Amazon plans over 3000 'Project Kuiper' satellites to spread global broadband
Roger Fingas
Amazon on Thursday revealed "Project Kuiper," an attempt to put thousands of satellites into low Earth orbit for worldwide internet access.
In all in the goal is to launch 3,236 small satellites, bringing "low-latency, high-speed broadband connectivity to unserved and underserved communities around the world," a spokesperson told CNBC. It's described as a "long-term" project, and one that will require partnering with other corporations. One of these will presumably be Amazon CEO Jeff Bezos' own spaceflight company, Blue Origin.
Kuiper will have to share orbit with similar networks like OneWeb and SpaceX's Starlink, the latter of which should eventually reach 4,425 satellites. Every company may have to contend with the problem of space junk — decades of public and private satellite launches have crowded orbit trajectories.
The project will ultimately further Amazon's own interests, letting more people shop from its online store, use its devices, and access services like Alexa and Prime Video.
Companies like Google and Facebook have experiemented with suborbital internet solutions with similar self-interest. None of these have advanced beyond the prototype stage, and in fact Facebook killed off its Aquila drone project last summer.
It's unknown how Amazon might deliver on its "low-latency" promise, since by definition satellite internet involves bouncing signals over extreme distances.
William Gallagher
David Schloss
Andrew Orr
Marko Zivkovic
Wesley Hilliard
Andrew O'Hara
22 Comments
It’s not unknown how Amazon can deliver on its low-latency promise. Low-altitude satellite arrays can be substantially faster than land networks because signals travel at the speed of light (signal over copper doesn’t and even fiber slows down due to frequent repeater hardware) and because the overall distance is shorter. Counterintuitive, but the 250 mile vertical round trip is much less than the inefficient path most signals take on the ground.
Here’s an excellent video that uses Starlink to demonstrate how these clusters achieve low latency: https://www.reddit.com/r/Starlink/comments/9voaoa/simulation_of_revised_starlink_phase_1/
The time to bounce a signal from ground to low earth orbit (2000km) and back at the speed of light is about 13ms. That's not terrible from a latency perspective, is it? In theory you could build a "low-latency" solution around that. Especially if the alternative is no access.
Google tells me that he low-end cost of launching a satellite is $50M. So launching 3000 would cost $150B. And obviously there are economies of scale available (could you launch 2? 5? 10? sats with one rocket?). It's an insane amount of money, but Apple has that much cash just lying around, so it's not impossible.
Mind boggling stuff.
The linked article mentions that other competitor can (or planning to) send up 30 satellites at a time, so now you're talking "just" a 100 or so launches (costing $10s of millions of dollars a pop) rather than thousands. It does seem feasible from a cost perspective for a company with deep pockets. Not clear how they recoup those costs, but that's another question.