[JoshCole]

It is so much worse than that. The video doesn't use SpaceX cost. It uses their cost to others. It categorically rules out using their new rocket, even though their stated plans call for them to use it. It rules out ride sharing, even though they have been doing that. The video creator /knows/ they are doing that. They know they are lying. I can tell, because I've looked at some of the articles they posted screen captures of. They edited the fucking headlines to remove the rocket launch price. The video is absolute garbage. It even posits a government conspiracy, because it has to, because it lied about the speed test numbers that are used to decide eligibility for funding. Yet despite all that when they talk about their choice of numbers they pretend they are being generous to SpaceX by using conservative numbers when nothing could be further from the truth.

[ceeplusplus]

Eh, you can do your own math. 5k satellites / 5 years lifetime = expect to replace 1k per year. Each launch of ~50 satellites costs around $35M (conservative estimate). So overall capex required to maintain the system is $700M/year. SpaceX would need 530k subscribers at $110/mo just to keep the lights on - and each satellite would have to handle ~106 customers concurrently, assuming a uniform distribution (best-case scenario for SpaceX). According to some sleuths on Reddit [1], each Starlink satellite can deliver around 20Gbps bandwidth, which gives each of those customers around 188 megabit speeds, assuming absolute best case geographical distribution. In practice people will be clustered so it won't be that great.

Surprisingly not that bad. The real cost could be double and they would still have enough capacity to serve high speed internet and break even.

[1] https://www.reddit.com/r/Starlink/comments/l20j2c/total_thro...

[JoshCole]

Just to give a sense of scale, because it is crazy. Your estimate? It is 0.000621118012422 of his one year estimate. You arrived at a figure more than three orders of magnitude lower than his figure. So I hope you'll understand why I have so little respect for his estimate.

https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_pr...

To just restate that to try to get across how extremely different your estimate is from his. If you convert to percentage and round at two decimal places... your estimate is 0.06% of his estimate. To try and make this different a little more real for people lets try to convert it into more familiar figures. If were talking about a house rather than rocket launches and cp thought the price was $100,000 for a house then CSS would have an estimate more than three orders of magnitude different. Something more like $100,000,000. Translating this level of difference in evaluation means the disagreement in 'real object' terms is roughly equivalent to this much difference in cost estimation: $120,000,000 home https://www.zillow.com/homedetails/329-Albion-Ave-Woodside-C...? versus $125,000 home https://www.zillow.com/homedetails/31235-Manton-Rd-Manton-CA...

The difference in estimates is absurd. The reason it is that absurd is that CSS tries to argue that SpaceX couldn't work because it would cost more than the entire economy in the world. It's like a comedy thing where he tries to use rhetoric about how stupid SpaceX must be to strengthen his argument that they could succeed without an overarching conspiracy helping them lie about everything. This is absolutely a conspiracy video, not a factual video.

IIRC, he moves on from this topic shortly after that to talk about how since Shotwell is a woman she isn't worth respect. It hits harder, if you believe him, because he just made a 'strong as long as you are delusional' case that anyone who thinks SpaceX might ever be profitable is insane.

[HALtheWise]

One key factor you didn't include in this calculation is "overprovisioning", the industry term for the fact that you only need to spec your total network capacity based on peak load times, and at peak times it's not actually going to be the case that every customer is using their maximum contractually available bandwidth. The typical industry assumption is that this is something like a 10x factor, although as network connections get faster, this becomes more pronounced.