The physical shape of the internet isn’t that different from what we had 50 years ago. It’s gotten bigger. We have better wireless and switched from copper wires to fiber optic cable. But it’s still one big, physically connected system. An arterial network strung between telephone poles and lace through underground conduits, connecting every computer in the world.
But that might be about to change. A handful of companies are trying to build a new type of internet that doesn’t run on physical cables. And instead of going underground, they’re building it in space. (light music) This is what a satellite-based internet would look like. Hundreds or even thousands of satellites orbiting the Earth. It’s expensive to launch and maintain a network like this, but a few are already up and running.
SpaceX’s Starlink, Amazon’s Project Kuiper, OneWeb
SpaceX’s Starlink project is probably the most famous. Thanks, Elon. But OneWeb is also fleshing out its satellite constellation, and Amazon‘s Project Kuiper secured a massive contract to launch a plan of 3,236 satellites.
There have been satellite internet services, but those have been geosynchronous satellites. Maintaining a stable orbit over a specific spot on Earth. That orbit only works if the satellites are at a specific altitude, just over 22,000 miles up, which is so far away that the services tend to have awful latency. The new generation of satellite internet systems is doing it differently.
They’re putting their satellites in low to medium Earth orbit between 200-1200 miles up, which means you can finally get a low enough latency to make internet service work. The lower orbit also means the satellites must go much faster to keep from falling. So they’re always moving around relative to a particular spot on Earth. To make it work, you need a whole network of satellites. So when one drops out of the line of sight, there’s already another one coming into view.
One thing that gets lost when we talk about space internet is just how reliant these systems are on ground-based infrastructure. So to tap into a constellation like Starlink, you need to use a terminal that relays signals to a satellite that’s orbiting overhead. That satellite then needs to relay those signals to a ground station connected to the existing fiber-optic infrastructure. That ground station then relays those signals back to the satellite, which then relays them back to you and your user terminal.
But without those ground stations, the entire system wouldn’t work. And so you need them populated in certain locations all over the globe. And you need to be within several 100 miles of those ground stations in order for you to get the communications that you need. – Think back to that big network we showed you at the beginning. We take for granted that major services like YouTube and Netflix will operate out of regional data centers. And those data centers will be central parts of the internet backbone. Places like Northern Virginia and southeast London. But if a large portion of the internet users connects through orbiting satellites, that starts to look different.
For terrestrial networks, the hardest part is the last mile. That last bit of fiber connects the backbone to your house. It’s the most expensive part of the network to build and also the most delicate to maintain. So if you lose service, there’s a good chance your neighbors did too. And restoring it, it’s just a question of how quickly the local service crew can get out to you.
But that’s not necessarily true for satellite internet. And it could really change the economics of how remote areas connect. Now, we’re still a really long way from that happening. Right now, the best scenario for these companies is having a small sliver of total users connecting through satellite networks. But keeping that many satellites in the sky is still a huge engineering challenge and can cause real problems.
The astronomy community is very concerned about large mega-constellations like this because of what kind of impacts that can have on their research and their study of the cosmos. So in order to study distant places in the universe, astronomers take long exposure images of the night sky. But the problem with these satellites is they’re very reflective and can catch light from the sun. And when they pass over a telescope or an astronomer trying to take an image of the distant universe, that satellite reflecting sunlight can leave a very long white streak through the image, effectively ruining their observation.
Another major issue is the fact that low Earth orbit is getting even more crowded than ever before. And these companies propose putting thousands of satellites into roughly the same area of space. Satellite trackers are concerned that the probability of collisions will increase. And when things collide in space, it can be pretty nasty. These things are moving at roughly 17,500 miles per hour. So if one satellite runs into another, they can create thousands of pieces of debris that sticks around in space, threatening other satellites.
There are also practical issues that are still getting ironed out. When our editor-in-chief Nilay Patel tested Starlink in 2021, he said the service could offer inconsistent connectivity and huge latency swings at best. The biggest problem is obstructions and the dishes’ line of sight to satellites. Trees or even poles can cause service to be cut out. And since the satellites are always moving, you need a big strip of unobstructed sky. The usual fix is to mount the dish as high as possible.
But Starlink also warns that roof-mounted dishes can be blown loose in a storm. The company says you may wanna take the dish inside during high wind events. To be fair, Starlink has launched a lot more satellites since we reviewed it in 2021. But a quick spin through Starlink’s Reddit community shows latency swings, flagging download speeds, and dropped service are still very much an issue. Still, there are places where this kind of satellite internet could make sense.
A map of America’s broadband problems
Lots of places in America are still hugely underserved for broadband. And there are even more places like that in the rest of the world. The existing satellite projects aren’t ready to fix that just yet. But the network keeps growing, and more areas keep coming online. Even if Starlink can’t make it work, there’s always OneWeb, planning to launch full service at the end of 2022, or Project Kuiper after that. And if he finally does, the internet could start to look different.