Am I missing something? I feel like this website has no actual information on it. It's just pretty pictures and a bit text that doesn't talk about anything new.
Very excited about this. What's interesting is the deployment mechanism, where they almost use the momentum of the rocket to "fling" the 60 tightly packed satellites away from the booster, and then after about 3 hours they fire up the Krypton ion thrusters and start the course correction phase. IIRC the only other times this deployment mechanism is used is on tiny CubeSats where they opt for this more conglomerating type of deployment due to the durability of the CubeSats.
Will definitely be getting service for some family members in Northern Africa where incredibly slow speeds are charged high prices if Starlink can compete on price.
Last night at the sat. launch there was almost no fling effect. They slowly floated away in a group, hardly separated. I thought they could have had some mechanical problem.
Limited fuel. Thing is that ion thrusters have very very high specific impulse so you get a lot of oomph from them, especially if all you are doing is LEO stationkeeping.
Limited fuel, but ion thrusters are at least an order of magnitude more efficient (specific impulse) than the usual chemical propellants. The downside is low thrust, but these satellites can be patient when maneuvering between orbits.
I think they have limited fuel. According to the site, at the end of life, the satellites de-orbit. Maybe they will be recovered and recycled, too? Either way, the Starlink grid needs to be replenished with new satellites from time to time.
> the satellites de-orbit. Maybe they will be recovered and recycled, too?
They definitely de-orbit. Half of the reason their orbit is so low is to ensure that they de-orbit quickly once they are out of fuel (the other half is latency).
> the satellites de-orbit. Maybe they will be recovered and recycled, too?
Deorbit is a fancy word for burn up in the atmosphere.
They are within earths gravitational pull, so if they have a hard failure or run out of fuel to the point they can't "push back" they get vaporized on reentry. It's a clever fail safe.
Every satellite going around earth is "within earth's gravitational pull". The starlink satellites are so low that they are inside the rarefied upper layers of the atmosphere, meaning they have low but constant drag pulling on them. When they stop thrusting, they will eventually fall out of the sky.
LEO satellites deorbit due to friction with residual atmosphere, not because of gravitational pull. Orbiting already means you are in an equilibrium with gravity.
Fuel implies a chemical reaction that releases energy, which is almost always burning. (Some people extend the term to cover non-chemical reactions, but in any case Krypton is not used that way either).
The krypton isn't being used as a chemical reactant, and in fact being largely chemically inert is a benefit for ion thrusters. I suspect they're using krypton instead of xenon due to the latter being more expensive, even though the higher atomic mass of xenon would be beneficial.
Lower atomic mass gives better Isp, although for power-limited situation you usually want to maximize thrust instead.
In the pre-flight press conference the topic of krypton thrusters came up, and at first he made a superman joke, but then said the real answer was the lower cost of krypton.
Lower atomic mass gives better Isp for chemical rockets, for which the exhaust velocity is proportional to the temperature of the exhaust divided by the square root of the atomic mass of the exhaust product. However, these satellites are using ion propulsion. In ion engines, the exhaust velocity is constant, so a higher atomic mass allows each atom / molecule to carry more energy. This makes higher mass fuels like Xenon better for ion engines. (Radon would be better still but the radioactivity poses problems for various systems.)
Why is the exhaust velocity constant? I would naively expect the engine to impart roughy equal energy into each atom, so that lighter atoms would be faster. But I know little of the specifics of ion engines.
They might have gotten launch costs so low that a live test with the possibility of doing it over was cheaper than expensive test runs. Another comment mentioned SpaceX was close to losing some key license. Launching too late was probably going to cost more than the worst-case scenario cost of launching early.
Having two will offer some redundancy. You could probably signal process the RF given off by them to get to exactly the right mechanical screaming sound. The ISS already has the right kind of segmented cupola window.
Is there any information out there about what the terrestrial requirements to actually use this service are going to be? Is it going to require antennas similar to line-of-sight microwave internet providers like Monkeybrains? How much is that hardware expected to cost?
The satellites themselves, the way they're being deployed, what altitude they'll be in orbit...all of that is really fascinating, but I'm very curious what this is actually going to look like to use.
I don't think they've announced much regarding the pricing or specifics of the terminals, but Musk has said in an interview they will require an antenna array with line of sight, and will be about the size of a pizza box.
I think there was an interview where they talked about aiming to get the cost of a base station down to ~$200 (some number of hundreds, think it was two). That was probably cost to SpaceX not an end user though, and it obviously doesn't include cost of service.
Most people probably won't buy the terminal outright, though, I expect? You'll just buy a Starlink subscription with a fixed initial term and they'll lease you the terminal as part of that.
Airplanes, cruise ships, ski (and other) resorts, rich hermits, expensive summer camps, consumer trains, etc all do. For lots of commercial activities a thousand dollars is a rounding error compared to the costs of their other equipment too.
No, but they don't need to each buy one - they can pool their money and have one receiver for an entire village, and that may be much better than what they have now, which is either nothing or spotty 3g access
who gave you that impression - ofc there is GPS in orbit. it is just that the usual commercial GPS software refuses to deliver coordinates above a certain height or a certain speed.
this is ostensible to prevent GPS from being used in non American ICBM and cruise missile weapons.
the GPS birds are in a 20k km orbit and you should be able to use them just fine at lower orbits (starlink sats are planned to be in several fairly low orbit shells 320/550km).
I suspect you can even use GPS all the way out to say the moon if you are a bit creative. (catching GPS sats coming aroud the earth for positioning, or using a special dish antenna to get reflected GPS signals)
A double metric "kilo" prefix like that feels awkward. I know it's common, but I've always felt that we should just say "megameter" instead. (Also, a metric ton should just be called a "megagram".)
Agreed with megameter, but disagree with tonne: the kilogram is the base unit that physical constants etc. are measured against (e.g. Newton's constant is measured in Newton meter squareds per kilogram squared). It can be less confusing to talk about microkilograms than grams when doing physics.
Strictly speaking, the star trackers don't provide any position information; what they provide is a very accurate measurement of the satellite's orientation. And yes, they're commonly used on satellites.
I do wonder how well Starlink's antenna will handle roll. Current VSAT require rather large boat, but this is probably different.
I can see price for older boats increase purely due to Starlink. It's already somewhat cheap way of living and I do hope special tax provisions are made for people out in the ocean (I don't use any social benefits for paying my taxes in country X...)
It does go sideways, all sailboats do. But usually less than1 knot when sailing 5-10 knots.
Roll on monohulls is pretty big. Current generation of Youtubers seem to have switched to catamarans, which makes me too envious - I am not sure I can save up that much before responsibilities catch up on me.
I'm far from an expert on this, but I believe the biggest benefit to phased array antennas is that they don't have any moving parts, and adjusting where the signal is pointing is almost instant.
As long as the "pizza box" can track the sats, it should be able to communicate through some pretty intense speed and acceleration changes.
Stationary Starlink antennas will already need to handle aiming at around one degree per second just because of how quickly a satellite directly overhead will move. So the roll of a sailboat in a marina should be no problem at all, and maintaining a connection while underway is probably doable without much redesign.
And that's assuming there's any difficulty to aiming. Phased array radars have shown that merely sweeping the beam around in a predetermined pattern can be very quick. I don't know how much harder it would be to adjust that based on input from gyros.
I'm no expert, but I would think that the first customers will be the same people using the very expensive but low-latency terrestrial microwave links. Though now that I actually researched it, I don't know if Starlink will be able to beat 8.5ms round trip of this Chicago-NYC link.[0]
Not likely. The microwave paths they use now are surprisingly close to optimal straight shot. An additional 400 km up and 400 km down is going to blow out the latency budget.
So, technically what do you need to use their offering down the road? A satellite dish on the roof? Just a small antenna as built in in most phones today?
It’s a “user terminal” about the size of a pizza that can be mounted in a roof. Since it’s a phased array the direction doesn’t matter (unlike a traditional dish for a geostationary satellite). Once they are producing them in volume they should be pretty cheap.
> Since it’s a phased array the direction doesn’t matter
Of course it does matter, a phased array isn't magic. A steered antenna will have much better gain when the target is at low elevation, when a fixed phased array will be operating at very oblique angles.
What I mean is you don’t have to point a dish within arcseconds to get a connection. You just have to make sure it’s facing upwards. Per the Starlink Wikipedia page:
> Instead, it will be linked to flat user terminals the size of a pizza box, which will have phased array antennas and track the satellites. The terminals can be mounted anywhere, as long as they can see the sky
Even so, the website is just static content basically (except the globe at the top that you can rotate) so requiring webgl for the entire website (or even JS) feels a bit too much. Could just have hidden the globe if webgl/js is not enabled.
> Starlink is on the leading edge of on-orbit debris mitigation, meeting or exceeding all regulatory and industry standards.
> At end of life, the satellites will utilize their on-board propulsion system to deorbit over the course of a few months. In the unlikely event the propulsion system becomes inoperable, the satellites will burn up in Earth’s atmosphere within 1-5 years, significantly less than the hundreds or thousands of years required at higher altitudes.
"Starlink satellites are equipped with efficient ion thrusters powered by krypton that enable the satellites to orbit raise, maneuver in space, and deorbit at the end of their useful life. Starlink is the first krypton propelled spacecraft ever flown."
It is mentioned within the article : Either they will be de-orbited by their own ion thrusters or they will fall on earth their own within a year to 5 years.
>At end of life, the satellites will utilize their on-board propulsion system to deorbit over the course of a few months. In the unlikely event the propulsion system becomes inoperable, the satellites will burn up in Earth’s atmosphere within 1-5 years
So, how long a lifespan do these satellites have if their propulsion system works properly to keep them in orbit?
As a foreigner living in China, I'm wondering if such network can be technically prevented from being used in covered land/countries? How cool would it be to finally get an unrestricted access to the Internet here.
I believe the issue in f.ex. India is not censorship and population control, but rather a slow bureaucracy and in-fights between different government departments and the military. That said, things are changing in India. BSNL were just issued a license for offering IFMC services, including KA-band and L-band services.
Most probably, to use Starlink, they would have to abide by local laws and regulation. So in the case of China, I don't think the connection would be any different from what you can find consumer ISPs today. Or Starlink simply skips China.
AFAIK foreign satellite phone operators don't modify their service over China, but satellite phones are illegal to possess in China. I'd expect Starlink terminals would also be illegal to possess in China.
Unless SpaceX has implemented high speed frequency hopping (which would 10x the cost of end user devices), they can easily be jammed. It isn't hard to overpower a single from space with a ground based transmitter.
It is also easy to control the direction your receiver see it's signals from, to properly jam a satellite you need to fly the jamming equipment high in the air on a plane/balloon. A ground based jamming will have to be very powerful or close to block a receiver that is pointed to the sky.
Probably feasible, but would create international tensions, in worst case retaliation - many countries jamming foreign satellites or taking other hostile actions. Yuck.
Even if you implement frequency hopping it'll do nothing for China blocking it. The end users need to hop to the same frequencies to receive the signal, which means that a jammer can do the same.
This sort of thing works for militaries (e.g. for GPS) because you can assume that the receivers are secure, i.e. they're guarded by people carrying guns.
That's not really a viable plan when a hostile state can just buy a receiver with the built-in hopping algorithm from a retailer.
It works for (the US) military because they assign unique NET codes to groups and unique keys to transceivers. The frequency schedule is seeded using part of the key, the NET, and the day. The system is actually designed with the assumption that transceivers will fall in to enemy hands.
Starlink could implement something similar where the schedule is seeded using a key stored on the device, the day, and the receivers serial number (or other unique value). Supporting a million receivers is the same as supporting one, it is basically just one more input to a hash function.
It depends on how conspicuous the transmitter/receiver is. The authorities could just roll up and confiscate your antenna if such is easily visible and uncommon. Or deploy a jammer in your area. I think it would be about as safe as operating a pirate radio station.
What if the whole group of satellites overhead acted as one huge phased array? If that could be made to work, then it could be very hard to detect the uplink and downlink. This would be fiendishly difficult to pull off, however.
They won't be operating in China, as China will blow their satellites up if they open up service in China. Their govt cant imagine a world where people have access to non-state controlled media. What an Orwellian world we live in.
Ignoring the fact that there are already plenty of Western companies providing satellite internet coverage in China who haven't had their satellites blown up yet...
Destroying the satellites--or enough of them to sufficiently degrade Starlink's ability to operate in mainland China--would be a very expensive bit of overkill with a great deal of international blowback. Destroying one satellite with an ASAT weapon annoys everyone, but the consequences are somewhat limited; the debris from destroying hundreds or thousands in orbit is pretty much a casus belli. China still maintains enough authoritarian elements that it'd be easier to crackdown any dissidents using Starlink instead. Nation states have been tracking rogue radio signals by spies for decades, and they've got the resources to monitor for them.
If Starlink were to pose a problem in China, that's how they'd go about it. Destroying the satellites would be way too flashy. There's also the added benefit--from their perspective--that Starlink's mere presence would be a filter for their domestic surveillance; anyone willing to go through the effort of buying a likely illegal antenna, setting it up, and then trying to disguise it is the exact sort of person they'd want to monitor. Which is a pretty depressing thing to think about.
This first batch of satellites doesn't have the planned laser links for communication between satellites, so service in China will only be possible with a ground station in or near China. (The satellites just launched also only operate on two out of the three planned radio bands for connections with the ground.)
Why would they not include the inter-satellite linkage capability? That doesn't make any sense, given that was a solved problem in the Iridium constellation 20 years ago. That was a huge benefit of Iridium, in that it didn't have to rely on the "bent pipe" of just relaying data between a client and a ground station within the same coverage area. It was literally, "if you could talk to one satellite, anywhere in the world, you would have service".
> it also helps with latency if the signal doesn't traverse the constellation.
No it doesn't. Satellite-satellite lasers can follow a direct route at the speed of light in a vacuum, much like those microwave towers used by high-frequency traders. Done right it'll be lower latency than ground-based fiber optics (which snake over the terrain and only transmit signals at about 2/3 the speed of light).
Iridium does inter-satellite linkage with RF, Starlink intends to use laser comms. The advantage being much higher bandwidth, and not having to license the spectrum.
The laser comms are very much on the roadmap, however they were not ready in time for the first launch, and SpaceX has a hard deadline to meet for getting sats up there, so the first generation is going to just bounce your signal back to a ground station. (Which will bounce it to the next visible sat to push it forwards, unless it has fiber.)
China can both jam the satellite signals or (as Musk said during a Q&A) blow the satellites up. Selling directly to consumers will not be Starlinks main line of business, it will be backbone connections to local providers. So even if Starlink is unwilling to do any filtering, the local ISPs will do the filtering.
> Selling directly to consumers will not be Starlinks main line of business
It absolutely will be.
What they have said is that it won't be Starlinks main line of business at the beginning, until the receiver costs have been driven down and the constellation is much more dense. At that point, they very much intend to provide internet access directly to millions of people.
I remember Neil Stephenson once compared the cutting of undersea fiber optic cabling to the use of nuclear weapons: once one nation does it, every one is going to do it, so no one ever does it.
So I think China will innovate and find ways to block access to starlink, without attacking the satellites.
Could be useful underground and underwater, and on the Moon too!
It is not reasonable to increase program cost by 50% just to provide internet to 1000 people living in Antarctica, who already have satellite internet.
I really doubt it would be a 50% cost increase. They're already planning a few hundred satellites to fly in orbits with 81° inclination, which should provide some coverage for the coast of Antarctica. Adding a few dozen in a polar orbit would be more like a 0.6% increase in the total number of satellites planned, which I can imagine the military paying for with pocket change if the system as a whole proves to be successful.
If they added a polar orbit to the constellation, they would populate it with just one or two launches. So the costs of adding that polar coverage would be incurred basically all at once, and would have to be recouped over the lifetime of those satellites.
This is an amazing achievement. The pace at which SpaceX is doing things is nothing short of a miracle. Congratulations team SpaceX. I can only feel envy at the super cool engineering being done there.
So, part of Musk's long-term play here has got to be, eventually, to include a Starlink terminal into every Tesla, right? Probably without even charging any subscription. You and your passengers get free broadband internet in the car, and Tesla get a high-bandwidth channel to every vehicle on the road, allowing them to stream live, high-fidelity telemetry data back to their HQ and push out software updates as and when they need to.
One day they could possibly teleop cars autonomously (streaming sensor data to servers through the starlink terminals, then processing and sending output data back to the cars through the starlink terminals). Obviously latency is a huge factor in this, but it could be possible.
Interesting possibility. I don't think it would make sense for teleop to be the normal mode of operation, since it would consume a ton of bandwidth, but maybe they could use human teleoperators to handle occasional edges cases between "level 3" and "level 5" autonomy, in particular for the autonomous ride-sharing network.
They could colocate the teleoperators at Starlink ground stations to minimize latency, and prioritize teleop traffic.
My initial reaction was to make a joke about this, but the math is on point.
At the speed of light latency to/from the satellite (3.5ms) and a vehicle speed of 60mph, that's about 4 inches of vehicle travel. Sounds pretty great-- probably better than almost any human driver.
At terrestrial latency (30-60ms), thats a few feet, maybe nearing 10. You're still in the margin for human reaction times.
Doubling that to account for compute latency and such is actually still in the noise for car brakes available today.
Most super chargers are within about 120 miles of each other, so at 65mph, you have about 2 hours between them. They've done a pretty impressive job building out the infrastructure
I thought that most people wanted to avoid having to stop their trip for charging while driving because it takes too long (I don't even like to stop for gas and it is much quicker)... it's weird that you would not choose the best solution to your problem (and spend more money while doing it), but I guess some people don't care about time or money.
Also, what are those white spots on the map? Unpopulated areas with no charger in a 2h radius?
Your idea of what a tesla is like is far off. My 2012 tesla can drive over 250 miles. If I really gun it in the winter in the mountains its over 200. Then 30 mins charging at a supercharger. The new S can go over 370 miles without charging. 60 miles an hour * 4 hours, people stop to eat or pee.
I'm talking about a round trips of course... so you have to cut that in half if you can't or don't want to charge while on your trip (if you want to be able to come back home).
Most of the car trips that I do are round trips... and I would guess that yours are too?
I see what you are getting at. I don't have to go round trip very often and worry about charging because I have enough range, but also because because there are chargers everywhere and I can charge on the way back if needed. Most trips I take are within round trip range of my car. of course it's more convenient not to have to charge.
plugshare.com is a good site that shows all the chargers. supercharge.info is a 3rd party site that shows tesla chargers only, plus where the ones under construction are. tesla.com has one too.
I go skiing a lot, and there's one thing you learn - you use a lot of energy going uphill. An ev has much more accurate energy usage so you notice, gas cars of course use more energy doing more work too. But the huge difference is going back, going downhill, your electric car basically glides down hills for free, and because you are recharging the battery when you do moderate breaking, I can drive 20+ miles back and end up with more range than I started out with.
In Seattle I can easily go round trip to the 3 ski areas around town.
Basically I don't worry about it because I could charge on the way back if I needed it, and in 6 years of day trips I only charged once or twice on the way home - the one I remember was when I went to mt rainier for a week with no charging.
Neat video, really drives home how essential super chargers are to the usefulness of electric vehicles. Sitting for 12 hours in a trailer park to get another 200 miles of range is insane.
Also, this guy has some serious driving PTSD or something, he's so scared of cars that he thinks its dangerous to drive on an empty highway at night. Yikes.
No, they can only service a limited number of customers within a given region. The only users of Starlink, et al. in major metropolitan areas will be backbone providers.
"This system, if successful, would provide people in low to moderate population densities around the world with affordable high-speed internet access, including many who have never had internet access before,"
- Tom Praderio, SpaceX engineer
He’s emphasizing the use case that matters. That he doesn’t mention something of little interest (yet another broadband provider in urban areas) doesn’t mean it’s not something possible.
In the developing world, we've seen a leapfrog effect in telecoms - they skipped over fixed-line phones and went straight to mobile phones. It's far easier to blanket a rural area or a dense slum with 3G signal than to run wires to every house.
That poses a problem when you need fast fixed broadband - there's only so much mobile spectrum to go around, you can't really cheat the Shannon-Hartley theorem, so you end up with tight data caps and unreliable performance. Rolling out fixed infrastructure can be prohibitively expensive, so a lot of areas are stuck on mobile-only.
Low-orbit satellite broadband potentially breaks this dilemma. The low orbit provides tolerable latency (within 30ms of fixed broadband), there's an abundance of spectrum in the Ku and Ka bands and you can steer a tight beam with a phased array. Starlink and OneWeb could potentially offer a satellite broadband product that is competitive with fixed-line broadband on cost, latency and bandwidth, but that is available almost anywhere on earth. Becoming the default broadband provider for half the world's population is Kind Of A Big Deal - maybe enough of a big deal to bankroll a Mars mission.
I think there's obvious potential for this to be Kind Of A Big Deal on a purely personal level. Is your local cable monopoly absolutely terrible? Within a couple of years, it might not be a monopoly any more. Huge amounts of rural real estate becomes a heck of a lot more attractive if literally everywhere has good broadband. This might just be one of those once-in-a-generation technologies that meaningfully changes human geography.
Not a chance. People will still live in the cities. It just means that city which went from 5,000 people to 2,000 people in the last 20 years might go back to 5,000.
I am part of a group setting a tiny house community. We found being more than a hour from any major city center was not a problem for most of the people interested in our setup.
But limited mobile phone service and no high speed internet limits a lot of people. StarLink hopefully comes along by the time we start our planned future expansion.
At present we are three owners each building own house, we have 5 different people interested right now joining us to build their homes and our planning is no more than either 50 quarter acre or half acre lots which will let us leave about 15 acres untouched.
The real short term is just Wifi, with ethernet cables for the further homes. Once we get more people, we will look into fiber for everybody.
Point-to-point Wifi is powerful and is being deployed as an alternative to fixed lines in some places. E.g. see the Ubiquiti products. https://www.ui.com/products/#airmax
> Low-orbit satellite broadband potentially breaks this dilemma. ... Starlink and OneWeb could potentially offer a satellite broadband product that is competitive with fixed-line broadband on cost, latency and bandwidth.
But it doesn't, despite the marketing of Starlink as a "broadband" network they simply can't service high population density areas.
So it might be a backbone to a 4G tower and possibly usable by consumers in rural areas, but it won't be installed on the roof of a car nor offer connectivity to individual consumers in urban regions.
It won't be a reliable connection because the car moves around. Stuff like trees, tall building, bridges and tunnels will disrupt the signal. So for cars, a ground based cellular connection will work more reliable.
Antenna alignment could also be a factor, which becomes more difficult if the antenna is attached to a moving object.
The antenna has to constantly realign anyway due to how fast the satellites in low earth orbit move. Given that they're using phased array antennas, which can be steered electronically nearly instantaneously, there should be no issue tracking a satellite from a moving vehicle.
As mentioned earlier by dsr_, at 340km, a 1.5 degree beam covers a circle of just under 9km diameter. Let's call it 50 square km.
If a single satellite can cover a 9km diameter, I think it will be just fine for vehicles. Especially so as the idea is for overlapping satellites. Also, much of the starlink bandwidth will be used for extra cellular / wireless / internet backbone areas. Areas that need long haul fiber optics to have bandwidth suddenly won't, so it will actually improve existing internet service providers as well since their addressable market will increase.
SpaceX are planning on launching a lot of satellites - 2800 Ku/Ka-band satellites at 1150km and 7500 V-band satellites at 340km. Both will be capable of steering a 1.5 degree spot beam.
I'm not saying that you're definitely wrong, but Starlink is designed to provide a truly phenomenal amount of throughput. The V-band VLEO constellation is clearly intended to serve urban areas.
At 340Km, a 1.5 degree beam covers a circle of just under 9Km diameter. Let's call it 50 square Km.
Population density in Boston is over 5000/sqKm, so 250,000 people have to share that satellite. If 10% are subscribers and they use it 10% of the time, that's 2500 simultaneous users.
Starlink has claimed 20Gb/s per satellite, which gives 1 Gb/s per 125 simultaneous users in this high-density area. That's not great, but it's also not awful.
In any less population-dense area, it can be reasonably competitive, especially as the end-user deployment model is expected to be "here's a pizza-box sized antenna: make sure it's pointed in this general direction of the sky, and give it power and ethernet."
Yes. I seem to recall either Elon or more likely Gwynne Shotwell saying they're shooting for three or more satellites in view for the system to truly work the way it has been designed.
3 satellites makes for 24 mb/s per person. Except—I think it's actually 20 Gbps per satellite, so 24 mbps per person, so really only 3 mb/s, which is...usable, but slow.
Reasonably competitive? The competition is non existent in most of the inhabited world. In addition I have a friend on a farm in the US who can’t get a reliable internet connection in spite of the Sprint fiber that runs right in front of his house. This would be a no brainer for a lot of the population. Starlink could have more customers than ATT and Vodaphone combined.
Starlink satellites spend most of their time above sea and other uninhabited areas.
If I remember correctly, Starlink could potentially serve roughly 40 million customers globally in far to reach areas. Including ships and aircraft. It's not economic solution for the urban population.
> you can't really cheat the Shannon-Hartley theorem
No, but you can work within it's limits to achieve some fairly spectacular things.
I predict that if low-orbit sat service starts bringing competition to rural areas, we will see the cell companies start using spatial diversity antennae to dramatically up the available bandwidth.
I'm not sure how solving problems with multi-path and/or LOS helps here. The fundamental problem with rural is that the $ invested : customers served($ out) is drastically lower as population density diminishes.
Spatial diversity does more than help with multi-path issues. It multiplies your available bandwidth by sub-dividing the space around the base-station.
Bandwidth doesn't help you with HAAT[1] and the fact that you need to build out towers every couple miles or more depending on your terrain.
Towers are not cheap to build or maintain, if you only get a couple hundred people per tower it may not make sense to do(and why you don't see a lot of community WISPs except in unique cases).
I read an HN comment before that convinced me it's physically impossible for this system to work for more than a couple million devices at VERY low bandwidth (I think the quote was like 0.1mbps). Is this true?
I was all aboard the SpaceX hype train until that comment. I remember looking into his argument, and it seemed like everything checked out.
I mean if you live in a place where you just dont get internet even 100 kbps can be game changing. Sure you cant stream netflix. But being able to some degree go on the internet is pretty big.
Did the arguement stem from the problems of sharing bandwidth due to free space broadcast (instead of wires where everyone can re-use the same bandwidth)? Because, if so, the spot beams for the Starlink sats down to the ground terminal are only ~15 km in diameter. The uplink back to the sats may be a bit more constrained and won't scale. But downstream should be fine.
I'm sure that very smart person thought of problems in 5 minutes in an internet comment that the engineers who just put 60 satellites in orbit with didn't consider about their total addressable market...
Satellite internet is already available in rural areas and it doesn't seem to make that big of a difference. Most people care more about the culture that populated areas provide more than internet speeds.
Personally I pay $30 a month for 15 Mbps. I remember EarthLink or Prodigy being like $15-20 a month for 56Kbps. I remember leaving the modem on all night to download a 30MB demo of Monster Truck Madness in Elementary School. Now speeds are ~250 times what they were and the price has only doubled in the past 20 years? That's f-ing amazing. Prices are not getting out of control if you stick to the bottom end. I'm not even sure what I'd do with gig internet.
For areas going from no internet to satellite internet it's probably going to be amazing, but going from high latency satellite internet to low latency satellite internet will not be a huge game changer.
Adoption of http2 would probably help the quality of existing satellite internet more and be cheaper than this endeavor. As everyone started moving to HTTPS, the handshakes take much longer on a high latency connection and multiplexing would help alleviate that issue.
Agreed, but at this point in time I can't think of any "killer apps" for it. It seems silly to pay so much for something I wouldn't be able to use today.
I couldn't imagine replacing TV with streaming HD video or downloading free multi-gigabyte sample packs just to see what's in them before before multi-megabyte connections and affordable half terabyte hard drives. Most new or improved technology is speculative. No one knows what people will use it for 10 years later. Guesses rarely pan out.
Widespread gigabit is a recent development, and it's still expensive. Whatever technology takes advantage of it is a few years out. I had megabyte cable ~5 years before YouTube launched.
I don't have it available and I live in an urban area. Hell, I don't even have a viable alternative to my cable company. There's a ton of bureaucracy around all of it.
Political nonsense aside, we rarely see the use for something when it first comes out. That includes really useful things. e.g. all of the super important inventions that were straight up accidents.
Sure, it could take years, but I can't imagine having faster communication between more people not having a huge effect.
>Satellite internet is already available in rural areas and it doesn't seem to make that big of a difference.
It absolutely makes a difference. Rural/isolated communities are being left behind in terms of access to what are now basic staples of modern life: inexpensive, high-quality VoIP and video telecom, diverse sources of streaming video and audio media, real-time online gaming, telecommuting, etc.
The satellite internet available to most consumers today is cost-prohibitive garbage. Most options are doing satellite downlink paired to dialup uplink, and even the stuff that does two-way satellite links provides abysmal performance. All of it usually has a $100+ USD per month cost. Compared against 3G mobile, the latter is a far better deal.
Just to be clear, I don't think anyone is not moving to a rural area because of the internet speeds.
I'm not sure if you're talking about the U.S. or not, but my parents have satellite internet and they don't even notice a difference unless the dish is full of snow, which I'm not sure is a problem Starlink will solve. They stream, use Skype, and wifi calling. I don't notice much of a difference when I visit. If I try to log into a remote terminal, sure, but I'm not doing that. Internet might be more expensive, but rural properties are cheaper.
What will it mean to youtube, facebook, google, amazon, netflix, if all those sparse clients with slow internet become a broadband client... And, even more, a client that is driven away by some sort of capping? Of redirection to another site? Ads from another company?
Because net neutrality isn't exactly a thing everywhere, worldwide.
No, that's not the plan. To work with these low-orbit fast moving satellites, the antennas would be then size of a model 3 hood, pretty fancy/expensive and probably more than a little power hungry.
The cars will use LTE, although the backbone of those base stations could very well be Starlink.
Maybe future iterations will have extra hardware for terminal friendly applications, but I'd guess you'll see level 4 autonomy in a model 3 before they can upgrade Stsrlink to do that.
I would also like to see low orbit shielded datacenters. These would have relatively high mass and streamlined design, which would allow for radiation shielding, and reduce the necessity of boosting orbital trajectory. They would also have very high bandwidth line of sight laser communications. This would allow them to migrate the state of virtual servers to the next satellite over. So applications consisting of a cloud of server processes can incrementally migrate to virtually "hover" at very low altitudes above the earth.
Assuming Starlink also works, I suspect the first company to make this work would make an absolute killing.
Can't be served a warrant. Arguably superior physical security. Can have line of sight comms with the entire region, which are both more direct and have faster signal speeds than fiber.
For certain global applications, a low orbit cloud consisting of many datacenters has some advantages. The virtual servers could virtually migrate to "hover" wherever they are needed, or to wherever would minimize latency.
Yes. But your reading comprehension is spotty. Look at the cousin comment:
The way the "hovering" would work, is that the entire state of one virtual server would be serialized, then sent via line-of-sight high bandwidth comms to the next satellite to go over the region of interest. This way, the entire application's server cluster could incrementally migrate to remain over one geographic area, but do so at a much lower altitude than geosynchronous orbit.
You're not understanding the proposal. The way the "hovering" would work, is that the entire state of one virtual server would be serialized, then sent via line-of-sight high bandwidth comms to the next satellite to go over the region of interest. This way, the entire application's server cluster could incrementally migrate to remain over one geographic area, but do so at a much lower altitude than geosynchronous orbit.
So wait, we're sending entire server states from one satellite to another every few seconds, rather than routing the data to the satellite when needed? Seems like it would be a trade-off that would be worthwhile only for a small number of edge cases where there's a small server size, a fast start-up time, and a huge amount of very local traffic. In which case, put the server on the ground near the traffic? Or alternatively, move just a bit of the data (caches and some processing) around.
Turn every supercharger station into a LTE provider then put a dish on it and you're there. I'm glad spaceX isn't on the public market though. So we'll actually have a chance of seeing some of these ambitious goals come true.
Latency, being LEO, should be ~30ms, though it definitely depends on the number of satellites / coverage. Bandwidth can go up to a gigabit, but it's not clear how it'll be packaged and what the price will be.
Krypton (from Ancient Greek: κρυπτός, romanized: kryptos "the hidden one") is a chemical element with the symbol Kr and atomic number 36. It is a member of group 18 (noble gases) elements. A colorless, odorless, tasteless noble gas, krypton occurs in trace amounts in the atmosphere and is often used with other rare gases in fluorescent lamps. With rare exceptions, krypton is chemically inert.
What I don’t understand is how the Starlink satellites will connect to network clients. For instance, if you connect your phone to Starlink, does your phone have antennas powerful enough to propagate signal back into low earth orbit?? I can imagine satellites having powerful enough antennas/RF, but your cell phone?
Client devices will likely use Wifi/4G/5G to connect to a base station with a Starlink terminal. The terminal is too big and power hungry for a handheld device. But still, the ability to place a base station anywhere, even on a moving vehicle, is a game changer.
So no broadband in Greenland or northern Norway? If the graphics are accurate, Denmark gets high density coverage while north America is pretty sparse.
maybe i over simplified it a bit let me clarify. A Widely accessible sophisticated AI (openAI) with a strong block chain based incentive structure coupled with unrestricted access to the internet could possibly create an artificial organism which is self sustaining.
Pretty cool, it doesn't even have to be fast to make a difference, even old modem speeds will allow for payments, messages etc. It would have made a big difference for these two girls [0].
Not only is this technological amazing, hopefully it will provide competition to areas with monopolistic ISPs. More choices, lower prices, faster speeds!
Depends on how far you want to go, the real advantages come if they have satellite to satellite laser links (which I believe are not on this first batch), that could significantly cut down latency between say New York and Sydney.
I don't get how upload is suppose to work..
In the 90's satellite internet the upload was sent via phone line / ISDN.
Are we suppose to have powerful directional transmitters attached to our laptops?
Or do they intend for the Starlink internet to be distributed via regional ground stations and just fail to mention it?
Initially, there will need to be a lot of regional ground stations, but not because of any asymmetry in the connection. End user base stations will use a phased array antenna similar to what the satellites themselves use, though probably a bit smaller. Ground stations will be very important early on, because these satellites do not include the planned laser links for communication between each other, so they can only bounce signals between end users and ground stations.
370 comments
[ 3.0 ms ] story [ 366 ms ] threadWill definitely be getting service for some family members in Northern Africa where incredibly slow speeds are charged high prices if Starlink can compete on price.
this is compulsory anyway due to physics.
Or, you just use thrusters, like the ones that launched from the Shuttle did.
> autonomously perform maneuvers to avoid collisions with space debris and other spacecraft
Pretty cool.
Also, the north and south pole are conspicuously avoided by Starlink, confirming my suspicions about the locations of intergalactic space ports.
They definitely de-orbit. Half of the reason their orbit is so low is to ensure that they de-orbit quickly once they are out of fuel (the other half is latency).
Deorbit is a fancy word for burn up in the atmosphere.
They are within earths gravitational pull, so if they have a hard failure or run out of fuel to the point they can't "push back" they get vaporized on reentry. It's a clever fail safe.
Every satellite going around earth is "within earth's gravitational pull". The starlink satellites are so low that they are inside the rarefied upper layers of the atmosphere, meaning they have low but constant drag pulling on them. When they stop thrusting, they will eventually fall out of the sky.
In the pre-flight press conference the topic of krypton thrusters came up, and at first he made a superman joke, but then said the real answer was the lower cost of krypton.
https://en.wikipedia.org/wiki/Specific_impulse
Are you saying that there will be intergalactic space ports at the poles in the future?
https://en.wikipedia.org/wiki/Nazi_UFOs
> Starlink is the first krypton propelled spacecraft ever flown.
So they're using non-flight-proven technology (Technology readiness level at most 6)... impressive!
I thought this was some scifi stuff, straight out of kerbal space program. But they appear to be actually viable!
https://en.wikipedia.org/wiki/ABS-3A https://www.youtube.com/watch?v=zI0Z68Yv5Ms
The satellites themselves, the way they're being deployed, what altitude they'll be in orbit...all of that is really fascinating, but I'm very curious what this is actually going to look like to use.
the GPS birds are in a 20k km orbit and you should be able to use them just fine at lower orbits (starlink sats are planned to be in several fairly low orbit shells 320/550km).
I suspect you can even use GPS all the way out to say the moon if you are a bit creative. (catching GPS sats coming aroud the earth for positioning, or using a special dish antenna to get reflected GPS signals)
A double metric "kilo" prefix like that feels awkward. I know it's common, but I've always felt that we should just say "megameter" instead. (Also, a metric ton should just be called a "megagram".)
Wouldn't reflected signals be inaccurate due to additional time of flight?
https://gdmissionsystems.com/en/communications/spaceborne-co...
This could have a huge impact on people wanting to live out of a van remotely and still be connected.
When I can get cheap high-speed low-latency internet in the middle of the ocean then I'm buying a boat and going to sea.
I can see price for older boats increase purely due to Starlink. It's already somewhat cheap way of living and I do hope special tax provisions are made for people out in the ocean (I don't use any social benefits for paying my taxes in country X...)
I'm thinking some sort of stabilization technology similar to how they manage to land rockets on ships. But not rocket-powered, of course!
Roll on monohulls is pretty big. Current generation of Youtubers seem to have switched to catamarans, which makes me too envious - I am not sure I can save up that much before responsibilities catch up on me.
As long as the "pizza box" can track the sats, it should be able to communicate through some pretty intense speed and acceleration changes.
And that's assuming there's any difficulty to aiming. Phased array radars have shown that merely sweeping the beam around in a predetermined pattern can be very quick. I don't know how much harder it would be to adjust that based on input from gyros.
Or an anarcho-capitalist hyper libertarian micro-nation perched on top of a spar buoy.
[0]https://www.zerohedge.com/news/chicago-new-york-and-back-85-...
And where do I sign up? :-)
Of course it does matter, a phased array isn't magic. A steered antenna will have much better gain when the target is at low elevation, when a fixed phased array will be operating at very oblique angles.
> Instead, it will be linked to flat user terminals the size of a pizza box, which will have phased array antennas and track the satellites. The terminals can be mounted anywhere, as long as they can see the sky
Hoping to see coverage added for Australia!
"Your browser does not support WebGL"
> KEEPING SPACE CLEAN
> Starlink is on the leading edge of on-orbit debris mitigation, meeting or exceeding all regulatory and industry standards.
> At end of life, the satellites will utilize their on-board propulsion system to deorbit over the course of a few months. In the unlikely event the propulsion system becomes inoperable, the satellites will burn up in Earth’s atmosphere within 1-5 years, significantly less than the hundreds or thousands of years required at higher altitudes.
So from a space debris standpoint, it's "walk away safe" in that a dead satellite can do nothing but fall into the atmosphere.
So, how long a lifespan do these satellites have if their propulsion system works properly to keep them in orbit?
As a foreigner living in China, I'm wondering if such network can be technically prevented from being used in covered land/countries? How cool would it be to finally get an unrestricted access to the Internet here.
But I'm probably just dreaming.
[0] https://www.outfittersatellite.com/Countries-with-Satellite-...
This sort of thing works for militaries (e.g. for GPS) because you can assume that the receivers are secure, i.e. they're guarded by people carrying guns.
That's not really a viable plan when a hostile state can just buy a receiver with the built-in hopping algorithm from a retailer.
Starlink could implement something similar where the schedule is seeded using a key stored on the device, the day, and the receivers serial number (or other unique value). Supporting a million receivers is the same as supporting one, it is basically just one more input to a hash function.
I can't imagine finding a transmitter capable of reaching LEO being all that hard for a country.
If Starlink were to pose a problem in China, that's how they'd go about it. Destroying the satellites would be way too flashy. There's also the added benefit--from their perspective--that Starlink's mere presence would be a filter for their domestic surveillance; anyone willing to go through the effort of buying a likely illegal antenna, setting it up, and then trying to disguise it is the exact sort of person they'd want to monitor. Which is a pretty depressing thing to think about.
No it doesn't. Satellite-satellite lasers can follow a direct route at the speed of light in a vacuum, much like those microwave towers used by high-frequency traders. Done right it'll be lower latency than ground-based fiber optics (which snake over the terrain and only transmit signals at about 2/3 the speed of light).
The laser comms are very much on the roadmap, however they were not ready in time for the first launch, and SpaceX has a hard deadline to meet for getting sats up there, so the first generation is going to just bounce your signal back to a ground station. (Which will bounce it to the next visible sat to push it forwards, unless it has fiber.)
!
They don’t just sit over China. There’s no way they could just start shooting them down without repercussions.
It absolutely will be.
What they have said is that it won't be Starlinks main line of business at the beginning, until the receiver costs have been driven down and the constellation is much more dense. At that point, they very much intend to provide internet access directly to millions of people.
So I think China will innovate and find ways to block access to starlink, without attacking the satellites.
It is not reasonable to increase program cost by 50% just to provide internet to 1000 people living in Antarctica, who already have satellite internet.
They could colocate the teleoperators at Starlink ground stations to minimize latency, and prioritize teleop traffic.
At the speed of light latency to/from the satellite (3.5ms) and a vehicle speed of 60mph, that's about 4 inches of vehicle travel. Sounds pretty great-- probably better than almost any human driver.
At terrestrial latency (30-60ms), thats a few feet, maybe nearing 10. You're still in the margin for human reaction times.
Doubling that to account for compute latency and such is actually still in the noise for car brakes available today.
Cool how far things have come.
Never have needed a faster Internet in a Tesla and all passengers will be carrying their phones anyway.
https://i.imgur.com/leKbJ4k.jpg (Map from April this year)
Most super chargers are within about 120 miles of each other, so at 65mph, you have about 2 hours between them. They've done a pretty impressive job building out the infrastructure
Also, what are those white spots on the map? Unpopulated areas with no charger in a 2h radius?
Most of the car trips that I do are round trips... and I would guess that yours are too?
Makes more sense?
plugshare.com is a good site that shows all the chargers. supercharge.info is a 3rd party site that shows tesla chargers only, plus where the ones under construction are. tesla.com has one too.
I go skiing a lot, and there's one thing you learn - you use a lot of energy going uphill. An ev has much more accurate energy usage so you notice, gas cars of course use more energy doing more work too. But the huge difference is going back, going downhill, your electric car basically glides down hills for free, and because you are recharging the battery when you do moderate breaking, I can drive 20+ miles back and end up with more range than I started out with.
In Seattle I can easily go round trip to the 3 ski areas around town.
Basically I don't worry about it because I could charge on the way back if I needed it, and in 6 years of day trips I only charged once or twice on the way home - the one I remember was when I went to mt rainier for a week with no charging.
I'm willing to bet that you are an outlier
The latest model goes 370. not 500.
Also, this guy has some serious driving PTSD or something, he's so scared of cars that he thinks its dangerous to drive on an empty highway at night. Yikes.
People have driven them around Australia. Can't get much further away from populated areas than that.
e.g. https://www.topgear.com/car-news/electric/woman-drove-her-te...
Source: https://www.scientificamerican.com/article/spacex-launches-t...
In the developing world, we've seen a leapfrog effect in telecoms - they skipped over fixed-line phones and went straight to mobile phones. It's far easier to blanket a rural area or a dense slum with 3G signal than to run wires to every house.
That poses a problem when you need fast fixed broadband - there's only so much mobile spectrum to go around, you can't really cheat the Shannon-Hartley theorem, so you end up with tight data caps and unreliable performance. Rolling out fixed infrastructure can be prohibitively expensive, so a lot of areas are stuck on mobile-only.
Low-orbit satellite broadband potentially breaks this dilemma. The low orbit provides tolerable latency (within 30ms of fixed broadband), there's an abundance of spectrum in the Ku and Ka bands and you can steer a tight beam with a phased array. Starlink and OneWeb could potentially offer a satellite broadband product that is competitive with fixed-line broadband on cost, latency and bandwidth, but that is available almost anywhere on earth. Becoming the default broadband provider for half the world's population is Kind Of A Big Deal - maybe enough of a big deal to bankroll a Mars mission.
I think there's obvious potential for this to be Kind Of A Big Deal on a purely personal level. Is your local cable monopoly absolutely terrible? Within a couple of years, it might not be a monopoly any more. Huge amounts of rural real estate becomes a heck of a lot more attractive if literally everywhere has good broadband. This might just be one of those once-in-a-generation technologies that meaningfully changes human geography.
With self driving cars you can go out, drink alcohol, socialize, and get back home whenever you want, you can even sleep in the car
By building a GSV and naming it Kind Of A Big Deal.
But limited mobile phone service and no high speed internet limits a lot of people. StarLink hopefully comes along by the time we start our planned future expansion.
At present we are three owners each building own house, we have 5 different people interested right now joining us to build their homes and our planning is no more than either 50 quarter acre or half acre lots which will let us leave about 15 acres untouched.
The real short term is just Wifi, with ethernet cables for the further homes. Once we get more people, we will look into fiber for everybody.
But it doesn't, despite the marketing of Starlink as a "broadband" network they simply can't service high population density areas.
So it might be a backbone to a 4G tower and possibly usable by consumers in rural areas, but it won't be installed on the roof of a car nor offer connectivity to individual consumers in urban regions.
How come?
Antenna alignment could also be a factor, which becomes more difficult if the antenna is attached to a moving object.
If a single satellite can cover a 9km diameter, I think it will be just fine for vehicles. Especially so as the idea is for overlapping satellites. Also, much of the starlink bandwidth will be used for extra cellular / wireless / internet backbone areas. Areas that need long haul fiber optics to have bandwidth suddenly won't, so it will actually improve existing internet service providers as well since their addressable market will increase.
I'm not saying that you're definitely wrong, but Starlink is designed to provide a truly phenomenal amount of throughput. The V-band VLEO constellation is clearly intended to serve urban areas.
https://licensing.fcc.gov/myibfs/download.do?attachment_key=...
Population density in Boston is over 5000/sqKm, so 250,000 people have to share that satellite. If 10% are subscribers and they use it 10% of the time, that's 2500 simultaneous users.
Starlink has claimed 20Gb/s per satellite, which gives 1 Gb/s per 125 simultaneous users in this high-density area. That's not great, but it's also not awful.
In any less population-dense area, it can be reasonably competitive, especially as the end-user deployment model is expected to be "here's a pizza-box sized antenna: make sure it's pointed in this general direction of the sky, and give it power and ethernet."
thats per satellite... wouldnt you, in theory, have access to multiple satellites? i though the theory was overlap, etc...
If I remember correctly, Starlink could potentially serve roughly 40 million customers globally in far to reach areas. Including ships and aircraft. It's not economic solution for the urban population.
No, but you can work within it's limits to achieve some fairly spectacular things.
I predict that if low-orbit sat service starts bringing competition to rural areas, we will see the cell companies start using spatial diversity antennae to dramatically up the available bandwidth.
Towers are not cheap to build or maintain, if you only get a couple hundred people per tower it may not make sense to do(and why you don't see a lot of community WISPs except in unique cases).
[1] https://www.fcc.gov/media/radio/haat-calculator
I was all aboard the SpaceX hype train until that comment. I remember looking into his argument, and it seemed like everything checked out.
Personally I pay $30 a month for 15 Mbps. I remember EarthLink or Prodigy being like $15-20 a month for 56Kbps. I remember leaving the modem on all night to download a 30MB demo of Monster Truck Madness in Elementary School. Now speeds are ~250 times what they were and the price has only doubled in the past 20 years? That's f-ing amazing. Prices are not getting out of control if you stick to the bottom end. I'm not even sure what I'd do with gig internet.
For areas going from no internet to satellite internet it's probably going to be amazing, but going from high latency satellite internet to low latency satellite internet will not be a huge game changer.
Adoption of http2 would probably help the quality of existing satellite internet more and be cheaper than this endeavor. As everyone started moving to HTTPS, the handshakes take much longer on a high latency connection and multiplexing would help alleviate that issue.
We're going to find out and that's the most interesting part.
I don't have it available and I live in an urban area. Hell, I don't even have a viable alternative to my cable company. There's a ton of bureaucracy around all of it.
Political nonsense aside, we rarely see the use for something when it first comes out. That includes really useful things. e.g. all of the super important inventions that were straight up accidents.
Sure, it could take years, but I can't imagine having faster communication between more people not having a huge effect.
It absolutely makes a difference. Rural/isolated communities are being left behind in terms of access to what are now basic staples of modern life: inexpensive, high-quality VoIP and video telecom, diverse sources of streaming video and audio media, real-time online gaming, telecommuting, etc.
The satellite internet available to most consumers today is cost-prohibitive garbage. Most options are doing satellite downlink paired to dialup uplink, and even the stuff that does two-way satellite links provides abysmal performance. All of it usually has a $100+ USD per month cost. Compared against 3G mobile, the latter is a far better deal.
I'm not sure if you're talking about the U.S. or not, but my parents have satellite internet and they don't even notice a difference unless the dish is full of snow, which I'm not sure is a problem Starlink will solve. They stream, use Skype, and wifi calling. I don't notice much of a difference when I visit. If I try to log into a remote terminal, sure, but I'm not doing that. Internet might be more expensive, but rural properties are cheaper.
Hi, I'm your counterexample. I held out moving to a rural area until the internet situation got better. It's great out here with a yagi on the roof.
Definitely not true. It is a big decision point for a lot of people.
Because net neutrality isn't exactly a thing everywhere, worldwide.
The cars will use LTE, although the backbone of those base stations could very well be Starlink.
Maybe future iterations will have extra hardware for terminal friendly applications, but I'd guess you'll see level 4 autonomy in a model 3 before they can upgrade Stsrlink to do that.
Assuming Starlink also works, I suspect the first company to make this work would make an absolute killing.
Why? Build on the ground with a fibre, far cheaper and far lower latency.
For certain global applications, a low orbit cloud consisting of many datacenters has some advantages. The virtual servers could virtually migrate to "hover" wherever they are needed, or to wherever would minimize latency.
Yes. But your reading comprehension is spotty. Look at the cousin comment:
The way the "hovering" would work, is that the entire state of one virtual server would be serialized, then sent via line-of-sight high bandwidth comms to the next satellite to go over the region of interest. This way, the entire application's server cluster could incrementally migrate to remain over one geographic area, but do so at a much lower altitude than geosynchronous orbit.
You're not understanding the proposal. The way the "hovering" would work, is that the entire state of one virtual server would be serialized, then sent via line-of-sight high bandwidth comms to the next satellite to go over the region of interest. This way, the entire application's server cluster could incrementally migrate to remain over one geographic area, but do so at a much lower altitude than geosynchronous orbit.
https://en.wikipedia.org/wiki/Krypton
Any RF nerds know if this is easy to do?
I look forward to seeing more details.
its unclear if elon wants to avoid an ai apocalypse or just get ahead of it.
[0] https://en.wikipedia.org/wiki/Deaths_of_Kris_Kremers_and_Lis...
the further your communication has to travel, the better starlink will be for latency