But if everyone gets 1Gbps then each satellite can only service 23 people. Of course people who have that kind of bandwidth don't use it continuously. But what if we all try to watch some 4k video at say 20mbps? That's 50x23 = 1150 users per satellite or around 5 million users to saturate the whole constellation. It's still cool, but certain things will not be possible.

Key phrase in the article: "up to 1Gbps per user". They aren't providing an SLA of guaranteed 1 Gbps, but a maximum possible 1 Gbps. They will oversubscribe just like all WAN's do because thats what makes the most economical sense. Now, I bet you will be able to pay for your guaranteed IOPS but everyone who's used EBS knows that costs money.

Oversubscription isn't exactly a rare thing in the ISP world.

No, but cables can be duplicated, wireless bandwidth can't.

lasers are wireless and don't compete for spectrum. beam-forming with phased arrays is not quite as good, but still gives you more capacity than naively shared spectrum would.

This is laser based? Are the building ground stations everywhere too?

The links to end-users use beam-steered microwaves, sat-sat links use laser.

I was more making a more general statement: that wireless is not necessarily synonymous with shared medium.

>An astounding O(54% of global internet traffic) [2]!

Thanks for working this out. This matches with Musk's stated goals for the project from January 2015 ("the majority" of internet traffic): https://www.youtube.com/watch?v=AHeZHyOnsm4&t=2m40s

>The focus is going to be on creating a global communications system. This is quite an ambitious effort. We're really talking about something which is... in the long term it will be like rebuilding the internet in space. The goal will be to have the majority of long distance internet traffic go over this network, and about 10% of local and business traffic. Probably 90% of people's local access will still come from fiber, but we'll do about 10% of business and consumer direct, and more than half I think of the long distance traffic.

>As you guys may know, the speed of light in a vacuum is somewhere between 40-50% faster than in fiber. So you can actually do long distance communication faster if you route it through vacuum than you can if you route it through fiber.

>And you can also go through far fewer hops. Let's say if you want to communicate from Seattle to South Africa. If you look at the actual path that it takes it's extremely convoluted[1]. It will follow the outline of the continents. It will go through 200 routers and repeaters, and the latency is extremely bad. Whereas if you did it with a satellite network you could actually do it in two or three hops. Well maybe four hops; it depends on what the altitude of the satellites are, and what the cross-links are.

>Let's say at least an order of magnitude fewer repeaters or routers, and going through space at 50% faster speed of light. So from a physics standpoint it's inherently better to do the long distance internet traffic through space.

[1] https://personalpages.manchester.ac.uk/staff/m.dodge/cyberge...

If it does actually work faster than fibre then some early high-paying customers could be high-frequency traders (HFT). Those folks already go to amazing lengths and great expense to shave milliseconds off latency. They could get a huge early cash injection by auctioning off exclusive, prioritised access between certain locations.

For [3], Guesstimate ( https://www.getguesstimate.com ) pretty much does that.

More numbers to crunch:

- Satellite with LEO orbit covers 1-2 percent of earth surface with 10 degree elevation.

- Land occupies less than 30% of earth's surface area. Urban areas occupy less than 3 percent of the Earth's land surface. That's where the customers are.

You have to divide at least by 100.

Urban areas already have high-bandwidth, low-latency internet service at more competitive rates, let alone 3-5 years from now. I'd think most target customers for this service are in rural areas, as well as those with a need for connectivity while traveling.

I'm not 100% sure I'm following your comment, but with LEO, you don't get to choose a location above earth for each satellite, as you would with GSO.

Half of those satellites will be over the Pacific Ocean at any given time.

On the bright side, this completely solves the bandwidth for adventurers. The farther you are from civilization, the less contention for the links.

it's about 100Tbs which is close to current US to Europe internet capacity. So it's definitely will not represent 1/2 of the world's traffic.

Per the wiki [1], Internet traffic is defined as: the amount of data sent and received by visitors of a particular web site.

The traffic between europe and the US is not solely dominated by the amount of data sent and received by visitors of a particular web site. It includes a lot of other background data such as backups between regions.

The analogy to economics would be the calculation of GDP in that you generally don't factor in B2B products/services when the end product/service is the consumer.

Obviously, this won't replace my "laptop -> router" communication nor "server -> server" communication but it's the consumer that matters.

[1]: https://en.wikipedia.org/wiki/Internet_traffic#Global_Intern...

you conveniently ignore: Internet traffic is the flow of data across the Internet. The phrase "Internet traffic" is sometimes used to describe web traffic.