain many separate focused beams.  With a constellation of 4000+ satellites you will have on the order of 50 satellites within 45% of the zenith in view of the ground antenna at all times. This means 50 data channels at once. Unlike a simple antenna the phased array can spatially discriminate between the various beams and so they can all use the SAME SPECTRUM. This is a 50-1 bandwidth multiplier over not using a phased array (you could also do it with 50 separate antennas, each tracking one satellite, but not cost effectively)


As to inter-satellite links, almost certainly they will go with optical. Nearest neighbor satellite connections will be less than 1000mi, the beams don't have to be that collimated to achieve adequate S/N, and you have essentially unlimited spectrum to work with at those frequencies. Also there is still some difficulty in making reliable millimeter wave systems, with long term amplifier degradation at the higher frequencies.

Iridium has tracking antennas for inter-satellite links. The antennas pointing at the leading and trailing satellites in the orbit are aimed and then don't move again since their orientation doesn't change. Only the antennas to the side need to actively track. The tracking antennas on Iridium have not been a problem area over the last 15 years.

Spacex constellation is not suitable for Cell use as the frequency is too high and has no penetration. Coverage could be implemented with a local cell site linked to a Spacex antenna for the backbone connection.

This is not a major threat to Iridium, the majority of their business is actually asset tracking, not phone or data. Little boxes that report home location and condition of containers etc. Although Iridium only has about 8 mhz of spectrum, which has strangled it's expansion and use as a data system, it's down around 1.6ghz which has some penetration. Spacex will certainly be much higher, 12-40ghz range.

The Spacex system is complex. Every receiver/antenna has to know it's exact location/orientation and time. It has to know the location of all visible satellites and their orbits to calculate pointing and adapt to the ever changing doppler shift of each satellite. The software to run the system is going to be very complex. The phased array system will be complex...

One can understand why OneWeb didn't want to go this route if they needed something working by 2019 or lose their spectrum rights.

[QuantumG]

The Gods Must Be Crazy http://spacenews.com/heres-why-coca-cola-is-investing-in-oneweb/

[R7]

The Gods Must Be Crazy http://spacenews.com/heres-why-coca-cola-is-investing-in-oneweb/

Coke wants to empower third world women ... to sell more coke.

And force their husbands to buy it so they can call home using the bottle caps as a currency.

Bottle caps as a currency...

It is time to start building the Vaults! 

[neoforce]

After reading through this thread there are a few comments and clarifications I want to make.

<removed details>

Spacex constellation is not suitable for Cell use as the frequency is too high and has no penetration. Coverage could be implemented with a local cell site linked to a Spacex antenna for the backbone connection.

<removed details>

One can understand why OneWeb didn't want to go this route if they needed something working by 2019 or lose their spectrum rights.

randomly: Thanks for the great explanation. However I'm still confused by spectrum. I understand OneWeb has rights to specific spectrum that must be in use by 2019. But what rights to spectrum does SpaceX have?  I remember Branson saying SpaceX didn't have rights to any spectrum. (Not that I trust what he says.)

So what do we know about SpaceX spectrum, especially given that you said their frequency is too high for penetration.

[Patchouli]

The Gods Must Be Crazy http://spacenews.com/heres-why-coca-cola-is-investing-in-oneweb/

Coke wants to empower third world women ... to sell more coke.

And force their husbands to buy it so they can call home using the bottle caps as a currency.

Bottle caps as a currency...

It is time to start building the Vaults! 

Funny theory but I can see Coca Cola getting something more practical out of this deal.
They can get a reliable and secure data infrastructure esp if they set aside a few sats for their own corporate use and it'll help them to expand in emerging markets without having to deal with local monopolies on data and telecom.

[meekGee]

After reading through this thread there are a few comments and clarifications I want to make.

<removed details>

Spacex constellation is not suitable for Cell use as the frequency is too high and has no penetration. Coverage could be implemented with a local cell site linked to a Spacex antenna for the backbone connection.

<removed details>

One can understand why OneWeb didn't want to go this route if they needed something working by 2019 or lose their spectrum rights.

randomly: Thanks for the great explanation. However I'm still confused by spectrum. I understand OneWeb has rights to specific spectrum that must be in use by 2019. But what rights to spectrum does SpaceX have?  I remember Branson saying SpaceX didn't have rights to any spectrum. (Not that I trust what he says.)

So what do we know about SpaceX spectrum, especially given that you said their frequency is too high for penetration.

IIUC, direct-to-handset spectrum is a rare commodity, especially bands that work world-wide, and Wyler's got a good chunk of it.

SpaceX's plans are not to go to the handset - their main thing is the long-haul, with stationary and car users. It's a completely different thing.

OneWeb is pretty much the same as was attempted before, only bigger.   commX is something completely different.

Branson either did not understand this, or was purposely ignoring it.

[oldAtlas_Eguy]

Although there may be some crossover between the two markets of broadband stationary and data mobile the two systems will have few other competitors for their customers. A secondary provider using commX could be a competitor for OneWeb by providing cellular service using commX as the network's backbone. Using solar panels and commX backbone remote towers could be placed anywhere and be nearly self sufficient. Again someone would have to put them up and OneWeb would already be operational. So for areas with enough customers and congestion using OneWeb this would be a business evolving direction (small town or groups of small towns close together). But for areas with very few customers it would not be a viable competitor to OneWeb.

[randomly]

randomly: Thanks for the great explanation. However I'm still confused by spectrum. I understand OneWeb has rights to specific spectrum that must be in use by 2019. But what rights to spectrum does SpaceX have?  I remember Branson saying SpaceX didn't have rights to any spectrum. (Not that I trust what he says.)

So what do we know about SpaceX spectrum, especially given that you said their frequency is too high for penetration.

There are no large chunks of spectrum available below 10Ghz, and above about 40ghz atmospheric extinction starts seriously attenuating the signal. By 100Ghz the atmosphere is essentially opaque.

Spacex needs at least several ghz of spectrum to do what they want to do. I don't know what is available but a possible source is to partner with somebody who already has such spectrum. Viasat-1 uses 4Ghz of spectrum in the 18-20Ghz and 28-30Ghz range. The Spacex system would completely replace what ViaSat is doing, it might be a smart move for them or anyone else in their position.

Acquiring available free spectrum or partnering with somebody who already has it, I think Spacex can find the spectrum they need.

[Robotbeat]

How much is Ariane's Soyuz cost? $60-90 million per launch? If so, that's $1.2-1.9 billion. They could have saved a pretty penny by launching on Falcon 9, considering it's nearly double payload and/or ability for reuse. Of course, that would also mean helping out their competitor. But I don't expect SpaceX to suffer much because of it.

Based on http://spacenews.com/40177arianespace-eyes-new-soyuz-opportunities-from-baikonur/, it looks like European Soyuz is 60 million euro, but Baikonur Soyuz could be a lot cheaper.

Right, and 60m Euro converts to about 60m-90m USD, depending on when in the last few years you pick the exchange rate. At the time, it was about 80m USD or so (that was from a 7 Soyuz group contract). Still could save money on a Falcon 9, not least because of the higher performance of a Falcon 9 (and the big price discount on F9R).