Hacker News Reader

Comment by DGAP | Hacker News Reader

Can someone explain to me the economics/ technology of Starlink and why it beats out traditional geosynchronous internet providers? Is it that owning your own cheap launch platform makes launching lots of satellites instead of one big one feasible? Is it some new antenna miniaturization technology? If that were the case, I would think this would have been invented by one of the myriad existing players in the space. Why is this apparently revolutionary progress in internet satellite technologies coming out of nowhere from a company (SpaceX) that has nothing to do with connectivity or satellites?

View on HN

It is due in part to them owning a cheap launch platform. The world has not caught up yet to how cheap space launches are going to be. SpaceX knew that with their Falcon 9 reusable rockets that launch costs for them were going to go down by a factor of 10 and with their new SuperHeavy, by a factor of 1000. Think $30/kg to orbit instead of $10000/kg. 14 of the 23 rockets launched by SpaceX this year were for Starlink. They want to build up a culture and skills for rapid relaunch and building StarLink gives them a hopefully profitable way to drive that.

Phased array antennas are necessary for this to work (no tracking of the low satellites necessary), because one wants to have the satellites in very low orbits to achieve low latency. Geosynchronous satellites are very far away from earth so that speed of light gives unacceptable delays for things like gaming, chats, remote server admin, etc. Very low orbit can get short pings (in theory better than fiber because light moves faster in a vacuum than in glass). But very low orbit means one needs a lot of satellites to have wide coverage and no dropped connections. They are thinking around 12,000 in the first phase and around 42,000 in the second phase. Having so many satellites also gives the system much higher bandwidth as each satellite is limited in data transfer rates.

Plus SpaceX is looking for a cash cow to fund their ultimate goal: a cheap, extensive Mars transportation system to facilitate building a city on Mars.

I agree with the cheap launch platform argument but from another, perspective.

I don't believe SpaceX numbers and BFR (it will always be BFR for me) is still a paper rocket (Raptor looks good though). But it is still cheap compared to traditional launches.

The problem is: SpaceX has now plenty of cheap, reusable Falcon 9s but not a lot of stuff to launch. There are not that many companies who want to build and launch satellites: it is expensive no matter what and the cost has to be justified. Falcon 9 rockets (including heavy) are also limited in capability: small fairing space and not the most efficient beyond LEO. Also, it may not be suitable for some sensitive payloads (military, etc...).

Ii means that SpaceX has good rockets, they already have the "culture and skills for rapid relaunch" but not enough market for it. So they did the obvious thing: create their own.

As for BFR. I made the choice not to take SpaceX communication into account, it is worthless. Same thing with everything by Elon Musk. I only believe what they show, which is already impressive enough.

I think there is still a pricing problem with reusability because spacex doesn’t have any competition that is reusable.

There is no downward pressure on pricing because spacex only has to be slightly cheaper than the cheapest non-reusable rocket. So for a marginal cost of, let’s say, $2m for a reflown booster, there are still able to charge $50m+.

But since their customers are seeing prices of $50m, space is still expensive... nothing much has changed.

(I’ve simplified this by omitting the fact that the upper stage is not reusable)

A lot of BFR development is being done in the open in Boca Chica, and you can follow along because it’s being live-streamed almost 24/7. BFR has had small incremental hops recently and will try for a 15km flight this week. We’ll see what happens.

Hopefully Bezos and Blue Origin can get New Glen on the launch pad soon. It is the only viable competitor to SpaceX that I see at the moment. One other possibility is for some company in China to appear on the scene.

Elon Musk has said he would love some competition because it helps increase the pace of progress. He even tried that internally for a bit by developing StarShip in two locations for awhile. I heard him in an interview saying that SpaceX might have too many great engineers and that it would be better for his goal of a city on Mars to have them spread over many companies.

By no definition is Starship a paper rocket. Either you have not been paying attention or you don't know what the definition of a paper rocket is.

> small fairing space and

Its not actually small. Its a little smaller then the biggest that you could launch on Delta 4 (that you can't buy) or a New Glenn (that you can't buy), but to just say its 'small' is just not accurate. Essentially its size is big enough for 99% of payloads.

There are like a few military sats in the next 5 years that need a slightly bigger fairing and they will have that for those launches.

> not the most efficient beyond LEO

By what definition? Just because it doesn't use hydrogen doesn't mean its not the suitable for beyond LEO. There is a reason the alternative for SLS is Falcon Heavy and not anything else.

> Also, it may not be suitable for some sensitive payloads (military, etc...).

That is just false. Falcon rockets are fully qualified for all military orbits and payloads.

The only thing they can't do right now is vertical assembly and that will be built by the time they have the first contract.

It's all about latency. Having broadband with 200+ms of latency to reach geostationary satellites is to slow for modern networks.

https://www.satsig.net/latency.htm

You have to move the satellites closer to Earth to reduce latency - hence low earth orbit (LEO). You can achieve much better latencies, <20ms, and support modern networking applications.

https://arstechnica.com/information-technology/2020/03/musk-...

But when you reduce the height to LEO, not only are the satellites no longer geostationary, they also have a much smaller coverage area. So you need more satellites to cover the same area, and fancy networking to handle the constant hand-off between satellites.

Bandwidth as well. Starlink can reach 100Mbps, and they're talking about 200Mbps. They can do that in part because they have more satellites.

> because they have more satellites

I wouldn't characterize this as a function of having more satellites, but rather having fewer users per satellite (an admittedly subtle difference). But, related to the above post(s), the smaller geographical area each satellite covers is a problem for coverage (you need more satellites in orbit), but is a benefit for speed -- you have less users vying for the full bandwidth of the satellite.

And, because the geographical areas are smaller, instead of having one base station/uplink, you now have geographically distributed base stations, so you have multiple points to access the Internet. This also helps with bandwidth (instead of X total uplink capacity, you need X/n capacity for each site, where n is the number of sites).

Once you get around the logistics of managing a fleet of satellites, many, many benefits flow from having lower orbits.

To grab a couple bits of info from that page + google, to give people a good idea of the distances involved at a glance:

    Starlink orbit:           550km
    Geosynchronous orbit:  35,786km
    Earth's circumference: 40,075km

Geosynchronous orbits are quite a way out. Latency at terrestrial distances can be painful already, bouncing through geosynchronous satellites will be multiples worse at a minimum. Meanwhile, Starlink's are at "normal" distances for networking - you probably connect to things further away all the time already.

The low latency is also a requirement to receive FCC subsidies related to rural broadband, e.g. see https://www.reddit.com/r/spacex/comments/h04on0/fcc_just_vot....

Have heard (IIRC in https://theamphour.com/518-satellites-and-evs-with-joris-aer...) that this could have been part of the reason for Starlink's decision to have their satellites in a lower orbit.

The subsidy angle is also of course a familiar playbook across SolarCity/Tesla/SpaceX etc.

SpaceX is super heavily vertically integrated which probably helps a lot with the financials. Also note that their financials aren't public, so there's all kinds of funny possibilities (what if SpaceX launches are sold below cost as a loss leader?)

I'd also imagine that they need a TON of subscribers to make it profitable. However by commoditizing their satellites the marginal cost of improving coverage is low (compared to GEO where every satellite is a billion dollar bespoke system).

Phased array antennas have been around for a while but I wouldn't be surprised if the starlink terminal is the first consumer device to actually use them (other than maybe boat radars?). They're still pretty expensive.

When you say boat radar and phased array, what is your experience there? The only radar I've seen on a boat has one of those spinning antennas.

No moving parts required, but you do need multiple units pointed in different directions to get full coverage:

https://en.wikipedia.org/wiki/Active_phased_array_radar

My experience is poor memory :) After looking into it a bit, it seems that only the military is using phased array radars on boats. I think I misremembered an article about Garmin and Raytheon introducing solid state radars a few years ago

I've heard it suggested that Starlink is used by SpaceX to keep their schedule fully booked, which helps them maintain quality standards, training, etc. If so, they could perhaps lose money on Starlink but still come out ahead when you factor in the benefit of having experienced workers for whom building and launching rockets is routine.

That's a great point. Similarly it's also a great opportunity for them to push the limits of reusability. A rocket failure with a paying customer's payload is Very Bad. But the risk tradeoffs look a little different if it's your own payload and you already have an assembly line churning out replacements

> they need a TON of subscribers...

This is where I get a bit suspicious of the whole thing, as the areas where this service makes the most sense for a consumer is sparsely populated rural areas. I don't see how the economics works out.

You don't see how the economics works out because you are not factoring the client with the deepest pocket: the military.

Putting tin hat

The low orbit large non-stationary constellation of Starlink makes harder for other countries to deny service by attacking the satellites.

Also, the sensor fusion from modern jets need a point to point (to avoid possible compromised infrastructure), high bandwidth low latency communication. Starlink seems a great fit.

This is only based on what we know the system is capable of. Going on the realm of speculation, it is not difficult to imagine how a system, at low altitude, that cover the whole planet, can be used to eavesdrop communication from other countries.

Also, the satellites are made to have a right turnover, they can be always updated with the latest technology.

Lets not forget that the Arecibo Observatory (that collapsed this week) was made to listen for russian radio waves reflected from the moon. Maybe Starlink is the Arecibo 2.0

In summary, I think the Starlink is a military project subsided with paying customer. I wasn't made to make profit, it was made to make SpaceX a viable company in the long run.

We know about how much investment was made in SpaceX. And the idea that SpaceX has for 10+ years sold launches as loss leader simply doesn't add up.

The opposite is more likely, they were profitable already for years and then the made re-usability happen and their margin should be huge. They have not really lowered the price since they introduced re-usability.

> If that were the case, I would think this would have been invented by one of the myriad existing players in the space.

A hundred-dollar bill is lying on the ground. An economist and an engineer walk past it. The economist nods at it and opines: "I thought I saw something over there, but I must've imagined it. If there had been $100 on the ground, someone would've taken it by now."

The engineer picks it up.

I feel like this joke would be improved by the engineer instead responding, "Yea, I did."

In the end, it's all of that and more. Geosynchronous is just too far away for good latency, plus the signal is weak and the bandwidth is saturated quickly.

So low earth orbit it is, but that meant doing several technology breakthroughs never seen before and at massive consumer scale.

First, you need to have massive launch capacity, otherwise satellites wouldn't be in view for most of the world, most of the time.

At this ground relative speed, they also stay in view pretty short at that speed, so you need to continuously roam from one satellite to the next every few minutes, hence the need for the cutting edge antenna array.

Lastly you need a truckload of satellites. Even the largest fleets are small series productions with specialized, expensive equipment such as radiation hardened processors, flywheels and propulsion to keep in orbit. Mass producing these from commercially available hardware was the third large challenge here.

Most companies would go bankrupt just tackling one of these and Starlink pulled them off in parallel. Pretty wicked.

What I don’t see talked about enough (in relation to SpaceX) is the one big customer who has been wanting global high bandwidth/low latency since at least the 90’s.

That is the US military. Think drones and all the fancy battle space information management stuff they’ve been talking about forever.

I don’t see how this gets built without the US mil wanting to use it. Furthermore I don’t see how this was designed without their input. It’s entirely possible that the US govt is covertly funding Starlink.

I believe even now hundreds of millis latency is quite enough for military applications, not even speak of 90's.

There's also a caveat: currently a satellite should have a ground station in view to relay data. It is rarely the case in areas of military activity.

It's because of the altitude of the satellites. Normal geostationary satellites are required to be at 35,786 kilometers in altitude above Earth's equator. Any signal has to travel that distance 4 times for a round trip ping, which equates to a minimum ping time to any service of 477 ms simply defined by laws of physics. 477 ms means any online games are unplayable and loading an ssl connection is going to push you into over a second delay.

Now the obvious idea is to put the satellites much closer to Earth, in that case they're rapidly moving across the surface and you need thousands/hundreds of them for full coverage, depending on the altitude the satellites are at. This wasn't really feasible until you had a rocket and satellite manufacturing designed for dumping large quantities of satellites into orbit at a time.

Building on the other comments here, SpaceX has more knowledge and certainty about recent and upcoming launch economics than other satellite providers, which opened up the possibility of a system like Starlink. Starlink is capable of such higher bandwidth and lower latency due to its lower orbit. Each satellite has a fixed amount of bandwidth capacity, and the area it covers is based on how high it is (how much of the Earth it can see at one time). High orbits mean they cover more area and spread that bandwidth out between more customers, low orbits mean less coverage area per satellite (so you need lots of satellites for global coverage) but more bandwidth per customer in that area.

Reply on news.ycombinator.com