SpaceX's next big commercial customer

[su27k]

At SXSW Elon was saying the goal is $5M for 150 tons to NEO with BFR ready by the early 2020s.

If he is using US Tons that is an amazing $16 per pound.

If he's using metric tons, that's $15.15 a pound. 

As somebody has pointed out, that's cheaper than the estimates for a space elevator:

For a space elevator, the cost varies according to the design. Bradley C. Edwards received funding from NIAC from 2001 to 2003 to write a paper, describing a space elevator design. In it he stated that: "The first space elevator would reduce lift costs immediately to $100 per pound"

Chances are he took current launch cost, dropped it by an order of magnitude and rounded. BOM on ficticious graphene/nanotube wire isn't exactly precise right now.

He got a quote from Mitsui on nanotube price. Also the $100/lb price is for pound to GEO.

[OccasionalTraveller]

SES, and other operators, will at the very least need more satellites and launches when their current on-orbit satellites' station-keeping fuel runs out. Either new satellites to replace the old ones, or on-orbit servicing in the mould of SSL's Mission Extension Vehicle. I suspect that 2017 was a lull for ordering largely because 2002 was (using data from Gunter's Space Page - GEO-Sat Contracts). Orbit lifetime for GEO satellites has traditionally been 15 years.

One-for-one replacement assumes that SES (and the other players) aren't going to order additional satellites for their existing fleets. They're usually referred to as 'communication' satellites, but the vast majority of the payloads are direct-to-home broadcasting by satellite (DBS). SES has as part of their global fleet four satellites at 19.2°E serving Central Europe (Astra 1KR, 1L, 1M and 1N), and three at 28.2°E primarily serving the UK and Ireland (Astra 2E, 2F and 2G). 1KR was ordered in 2003 and flew in April 2006, with an anticipated 15 year lifetime, so unless it's performed much better than design lifetime, SES had better get on with ordering a replacement.

We're seeing a slow shift to 4K broadcasting as content becomes available. Channels tend to run the newest standard in parallel with older ones, so 4K requires new capacity. 4K requires four times as many pixels as HD, which (if using the same compression method and parameters) requires four times the bitrate. 4K content generally is using the newer HEVC/H.265 compression method rather than the MPEG-4 AVC/H.264 used by most HD content, but while this has an eventual goal of using half the bitrate of H.264 for the same content, that implies that 4K will still need twice as much capacity as HD. Typically, early compressors (particularly real-time compressors for live content) barely achieve much improvement over the previous generation; I can't find anything particularly reliable on how HEVC is doing so far.

On the other hand, the 'cord-cutting' trend might be generalised to cancelling of subscriptions of all types. In some countries DBS subscriptions have dropped away - although in the UK this means that the viewer can still view BBC, ITV and other free-to-air channels using the same equipment.

On-demand Internet access to TV content through services like BBC iPlayer or Hulu is growing, but it has a major problem - each connection from a viewer to a content server is an independent connection, with the data being sent once for each viewer. Even live streaming content from a TV content provider, YouTube or Twitch is sent individually to each viewer. Multicasting does not yet work at internet scale, as far as I'm aware, it only works within a single ISP (e.g. AT&T U-verse).

There may be a shift from the current dumb birds - which know nothing of the format of the data being broadcast, and simply transpose from the uplink frequency to the downlink and rebroadcast - to smarter satellites that employ digital signal processing. This press release indicates that SES-14 was to be the first such satellite for SES. Whether that changes the equations for payload size, lifetime, and technology obsolescence, I don't know. It may mean smaller payloads that can then stay on orbit for longer; it may mean that the hardware is obsolete within a few years of being launched. I think SES suggested that it would be cheaper to build and launch the satellites and they would be replaced more frequently, but I can't find the slide that I saw that on.

It's probable that current analogue satellites will be replaced by digital ones only when their station-keeping fuel runs out or something else breaks (e.g. a steerable dish gets stuck, a reaction wheel fails), unless more capacity is required before then.

[Zed_Noir]

....
Big customers in the next 5-10 years (beyond SpaceX), in my mind, could be non-traditional commercial companies / pursuits. These are all probably low probability occurrences (say sub<35%), but don't need all to work:

Planet or their ilk: They are pushing the constant world coverage concept and if they manage to develop the algorithms they hope, their demands for higher resolution and better coverage should ramp up. With lower costs, it seems likely they could ramp up to a lot more satellites and slightly bigger (albeit still small vs traditional telescopes).
....

Planet and their irk might not survive competition from SpaceX. The VLEO portion of the Starlink constellation are good platforms for observation. It will not be hard or expensive to add a mass produced optical sensory package to the later iterations of the Starlink VLEO birds. It seems to me that SpaceX will just absorb the market segment of LEO & VLEO observation constellations eventually.

[Asteroza]

....
Big customers in the next 5-10 years (beyond SpaceX), in my mind, could be non-traditional commercial companies / pursuits. These are all probably low probability occurrences (say sub<35%), but don't need all to work:

Planet or their ilk: They are pushing the constant world coverage concept and if they manage to develop the algorithms they hope, their demands for higher resolution and better coverage should ramp up. With lower costs, it seems likely they could ramp up to a lot more satellites and slightly bigger (albeit still small vs traditional telescopes).
....

Planet and their irk might not survive competition from SpaceX. The VLEO portion of the Starlink constellation are good platforms for observation. It will not be hard or expensive to add a mass produced optical sensory package to the later iterations of the Starlink VLEO birds. It seems to me that SpaceX will just absorb the market segment of LEO & VLEO observation constellations eventually.

Starlink may provide secondary payload slots on Starlink sats to third parties, like the new Iridium sats. A smart player would contract for a large quantity of VLEO secondary slots, then reuse the beam director for the intersat lasercomm as their optical sensor telescope frame basis (though would likely need something like a membrane lens to increase aperture). One could essentially become a third party earth observation organization with nation-state capabilities, like an outsourced NRO. I get the impression Musk didn't want to involve himself in that, as he wanted to be the satellite equivalent to a dumb pipe provider competitor, much in the way cellular carriers began to undercut each other approaching marginal cost of data transmission once a low cost player entered the field. For certain customers, having near 24/7 on-demand optical coverage would be game changing. As the number of sensors goes up, that coverage approaches 24/7 realtime continuous coverage as well.

[IainMcClatchie]

... reuse the beam director for the intersat lasercomm as their optical sensor telescope frame basis (though would likely need something like a membrane lens to increase aperture).

Are membrane lenses a thing yet?  Are you talking about a photon sieve or a Fresnel lens?

I know folks who worked on in-space laser communications.  They just used smallish silicon carbide mirrors.

For certain customers, having near 24/7 on-demand optical coverage would be game changing. As the number of sensors goes up, that coverage approaches 24/7 realtime continuous coverage as well.

How big is this market?  My understanding is that Planet Labs is having a hard time making this case to anyone with a lot of money.  The usual examples are for things like ports where you'd like to track containers.  The folks that really care about that have video cameras mounted on nearby buildings, which is a lot cheaper than satellites.

[gongora]

The GTO satellite market is in a lull. Are their expectations that one of the big players is going to announce a new set of GTO satellites (5+) due to launch in 3-4 years? Or are they all building and launching one at a time?

Is the chance of SpaceX launching some OneWeb sats greater than 0? For example, if LauncherOne isn't able to meet the schedule?

Is LauncherOne really going to be cheaper than SpaceX? Would OneWeb pay more than SpaceX rates just because they perceive StarLink as a competitor. I imagine SpaceX would have no problem launching OneWeb satellites.

The OneWeb constellation is being deployed on Soyuz rockets.

Not all of them.   http://spacenews.com/blue-origin-gets-oneweb-as-second-new-glenn-customer/

The first ~700 Oneweb sats will launch on Soyuz.

I don't think there has been enough Soyuzs booked to launch that many, hence the Blue Origin deal.

They booked 21 Soyuz and were looking at 32-36 sats per flight.

[Roy_H]

The OneWeb constellation is being deployed on Soyuz rockets.

Not all of them.   http://spacenews.com/blue-origin-gets-oneweb-as-second-new-glenn-customer/

The first ~700 Oneweb sats will launch on Soyuz.

I don't think there has been enough Soyuzs booked to launch that many, hence the Blue Origin deal.

They booked 21 Soyuz and were looking at 32-36 sats per flight.

So that works out to a quite reasonable $40M to $45M / flight. Soyuz mass to LEO ~7,100kg, Falcon 9 (with recovered booster) ~ 15,000kg or about twice Soyuz although I don't know if 70 satellites would fit in the F9 fairing. Just trying to figure out how much of a premium One Web is paying vs launching via SpaceX.

[gongora]

The OneWeb constellation is being deployed on Soyuz rockets.

Not all of them.   http://spacenews.com/blue-origin-gets-oneweb-as-second-new-glenn-customer/

The first ~700 Oneweb sats will launch on Soyuz.

I don't think there has been enough Soyuzs booked to launch that many, hence the Blue Origin deal.

They booked 21 Soyuz and were looking at 32-36 sats per flight.

So that works out to a quite reasonable $40M to $45M / flight. Soyuz mass to LEO ~7,100kg, Falcon 9 (with recovered booster) ~ 15,000kg or about twice Soyuz although I don't know if 70 satellites would fit in the F9 fairing. Just trying to figure out how much of a premium One Web is paying vs launching via SpaceX.

The contract value is somewhere in the $1B to $1.5B range.  It includes options for 5 more Soyuz and 3 Ariane 6.  The initial OneWeb deployment is 648? satellites, which is covered by having at least (10 on first flight) + (20 x 32) = 650 on the 21 Soyuz flights.

[AncientU]

Middle of that range ($1.25B) equates to $1.86M/sat.  NG is launching 400 on five flights... a price per launch of $150M would match the Soyuz per sat price.  Likely Blue significantly undercut this price with launches in the $100M-$120M range.  Falcon might be able to lift half the NG payload at roughly half the price, so comparable value to a constellation operator.  Doesn't seem anyone else can compete at this price point.

[Robotbeat]

The OneWeb constellation is being deployed on Soyuz rockets.

Not all of them.   http://spacenews.com/blue-origin-gets-oneweb-as-second-new-glenn-customer/

The first ~700 Oneweb sats will launch on Soyuz.

I don't think there has been enough Soyuzs booked to launch that many, hence the Blue Origin deal.

They booked 21 Soyuz and were looking at 32-36 sats per flight.

So that works out to a quite reasonable $40M to $45M / flight. Soyuz mass to LEO ~7,100kg, Falcon 9 (with recovered booster) ~ 15,000kg or about twice Soyuz although I don't know if 70 satellites would fit in the F9 fairing. Just trying to figure out how much of a premium One Web is paying vs launching via SpaceX.

Soyuz has a much smaller fairing, so double probably would fit just fine in Falcon 9. And for 10 flights or so, SpaceX probably could be bargained with.

[Robotbeat]

Middle of that range ($1.25B) equates to $1.86M/sat.  NG is launching 400 on five flights... a price per launch of $150M would match the Soyuz per sat price.  Likely Blue significantly undercut this price with launches in the $100M-$120M range.  Falcon might be able to lift half the NG payload at roughly half the price, so comparable value to a constellation operator.  Doesn't seem anyone else can compete at this price point.

Of course, if you can fit them in the FH fairing (which could get a little larger), for a fully recovered FH, that's 140 satellites for $90m, or just $640,000 per satellite.

Fully expendable FH, at $150, would be cheaper still, but you'd likely run into volume constraints. Still, for $150m, you could probably launch 310 of them, equating to $480,000 per launch.


...and heck, while we're at it... BFR could launch like 750 of them for less than a $10 million Falcon 1, so $13,000 per satellite. (Of course, I expect BFR will be priced similar to F9 and FH until development is paid off, unless you have some bargaining power.)

[AncientU]

Middle of that range ($1.25B) equates to $1.86M/sat.  NG is launching 400 on five flights... a price per launch of $150M would match the Soyuz per sat price.  Likely Blue significantly undercut this price with launches in the $100M-$120M range.  Falcon might be able to lift half the NG payload at roughly half the price, so comparable value to a constellation operator.  Doesn't seem anyone else can compete at this price point.

Of course, if you can fit them in the FH fairing (which could get a little larger), for a fully recovered FH, that's 140 satellites for $90m, or just $640,000 per satellite.

Fully expendable FH, at $150, would be cheaper still, but you'd likely run into volume constraints. Still, for $150m, you could probably launch 310 of them, equating to $480,000 per launch.


...and heck, while we're at it... BFR could launch like 750 of them for less than a $10 million Falcon 1, so $13,000 per satellite. (Of course, I expect BFR will be priced similar to F9 and FH until development is paid off, unless you have some bargaining power.)

It seems that these constellation sats are going to generally be volume limited -- indications from Blue that customers love the 7m fairing, and Falcon's notably volume-limited fairing.  The mass-limited end of the spectrum will be propellants, bulk supplies like food and water, maybe heavy equipment for construction...

[Robotbeat]

Only LEO payloads are likely to be volume limited on Falcon 9 and even most on Falcon heavy.

SpaceX can always make a larger fairing. But there has to be sufficient demand. I just don’t buy this being a significant constraint for spaceX, even though people keep trying to make it so. If it were, they’d simply make another bigger fairing.

[guckyfan]

Only LEO payloads are likely to be volume limited on Falcon 9 and even most on Falcon heavy.

SpaceX can always make a larger fairing. But there has to be sufficient demand. I just don’t buy this being a significant constraint for spaceX, even though people keep trying to make it so. If it were, they’d simply make another bigger fairing.

I guess a bigger fairing is just as impossible as getting a rover out of RedDragon. Though that is out now I just shake my head remembering the endless discussion about that.

[AncientU]

Only LEO payloads are likely to be volume limited on Falcon 9 and even most on Falcon heavy.

SpaceX can always make a larger fairing. But there has to be sufficient demand. I just don’t buy this being a significant constraint for spaceX, even though people keep trying to make it so. If it were, they’d simply make another bigger fairing.

External demand?  I don't think so.

I suspect they'll build the fairing next year, qualifying for Class C payloads and preparing for constellation initial deployment.

[Asteroza]

... reuse the beam director for the intersat lasercomm as their optical sensor telescope frame basis (though would likely need something like a membrane lens to increase aperture).

Are membrane lenses a thing yet?  Are you talking about a photon sieve or a Fresnel lens?

I know folks who worked on in-space laser communications.  They just used smallish silicon carbide mirrors.

For certain customers, having near 24/7 on-demand optical coverage would be game changing. As the number of sensors goes up, that coverage approaches 24/7 realtime continuous coverage as well.

How big is this market?  My understanding is that Planet Labs is having a hard time making this case to anyone with a lot of money.  The usual examples are for things like ports where you'd like to track containers.  The folks that really care about that have video cameras mounted on nearby buildings, which is a lot cheaper than satellites.

The Falconsat demo was a photon sieve lens, and DARPA MOIRE project originally was a photon sieve but switched to segmented membrane fresnel lens.

Starlink lasercomm has been described as using a 6 inch SiC primary mirror.

As for the on-demand surveillance market, Planet still is ridesharing, so their distribution isn't great. Planet is now trying to add propulsion capabilities, which should improve their sat distribution. Shortening the tasking cycle and lowering cost will change the nature of market. There's a group trying to get tasking down to 90 minutes right now, orderable through a smartphone app, to give a sense of the changes occurring in this space. When you get to full continuous coverage though, the economics change significantly enough that the very nature of the market changes.

[IainMcClatchie]

The Falconsat demo was a photon sieve lens, and DARPA MOIRE project originally was a photon sieve but switched to segmented membrane fresnel lens.

Thanks for that pointer.  I'll go read about those.

As for the on-demand surveillance market, Planet still is ridesharing, so their distribution isn't great. Planet is now trying to add propulsion capabilities, which should improve their sat distribution. Shortening the tasking cycle and lowering cost will change the nature of market. There's a group trying to get tasking down to 90 minutes right now, orderable through a smartphone app, to give a sense of the changes occurring in this space. When you get to full continuous coverage though, the economics change significantly enough that the very nature of the market changes.

But... who's going to buy this imagery?  And how much are they going to buy? Remember that it's 3-5 meter GSD.  What problem does that solve?  I have talked to farmers and other folks in Ag about this.  They can't figure out what Planet's value proposition is.

[Kansan52]

Well, FarmersEdge is a customer now that this farmer down the road from me use the imagery to say Hi to space with his cows:

http://www.kansas.com/news/business/agriculture/article205574004.html

What he uses it for is kinda silly. But the resolution is enough for farmers in that group to benefit by applying the images to their needs.

[TrevorMonty]

... reuse the beam director for the intersat lasercomm as their optical sensor telescope frame basis (though would likely need something like a membrane lens to increase aperture).

Are membrane lenses a thing yet?  Are you talking about a photon sieve or a Fresnel lens?

I know folks who worked on in-space laser communications.  They just used smallish silicon carbide mirrors.

For certain customers, having near 24/7 on-demand optical coverage would be game changing. As the number of sensors goes up, that coverage approaches 24/7 realtime continuous coverage as well.

How big is this market?  My understanding is that Planet Labs is having a hard time making this case to anyone with a lot of money.  The usual examples are for things like ports where you'd like to track containers.  The folks that really care about that have video cameras mounted on nearby buildings, which is a lot cheaper than satellites.

The Falconsat demo was a photon sieve lens, and DARPA MOIRE project originally was a photon sieve but switched to segmented membrane fresnel lens.

Starlink lasercomm has been described as using a 6 inch SiC primary mirror.

As for the on-demand surveillance market, Planet still is ridesharing, so their distribution isn't great. Planet is now trying to add propulsion capabilities, which should improve their sat distribution. Shortening the tasking cycle and lowering cost will change the nature of market. There's a group trying to get tasking down to 90 minutes right now, orderable through a smartphone app, to give a sense of the changes occurring in this space. When you get to full continuous coverage though, the economics change significantly enough that the very nature of the market changes.

Planet business is packaging and selling of data from earth observation images of varies spectrums. Satellite business is just an necessary evil to collect this data. They are not opposed to buying data from other constellation operators.

[IainMcClatchie]

Well, FarmersEdge is a customer now....What he uses it for is kinda silly.

That's my point.  I've yet to hear about a non-silly need for Planet Labs style imagery.  The kind of need where lots of folks in the same situation say, yeah, that's a good idea, and buy as well.