The Offshore Voyaging Reference Site

New Satellite Communications Systems

By now, you’ve probably heard reports, rumours, and press releases about a new generation of satellite communications systems that are going to revolutionize everything.

Here’s what we know so far.

Who’s building this?

The main incumbents that a cruising sailor might consider are:

  • Iridium, which recently finished launching the 75 satellites of the Iridium NEXT constellation, providing low-throughput voice and data to compact portable devices.
  • Inmarsat, whose geostationary constellation provides higher throughput than Iridium, but at the cost of requiring larger terminal devices with directional antennas.

The major new players on the scene are:

  • SpaceX, which has just launched the first prototypes for the Starlink constellation of nearly 12,000 satellites.
  • OneWeb, currently building 650 satellites with a plan for 1,972 more.
  • Amazon, which is backing the Project Kuiper constellation of 3,236 satellites.
  • Samsung, which has plans for a 4,600 satellite constellation.
  • Facebook, whose Athena project is still largely under wraps, but is similar in scope to the other new entrants.

The Present Way

Iridium and Inmarsat devices provide nearly global coverage, but with very limited capabilities. Quasi-affordable Iridium mobile terminals max out at 128 kbit/s and usually operate at about one-tenth of that, with the best of the Iridium Certus units hitting 704 kbit/s at price points that make first-class airfare look like pocket change; Inmarsat BGAN and FleetBroadband terminals can reach 448 kbit/s.

Both have latencies in the high hundreds, and sometimes low thousands, of milliseconds. That’s good enough for a highly-compressed voice call, a few text-only emails, or downloading a weather map, but basic Web surfing ranges from painful to impossible, and video of any kind is simply not going to happen.

These limitations stem directly from the technology used by each network. To provide moderately high bandwidth, Inmarsat uses high-gain directional antennas and geostationary satellites. The only orbit that works is 35,785 km up, so each piece of data needs to travel at least 144,000 km; that’s nearly half a second of latency caused purely by the tardiness of light over such vast distances. What’s more, that one satellite is shared among everyone on your face of the globe. Equipping the satellite with dozens of super-narrow spot beams helps, but even so, hundreds of thousands of users are competing for that one orbiting antenna at any given time.

Iridium’s satellites are in much lower orbits, just 781 km above us. This is close enough for a compact handheld device to reach them, but their apparent speed is too fast to easily track with a high-gain dish; Iridium antennas are (nearly) omnidirectional. Their signal strength is therefore extremely low, limiting their throughput.

Combine the enormous capital expense of building and launching a small number of very sophisticated satellites, and real-world performance that is far below any terrestrial cellular or hard-wired option, and you’re left with a service that will, invariably, be very expensive relative to the size of its user base. Data and airtime on the existing networks are priced accordingly.

What Makes The New Ones Better?

The new entrants have something different in mind. Each of them is planning to launch far more communications satellites than have ever existed. Starlink alone will comprise six times more satellites than are currently in use by all nations, for all purposes, combined. The new networks will operate in low orbits, from roughly 350 km to 1500 km, eliminating the speed-of-light delay that is the Achilles heel of geostationary satellites.

The sheer number of satellites means that a dozen or more will usually be in view, with a clear line-of-sight link to at least two or three of them, from any point on earth. To track them, user terminals will be equipped with phased-array antennas, which can provide high gain, high signal strength, and extremely fast beam steering – they will compensate for the motion of the boat and of the satellite in milliseconds, rather than the several-second slew of a motorized Inmarsat dish.

Such systems will be capable of sustained low-latency streams at tens, possibly hundreds, of megabits per second to each user terminal. Theoretical speeds on the order of 1 Gbps might be possible. That performance makes them a viable alternative to fixed wireless, and in some cases, to hard-wired connectivity. The resulting huge user base should yield a much more favourable cost structure than currently exists.

Why Now?

The basic concept – a huge low-Earth-orbit constellation, providing fast high-bandwith communications to ships, aircraft, and remote explorers worldwide – has been around for decades. It’s only now, though, that it is becoming not just possible, but economically favourable, to build it.

Electronics manufacturing has reached truly incredible levels of efficiency in recent years. For $300, including the profit margins of everyone involved in making it, you can get a device that, in 20 seconds, churns through more calculations than all of the computers of the Apollo-Saturn spacecraft, NASA’s Mission Control, MIT, and all the subcontractors – combined – could perform during an entire nine-day mission. Applying that mass-produced efficiency to this problem makes the satellites, and all the ground infrastructure to go with them, affordable to build.

Rocket Science

The other big change is the Falcon 9 rocket. Launch prices plummeted when SpaceX’s big booster came on the scene; flights that might have cost $150M to $200M were suddenly being booked at $65M to $100M. Now that SpaceX is recovering and re-flying their rockets, instead of sinking a hundred-million-dollar vehicle in the Atlantic after each flight, the cost of launch is likely to come down further, and competing rocketry companies are all preparing for a new era of low-cost reusable rockets.

It’s All About…Money

What’s more, the commercial motivations have changed. SpaceX wants to go to Mars. They see the wireless communications sector, which pulls over $840 billion a year in global revenue despite having massive gaps in coverage and massive shortfalls in performance, as their best bet for making the money to go there.

Facebook and Amazon, per their nature, want dominant control over what people see, think, do, and buy in markets that are just beginning to open up. All players, recognizing the huge potential of this market and the value of first-mover advantage, are jockeying to be The One Who Connects Everybody.

When will they be available?

Login to continue reading (scroll down)

Inline Feedbacks
View all comments
Andre Langevin

Thanks Matt. Very interesting. It is indeed the promise of the current lowcost, parallel computing chipset architecture and all the improvment made in the last decade by signal processing. Don’t forget we went from ADSL to 500 Mbps (this is what i have at home right now) and i’m shy of paying 30 $more per month to have 1.5 Gbps internet at home. But where i work at the Quebec Automobile insurance we have patrol car on the road and we are still relying on 2G coverage for giving data access to patrol car all over the province because there are still LTE “blind spot”. 5G is arriving soon at hundreds of Mbps but no promise for extended coverage, millimetric wave goes far less than the current frequency.
But within 4 years we can expect that 5G will be available to coastal navigation as well. I’m pretty sure Murphy is working in these satellite projects and just wait for the first or second deployment before we get an unprecedented massive solar irruption that will wipe all satellite out there. Nobody can argue not, nobody can argue yes. We take for granted that a 4 B$ years old star has been under observation for 300 years and that is it “stable”. Good luck humans…

Philip Copeland

Thanks – very informative article. Let’s cross our fingers that this does become disruptive in2-3 years time.


Chris Steingraber

Thank’s Matt for putting this article together. This does seem promising and although I watched the launch of the first 60 starlink satellites I didn’t know about the tracking capabilities of the antenna which could give much more likelihood to applications on boats where they are rocking and rolling. Even though they are not planning for such applications directly to me it seems the most promising. The fact that Elon Musk is personally very skeptical of the risks of AI used wrongly would, I hope, mean that the ethical and security features are implemented in an acceptable way but it’s always hard to know what’s going on with so much technology. A new source of internet is always welcome to fight against monopolization of access to the internet and control of data. I know John was skeptical of his timelines which I just take with a grain of salt as it’s just marketing but Tesla regardless of their timelines has done incredible things to provide technology in the form of electric cars that perform better and cost less and SpaceX has done the same in their industry so I have a good impression of the general market potential as overall he seems to deliver just that, lower prices, better technology and better performance and that inspires some hope. I think the fact that regardless of when it comes out that is not the biggest thing to gripe about. The biggest impact in the greater context of things is that internet access to any land or sea remote parts of the world contributes to orders of magnitude higher connection and communication of those places and the rest of the wold and I believe that has the biggest impact on improving the world and those in worse circumstances. Education for remote areas, voice to areas in turmoil and injustice, and connectedness to the world allows places where health, education, work, technology, injustice and other issues are problems to catch up and close gaps to better standards of living in those areas over the years. Humans die in their lifespan and knowledge get refreshed and the more people can be education and work together with faster and vaster communication the better that can be for the quality of life for people and our caring for the planet too. A good resource to see how over the past 200 years quality of life across the globe has closed it’s gaps more and more from the third world countries to the 1st visit incredible site worth a visit. So yes I root for that and if it helps us too then that is a really profound way to improve the quality of life and connectedness and possibility to work remotely for those wishing to have a life with more freedom.

Marc Dacey

Well done. Andy Schell has persuaded me to get an Iridium phone, and I don’t care about the bandwidth or latency issues that much (I tried a primitive VoIP phone way back in the ’90s), as I consider it more or less an emergency item. Having an external antenna for it is easy enough to do and makes good sense. With an SSB/Pactor on our steel boat, a registered current EPIRB, an Iridium for emergency calls/taking to the raft/GRIBs when propagation stinks, and AIS tags on the well-tethered vests, I feel most of the bases are well-covered. Now, to get that VPN set up on the laptop….

Ben Logsdon

Andre, good last point there. One gnarly solar flare could trash all the satellites that we spent so much time and money to put up there. Luck is needed when gambling with nature!

Brent Cameron

Yesterday Starlink (Elon Musk’s latest venture) asked for approval to launch an additional 30,000 satellites (over the 12,000 they already have approval for). They plan on starting the launches in batches of 60 this fall – they already did one batch of 60 in the spring). They need another 6 launches (at say $75M/pop) to be able to launch the service. Musk says that economic viability kicks in at around 1,000 satellites (or about 16 more launches). There are currently about 2,000 total satellites in orbit right now with about 9,000 launched since Sputnik in 1957. I always about double any estimate on time that Musk gives and I’m still usually very optimistic but I don’t see a lot of new technology here that they haven’t already tested and put into orbit since they launched the second batch this spring (they launched 5-6 of them on an earlier launch to test the technology – the 60 were meant to be the first production satellites but they had a 5% failure rate (they have redundancy in their networks to cover this) and Musk has proven that the Falcon 9 launch systems are now quite reliable and able to launch on a very frequent basis – they’ve been constrained more by customers than by launch capabilities. Now I suspect they will be constrained by how many they can afford to throw up at $75M/launch. We don’t know what the actual cost of the satellites are but just the required launches will cost Starlink $1.2B for “economic viability”. If we assume that a customer would pay $50/month then you need almost 500K customers just to pay off the launch costs amortized over 5 years. The launches for all 42,000 would cost $52.5B excluding the cost of the satellites or of running them. Even at that rate, Musk believes that Starlink would only have about 3% of the total market (currently about $1T/year). Still, if they can bring in $30B/year on communications, it should be a pretty lucrative business for them – which is why Musk believes this is the best vehicle to fund his Mars ambitions.

Marc Dacey

I would just add that hundreds to thousands of miniature satellites all orbiting at the same LEO heights gives me the chills, however, and seems to me to be the potentially fatal flaw of any ambitious comsat constellation proposed to exponentially become future space junk. Yes, I know space is very, very big, but at these altitudes, it’s quantifiably so, and is an obvious target, as well.

Brent Cameron

One of the nice things about LEO though is that any “junk” very quickly falls out of orbit and burns up… we still have many dead satellites from the 60’s floating around up in geosynchronous orbits not to mention the debris from the Chinese and Indian ASAT tests. A bigger concern is how much it will change the night sky for astronomers… at 42,000 of these things, it will look like a highway up there!

Ivo Stankus

Wrong. Those satellites going up ARE production Starlink satellites, not some prototypes. They had no idea the satellites are going to be so bright, and they now recognize it as a major problem without knowing how to mitigate it. Starlink plans on applying an antireflective coating experimentally to ONE satellite going up in one of the batches next year, but they have no idea if it will work, or if it it will adversely affect the thermals. It’s a pointless service to boot, mobile broadband already covers about 80% of the world’s population, and will cover about 95% before Starlink is even functional…

John Harries

Hi Ivo,

That’s interesting, but please do not use that tone when disagreeing with someone. See (#4)

Andre Langevin

Good understanding of the problem of all the objects around earth

Philip Copeland

An interesting update from the President of SpaceX yesterday.

In a nutshell – she said they are aiming to start offering a consumer service in 2020 (in North America). To do that they need to launch 6 batches of satelittes (60 per launch??) – but are aiming for 24 launches by the end of 2020. Some interesting comments about US Military already testing – and achieving 610 megabits to a U.S. military C-12 twin-engine turboprop aircraft.

Worth reading the article.

Looking forward to that beer John – even if I have to wait until 2021!

John Harries

Hi Philip,

One nice possibility in all of this. If I am wrong and Elon pulls this off, it will give him huge positive cash flow. Maybe enough to take Tesler private, which I think would be a very good thing. Elon and public companies are not a good mix since being public stifles the man’s genius.

Philip Copeland

Hi John,

Tesla announced their results yesterday – “Tesla shares pop the most in 6 years after earnings show it’s building cash, boosting margins”. Worth reading the article –

As others have stated – the motivation behind Starlink is to provide cash flow to fund Musks “mission to mars”.

If you haven’t driven a Telsa I’d strongly recommend that you go rent one for a week or 2 to understand why they are being so disruptive. Its a little like when Apple introduced the first iPhone – it changed an entire industry forever – unless you happen to still use a flip phone (: !

John Harries

Hi Philip,

One quarter does not make a sustainable business. Still sticking with my prediction on Tesla. Now that the German Lux brands are serious about electric at the top end and others at the bottom Tesler will end up a niche brand at best, and probably bought by another car maker.

Flip phone! We have so much cool Apple tech that we get personal Christmas cards from Tim Cook. 🙂

Philip Copeland

Hi John,

Hope you have been keeping an eye on the Telsa results and stock price. I’m not seeing much chance of Elon taking this private!

On a serious note – keep your eye open for the upcoming Telsa “Battery Day”. Talk is that they have some great breakthroughs on the horizon – which will be fantastic for all of us – across every kind of device. Telsa recently acquired Maxwell Batteries – who have been leading much of the development of John Goodenough’s new solid state battery. (Goodenought was the original inventor of the LithiumIon battery and has just won the Nobel Prize for chemistry at the age of 97 – pretty amazing guy!) This is a good intro video on

John Harries

Hi Philip,

Yup, it’s either one of the greatest business success stories of all time, or the biggest stock bubble since the South Sea company. I’m still guessing the latter, but I could easily be wrong. And what Maxwell is doing is indeed interesting.

Philip Copeland

Also – first comments I have seen from a SpaceX/Starlink Exec on consumer pricing. Its vague – but nevertheless interesting.

John Harries

Hi Philip,

Thanks for the very interesting links. Still, I stick with my prediction of 2025 at the earliest for usable world wide internet on a boat at a reasonable price.

To me the key quote is this one:

SpaceX will have to hire a whole new workforce to deal with sales, tech support and product engineering. User terminals are a major concern. “The more engineering we do on the user terminal, the less service people we will have to hire,” said Shotwell, Terminals are one aspect of the Starlink business that the company has to “get right,” she said.

My thinking is that there is no question SpaceX can do amazing engineering far faster than most of us would believe but the delays will be in a high volume consumer roll out. The CEO herself acknowledges the difficulties.

(As the son of an engineer and inventor I can tell you that the chasm from proof of concept (first tweet) to viable consumer product is huge.)

That said, they do seem to be further ahead than I expected, so I could be wrong, and no one would be happier than me if I am. After all a beer is a pretty decent price for me to pay for world wide internet at a reasonable cost!

Stefan Smith

I agree with John, although would love to be proved wrong!

The key line is this: “For global coverage, which SpaceX would like to see up and running within 10 years, 24 launches would be required.”

There’s also a mention of military interest… so I’m guessing the military, followed by other big spenders (mining, oil, shipping) will take precedent.

I’ll bet on this tech trickling down to cruisers by 2035!

David Hoy

I bumped into the CEO of KVH at the Fort Lauderdale boat show last Friday (full disclosure, I used to work there years ago on the firmware for their very first Inmartsat sea-mobile antenna system). He does not seem worried about Starlink, but rather was talking about partnering with them and selling airtime. Starlink (and the other similar systems) would be extremely disruptive to Inmarsat and Iridium, and hence KVH’s core business that stands upon those platforms. Also talked with the CEO of Wave WiFi, not directly about Starlink, but the general state of cruiser-friendly Internet access. It would be awesome to have a system that would integrate WiFi, 4G/5G, and broadband satellite, and do automatic least-cost-routing between the various interfaces. That way you’d always have access, and always use the cheapest method for that access. Interesting times ahead 🙂

John Harries

Hi David,

Interesting that KVH are not too worried about Starlink. As you intimate, wheels within wheels on that one.

Christopher -

Fun read about Starlink and their business case.

Philip Copeland

Thanks – great article. I have worked in the Tech industry since the early 1980’s and started three different companies over that time. The magnitude/scale of what they are doing with SpaceX/Starlink is pretty incredible – I don’t recall anything that comes close to this in terms of sheer scale (or cost).


Philip Copeland

Latest news on Starlink.
Just had another successful launch yesterday (April 22) bringing the total to 420 Satelites in orbit. Most relevant was Elon Musk tweeting that a private beta would start in aprox 3 months followed by a public beta in 6 months “for high lattitudes”. We’ll see what that means before too long hopefully.

John Harries

Hi Philip,

Sounds good, but let’s not forget that they need thousands of satellites for a fully functioning global system. Still betting that’s a few years away, although I would love to be wrong.

Philip Copeland

Starlink is taking applicaitions for Beta Testers – starting later this summer. I believe they are interested in finding testers in more remote & and northern latitudes.

Sign up at


Philip Copeland

Some further information including shots of the small dish being used for beta testers and speculation on both speed (500mbs up & down) and pricing ($80 month) – still speculation however.

Philip Copeland

Speculation that Starlink to be priced at $99 / month with a $499 upfront cost for the dish.

John Harries

Hi Philip,

Hope it works out. That said, I won’t be signing up. Several decades in the computer industry taught me what a frustrating time sink being a beta tester can be, so I recommend that those with busy lives stay away from this for quite some time to come. Mind you, I don’t expect many to listen to me since new and cool is so seductive.

Andre Langevin

Remember John we use to call that “bleeding edge” because many IT enterprises were caught with the first iteration of a technology that was very costly to iron out the bugs.

The 499 $ dish will be for fixed installation. For a boat at sea a motorized antenna will be needed adding to the cost and complexity. I’m pretty confident that by 2023 it could be a good offering if someone can live with the entry price and the limitation. However…its always a matter of product positioning. There isn’t much R&D in a motorized antenna. The algorithms are known and many on open source. So i expect within a few years to have quality low cost motorized antennae’s.

I will wait at least 2 years…

John Harries

Hi Andre,

The news maybe better than that since I’m not sure a motorized antenna is required. Matt wrote on this a few months ago, and I know he follows it closely:

Even so, your 2023 guess may be a good one.

Andre Langevin

Hello John; directly from the mouth of the horse. Phased array (not moving) are much more costly and complex than a motorized one.

“At least before Musk’s January 7th, 2020 comment, it was believed that Starlink user terminals would have to rely almost entirely on high-performance phased-array antennas, referring to antennas that are steered electronically – i.e. without physically moving.

100% phased-array steering would likely result in the best possible user terminal from the standpoint of reliability and performance. However, full phased-array antennas – while making rapid progress – are still extremely expensive to manufacture compared to more basic alternatives, meaning that it could be an immense challenge – possibly much harder than building and launching Starlink satellites themselves – to mass-produce affordable user terminals under that paradigm. It’s possible that SpaceX has actually come to the same conclusion and is choosing to compromise with its first-generation user terminals, prioritizing time to market and cost per unit at the expense of peak performance and optimal reliability.”

In terms of dates, here is the calendar:
2020 – Service to be offered in northern US and southern Canada

Early 2021 – Service to be offered across the US and Canada up to 57° latitude

2021 – Continuous coverage between 57° South and North latitudes

Mar 2024 – FCC deadline to deploy half of the LEO constellation (2,213 satellites)

Nov 2024 – FCC deadline to deploy half of the VLEO constellation (3,759 satellites)

Mar 2027 – FCC deadline to deploy the LEO constellation (4,425 satellites)

Nov 2027 – FCC deadline to deploy the VLEO constellation (7,518 satellites)

John Harries

Hi Andre,

Interesting. On the time line, let’s not forget that “the mouth of the horse” has been a little erratic on delivery dates in the past…to put it kindly.

Philip Copeland

Good to see that Starlink are starting to do some testing themselves on their own fleet of boats. Also interetsing to note in the story the reported 700,000 requests to beta test. That’s a lot of demand if that number is correct. The market for RV’s and boats is going to be very large.

John Harries

Hi Philip,

I would not have any trouble believing that level of interest given how addicted we all are to high speed internet.

Robert Withers

Starlink has reduced the price of their “Maritime” option from $10k per month to £250 per month. The data limit for that is 50Gb – which is an awful lot of grib files. Whilst still not cheap, that price will make it viable to many bluewater sailors – especially as there is the option to stop/start subcription every month.

First report from a Boreal owner who used it while crossing the North Sea from Norway to Lerwick in lumpy conditions was very positive. V High Bandwidth (120mps), frequent dropout – probably caused by the seatate, but connection reestablished very quickly.

Maybe, just maybe, the promise of the last 25 years is starting to be realised

Mark Wilson

My fifteen year old was onboard with me over the easter holidays. She got through 15Gb in a day.

John Harries

Hi Mark,

Yikes, that could get expensive. Tic Toc I bet.

John Harries

Hi Robert,

Yes, I read about that over at Panbo. The interesting thing will be to see if this works mid ocean. Starlink relies on multiple ground stations so the big thing will be if they have satellite to satellite mesh working reliably and with enough band width. The other issue for yachts is going to be power use. And the final thing to think about is how erratic Elon is, so I would want a GO! along for backup.