3 ms·

$44M/mo revenue from his estimate is $528m/year; your estimates support, say, $150m depreciation per year. And there's significant operating costs. IMO the bi

by mlyle 4y ago

$44M/mo revenue from his estimate is $528m/year; your estimates support, say, $150m depreciation per year. And there's significant operating costs. IMO the biggest remaining factor is how long it takes to breakeven on the terminal hardware. While it's sold to end-users, it's a significant loss-leader (perhaps $2k of subsidy at this point).

toomuchtodo 4y ago
Terminals cost SpaceX $1500, breakeven is under a year considering the upfront cost to the user. Profitability will be determined by how many turns a Falcon 9 can make during its service life and if Starship can replace it fast enough (further driving down cost per kg to orbit). Everything else is mostly a fixed cost (ops, perpetual depreciation to account for satellite replacement on a cadence, etc).
- mlyle 4y ago
  > Everything else is mostly a fixed cost I disagree. They have all the normal variable ISP cost structures, too. They need to scale uplinks and the size of the constellation linearly with the number of subscribers they allow in an area, etc. IMO, what it will really come down to is how big the market is. Terrestrial infrastructure, including fixed wireless, should win most places-- it's a question of where the mobility benefits and/or population density make it viable, and how distributed these places are over the globe.
  - toomuchtodo 4y ago
    Uplinks and the huts for their gear, while a cost, are not exorbitantly expensive (and likely solved for long term with laser links between satellites for load balancing downlink across the constellation versus the current suboptimal bent pipe operating model). SpaceX has both US military and commercial airliners testing their service; the TAM is enormous for high speed global internet connectivity. Terrestrial service is ancillary revenue in markets where there is lower utilization (the same way T-Mobile offers home internet with no cap at lowest prioritization on their network to monetize their over investment in 5G). (disclosure: StarLink customer and SpaceX investor)
    - mlyle 4y ago
      > Uplinks and the huts for their gear, while a cost, are not exorbitantly expensive The question isn't whether it's exorbitantly expensive, but whether it's more expensive infrastructure per data moved to subscriber compared to other technologies that could reach those subscribers. That is, all providers need transit, backhaul, head-ends, etc. SpaceX will have all of these costs, too. > SpaceX has both US military and commercial airliners testing their service Yup, those are exciting markets, no doubt. They really depend upon inter-satellite links, as you mention. And, of course, they're not free of competitive pressures in these markets. > Terrestrial service is ancillary revenue in markets where there is lower utilization I haven't heard it positioned this way.
  - walrus01 4y ago
    One example of how large the addressable market is, is to look at the combined total revenue from all of the commercial geostationary satellite owning companies out there right now. And then add a good portion of the revenue from o3b, and iridium.
    - mlyle 4y ago
      > One example of how large the addressable market is, is to look at the combined total revenue from all of the commercial geostationary satellite owning companies out there right now. Unlikely. Starlink won't be able to segment the market perfectly. They're going to collapse the pricing for a lot of those uses. Not to mention that continued development of terrestrial networks is, IMO, going to further eat geostationary comms' lunch. > And then add a good portion of the revenue from iridium. Most of this Starlink can't capture, and it's not huge (~$500M/year).
      - walrus01 4y ago
        > Not to mention that continued development of terrestrial networks is, IMO, going to further eat geostationary comms' lunch. I absolutely agree on this part, having been directly involved in geostationary in the past one of the fascinating things to watch was a good portion of Africa disappearing from the target market for two-way geostationary data services, because of all the submarine cables that were installed along the west and east coast of africa, hitting major cities with landing stations, from about 2009 to the present date. Even Mogadishu, Somalia got a cable. There's a big old ugly defunct C band satellite Earth station on the hill above freetown, Sierra Leone that is a perfect example of this..
tsimionescu 4y ago
The $200M/year figure for the launch is also wrong - that's the estimated cost from one month of launches, not one year. It's not fair to call it a monthly cost either, but it's definitely not a once per year event. According to Wikipedia [0], they have had 53 successful non-test launches so far, starting in November of 2019. So 53 launches divided by 31 months = ~1.7 launches/month. At Soyuz costs, that's 1.7*$40M = $68M/month to launch the satellite fleet that they have today, which they would have to repeat every 5 years. Note that this is only an estimate of the launch cost, not the satellites, ground stations, or the cost of the individual devices. It also doesn't include bandwidth and other peering costs, personnel, sales and marketing, insurance etc. [0] https://en.m.wikipedia.org/wiki/List_of_Starlink_launches https://en.m.wikipedia.org/wiki/List_of_Starlink_launches

by mlyle 4y ago

toomuchtodo 4y ago

Terminals cost SpaceX $1500, breakeven is under a year considering the upfront cost to the user. Profitability will be determined by how many turns a Falcon 9 can make during its service life and if Starship can replace it fast enough (further driving down cost per kg to orbit). Everything else is mostly a fixed cost (ops, perpetual depreciation to account for satellite replacement on a cadence, etc).

mlyle 4y ago
> Everything else is mostly a fixed cost I disagree. They have all the normal variable ISP cost structures, too. They need to scale uplinks and the size of the constellation linearly with the number of subscribers they allow in an area, etc. IMO, what it will really come down to is how big the market is. Terrestrial infrastructure, including fixed wireless, should win most places-- it's a question of where the mobility benefits and/or population density make it viable, and how distributed these places are over the globe.
- toomuchtodo 4y ago
  Uplinks and the huts for their gear, while a cost, are not exorbitantly expensive (and likely solved for long term with laser links between satellites for load balancing downlink across the constellation versus the current suboptimal bent pipe operating model). SpaceX has both US military and commercial airliners testing their service; the TAM is enormous for high speed global internet connectivity. Terrestrial service is ancillary revenue in markets where there is lower utilization (the same way T-Mobile offers home internet with no cap at lowest prioritization on their network to monetize their over investment in 5G). (disclosure: StarLink customer and SpaceX investor)
  - mlyle 4y ago
    > Uplinks and the huts for their gear, while a cost, are not exorbitantly expensive The question isn't whether it's exorbitantly expensive, but whether it's more expensive infrastructure per data moved to subscriber compared to other technologies that could reach those subscribers. That is, all providers need transit, backhaul, head-ends, etc. SpaceX will have all of these costs, too. > SpaceX has both US military and commercial airliners testing their service Yup, those are exciting markets, no doubt. They really depend upon inter-satellite links, as you mention. And, of course, they're not free of competitive pressures in these markets. > Terrestrial service is ancillary revenue in markets where there is lower utilization I haven't heard it positioned this way.
- walrus01 4y ago
  One example of how large the addressable market is, is to look at the combined total revenue from all of the commercial geostationary satellite owning companies out there right now. And then add a good portion of the revenue from o3b, and iridium.
  - mlyle 4y ago
    > One example of how large the addressable market is, is to look at the combined total revenue from all of the commercial geostationary satellite owning companies out there right now. Unlikely. Starlink won't be able to segment the market perfectly. They're going to collapse the pricing for a lot of those uses. Not to mention that continued development of terrestrial networks is, IMO, going to further eat geostationary comms' lunch. > And then add a good portion of the revenue from iridium. Most of this Starlink can't capture, and it's not huge (~$500M/year).
    - walrus01 4y ago
      > Not to mention that continued development of terrestrial networks is, IMO, going to further eat geostationary comms' lunch. I absolutely agree on this part, having been directly involved in geostationary in the past one of the fascinating things to watch was a good portion of Africa disappearing from the target market for two-way geostationary data services, because of all the submarine cables that were installed along the west and east coast of africa, hitting major cities with landing stations, from about 2009 to the present date. Even Mogadishu, Somalia got a cable. There's a big old ugly defunct C band satellite Earth station on the hill above freetown, Sierra Leone that is a perfect example of this..

tsimionescu 4y ago

The $200M/year figure for the launch is also wrong - that's the estimated cost from one month of launches, not one year. It's not fair to call it a monthly cost either, but it's definitely not a once per year event. According to Wikipedia [0], they have had 53 successful non-test launches so far, starting in November of 2019. So 53 launches divided by 31 months = ~1.7 launches/month. At Soyuz costs, that's 1.7*$40M = $68M/month to launch the satellite fleet that they have today, which they would have to repeat every 5 years. Note that this is only an estimate of the launch cost, not the satellites, ground stations, or the cost of the individual devices. It also doesn't include bandwidth and other peering costs, personnel, sales and marketing, insurance etc. [0] https://en.m.wikipedia.org/wiki/List_of_Starlink_launches https://en.m.wikipedia.org/wiki/List_of_Starlink_launches