Starlink speeds past terrestrial networks – and regulators

Starlink can sometimes shift data more quickly than is possible on terrestrial networks, and improves connectivity in remote areas. But the space broadband service also presents new technical and regulatory challenges, according to speakers who took to the stage on Tuesday at the Asia Pacific Regional Internet Conference on …

[Tom7]

Jakarta are still a bit pissed because they spent several billion on a geostationary bird for remote internet access just before starlink became available.

[Michael Hoffmann]

<NBN Skymuster entered the chat>

[DS999]

Starlink is not intended for urban areas like Jakarta

It can never provide sufficient bandwidth to be a meaningful competitor in such places, and wouldn't be competitive price wise anyway.

I wonder if the reason why you see fewer Starlink satellites in Jakarta is because it is an urban area. Buildings block signals, while in rural areas so long as you aren't in a forest or in a deep valley you can see a lot more sky than in a city.

[MachDiamond]

Re: Starlink is not intended for urban areas like Jakarta

"I wonder if the reason why you see fewer Starlink satellites in Jakarta is because it is an urban area."

Just about any other option is going to be faster and less expensive if there is a choice. The key metric is customer density. In a city, it's a no-brainer to install cable/fiber. In the suburbs, it's not as rich, but good enough to turn a profit. In the country it can take miles of cable to hook up one person. If the terrain allows it, terrestrial wireless internet is simple and works well. For satellite to make a play, it needs to be servicing people way out and in difficult geography.

What concerns me is Starlink delaying or shelving projects where the margins on the fence for terrestrial connections. I don't think that Starlink is viable long term once they are out of the investor/startup phase and have to earn everything through paying customers. The cost to build those other systems is only going to go up over time, at least in numerical amounts. A Falcon rocket with a load of replacement satellites is a huge pile of money in infrastructure costs that is built into how the system works. I doubt my cable company has put $5mn worth of work into my local system over the last 5 years.

[Taliesinawen]

Re: Starlink is not intended for urban areas like Jakarta

ClippyAI: “Indonesia's regulations require Starlink to focus initially on remote locations as part of its licensing, targeting health clinics and education in far-flung areas before full urban rollout.”

[Oneman2Many]

Re: Starlink is not intended for urban areas like Jakarta

With Falcon, my guess is its not viable long term but with Starship they can launch 5x the amount of bandwidth per launch at less cost.

[MachDiamond]

Re: Starlink is not intended for urban areas like Jakarta

"With Falcon, my guess is its not viable long term but with Starship they can launch 5x the amount of bandwidth per launch at less cost."

It would be good to be able to spam long strings of satellites with Starship today, but that might not be viable when the constellation is needing a few replacements in certain strings. Starship is less expensive when using the full promised payload. There's a certain cost to launching that beast that doesn't vary with the mass of the payload. It's mostly volume that's a factor with Starlink or putting them up with F9H would make sense. I don't see "bandwidth" as a useful metric. Coverage is going to be more important since not having any is more of an issue over being throttled. My cable connection is nice and fast or doesn't work. I'd happily trade slower for more reliable.

[Oneman2Many]

Re: Starlink is not intended for urban areas like Jakarta

Bandwidth is the best metric, ultimately its is what is needed to satisfy your customers. The key is start filling the shells where there are the most customers and where you are bandwidth constrained. You leave the existing 12,000 satellites to provide global coverage. In fact they already have global coverage with the 10,000 sats they will have by March, so coverage isn't an issue

As for replacements, Assume 5 year lifespan, that is 50 a month they need to keep launching, or one Starship launch. SpaceX would like to get around 30,000 v3 sats including direct to cell, we'll see about that.

Cost of F9 launch for Starlink is around $50m, cost of Starship is thought to be closer to $20m once they have regular cadence.

[MachDiamond]

Re: Starlink is not intended for urban areas like Jakarta

"In fact they already have global coverage with the 10,000 sats they will have by March, so coverage isn't an issue"

Total build out is supposed to be around 48,000 satellites according to Elon. I know, he's not that good with numbers. The replacement requirements also have to include sats that have failed early. It's closer to 26 replacements per day. 5 years is 1,825 days. Divide 48,000 satellites by 1,825 and you get 26.3 satellites per day hitting their best by date. I'm going to assume that a few will last a bit longer and that will balance out the ones that fail early.

You may have missed my point that having the binary choice of lots of bandwidth or nothing at all isn't a good choice to make if you can have less bandwidth that's on all of the time.

Starship will never launch for $20mn. Just what they'll have to figure in for amortized development and update costs will be more than that much. Propellant isn't hugely expensive, but SS uses thousands of metric tons which requires another massive investment in ground infrastructure to make it since there is a risk in building out that much capacity for one customer. If SpaceX does it in-house, that's another big sunk capex. Elon's glorious new future where Starships are launching every few hours ignores things such as supplies.

[Anonymous Coward]

Re: Starlink is not intended for urban areas like Jakarta

That had me confused as well... but Jakarta is near the equator, which presents two issues: 1) the satellites seem to "bunch up" at higher latitudes due to the inclination of the orbits and 2) the user terminals are prohibited from "pointing" to a band over the equator, so a user in Jakarta can't communicate with satellites directly overhead.

Of course, urban canyons won't help either.

#2 is because Starlink shares frequencies with some geostationary satellites, and the GEO get priority.

[vtcodger]

Re: Starlink is not intended for urban areas like Jakarta

"I wonder if the reason why you see fewer Starlink satellites in Jakarta"

It's just an artifact of the way that near circular satellite orbits work. It's not that easy to describe in words, but the result is that the density of satellites is lower near the equator than in higher latitudes. European, North American, and East Asian locations will see more Starlink satellites at any given time than will equatorial locations like Singapore, Jakarta and Quito.

[_wojtek]

"remote areas"

I kinda fear that those remote areas with further urban sprawl becase people won't feel limited by access... which is kinda a bit worrying...

[stiine]

Re: "remote areas"

Did you miss the 17000 islands part of the article? The country itself is sprawling.

[Ol'Peculier]

I;ve only used Starlink once, in a remote bar on one of the Galapagos islands, and despite the fact I despise Musk and everything he tries to f@$k up, was hugely impressed with the speed.

If I was to move to somewhere remote with no broadband access, I'd need two things. Access to a decent pub and satellite...

[Roland6]

>” If I was to move to somewhere remote with no broadband access, I'd need two things. Access to a decent pub and satellite...”

I take it, Starlink was the only redeeming feature of that “remote bar”.

[Ol'Peculier]

The pub on Isabela brewed it's own beer, and also was been visited by a very nice lass from Germany who I very much enjoyed talking to.

Here, I'm looking at somewhere on the Moors. The nearer to Blakey Ridge the better. If you know, you know...

[Roland6]

Whilst the Lion Inn does look interesting, I’m a little concerned the (decent) beer selection only warrants a single sentence in the menu.

For your consideration a pint from the Potbelly brewery ->

[Mozzie]

Zaroa Brewery by any chance?

Sad if Starlink was the redeeming feature. Between that and the Lion Inn I know which I'd rather be sat outside today but I'd guess the signal was better in Puerto Villamil too?

[Ol'Peculier]

Re: Zaroa Brewery by any chance?

Gold star to you, sir!

Was just a shame my knees were knackered by all the boat transfers. But it's on the list of places I'd love to go back to. They seem to have doubled the entry cost as well, but TBH if it gets rid of some of the more, um, crass people that were in the bars on San Cristobal then it could be only a good thing...

[Sam not the Viking]

I was thinking that you need to get out more (moor?). But Galapagos and North Yorkshire, despite both being remote, distant and inhabited by strange beasts (I do have a connection....), indicate the search for the perfect pint is never ending. Like software, needing constant updates.

Cheers ---->

[Alumoi]

I've bought a Starlink mini for my ham radio POTA and SOTA. Works flawlessly. So go Starlink!

Musk? Fuck him!

[DS999]

Musk is busy combining all his companies

First xAI bought X (at 3-4x what it was actually worth) and now Starlink and xAI have merged meaning Starlink owns X. I'm sure at some point he'll do the same with Tesla. His dream of an "everything app" is a massive failure, but he's gonna do an "everything corporation" as a consolation prize.

[MachDiamond]

Re: Musk is busy combining all his companies

"but he's gonna do an "everything corporation" as a consolation prize."

Maybe Elon sees that he's going nowhere but, if he can make one company that is really huge, it will far into "too big to fail" and governments will bail him out of any grievous mistakes.

[Anonymous Coward]

Re: Musk is busy combining all his companies

> "too big to fail" and governments will bail him out of any grievous mistakes.

Not "too big to fail" but "US government has relied too much on it as a way to boast to the world to let it fail" - i.e. the US still wants SpaceX despite Musk, it (DJT) no longer give a dam about making Elon happy or about what his other companies do, so the safest thing is to have SpaceX own the lot. But (to hide the he obvious fraud) he still needs a flimsy excuse why SpaceX would want Twitter, so make xAI buy X then come up with a "reason" that SpaceX "needs" xAI.

Next week: SpaceX needs the secret behind the amazingly robust Tesla Truck* so they have to buy that as well.

* a smaller percentage of those have blown up compared to Starship; so far.

[David Hicklin]

Re: Musk is busy combining all his companies

> Not "too big to fail" but "US government has relied too much on it

US company, US problem to bail it out - leave the rest of the world out of it and others are building alternatives to starlink

Yeah getting to space might a an issue but there are again alternatives for non-human launches

[Jellied Eel]

If I was to move to somewhere remote with no broadband access, I'd need two things. Access to a decent pub and satellite...

IPA was originally invented as a beer that could survive transit by sea to the colonies. So we just need Space-IPA and Spac-X to invent sub-orbital or Fractional Orbital Beer deliveries. Some Dutch scientists already invesigated the effects of microgravity on beer onboard NASA's vomit comet, so it's a simple matter of building on that reasearch.

[MachDiamond]

"I'd need two things. Access to a decent pub and satellite..."

What might be keeping those remote areas worth visiting is a lack of connectivity.

[Dagg]

lack of connectivity

So, so true! When I go on holiday I DO NOT what to be contacted especially by WORK and certain family members...

[Anonymous Coward]

I disagree and agree. My goal in life, 10 years ago, was to move to a beach in the south pacific and call it 'work from home' with a solar panel to charge my laptop, but I'm not a very good fisherman.

[MachDiamond]

"and despite the fact I despise Musk and everything he tries to f@$k up, was hugely impressed with the speed."

Given where you were, I expect you weren't sharing with loads of other people. Try the same thing in London.

[Anonymous Coward]

And end to geofencing ?

Is it a bit dim to ask how you will tie end IP addresses to any particular .location ?

Presumably it's "trivial" to route a starlink modem to your network 1,000s of miles away ?

[Kurgan]

Re: And end to geofencing ?

In fact starlink users are experiencing every kind of anomaly with geoip, which is to be expected.

[Anonymous Coward]

Re: And end to geofencing ?

To be fair, geofencing was never part of the technical specifications for TCP/IP. That's something governments invented all by themselves.

[Anonymous Coward]

Re: And end to geofencing ?

Governments?

IP geolocating was dreamt up by salesmen trying to track you and/or change the terms if your subscription, back in the days before Starlink etc when the governments could rely on just controlling what went over the cables near the borders.

That governments would then latch on to the salespeople's entirely mythical "geolocating" and "geofencing" is no surprise, nor is their dismissal of any failures (ring any bells there).

[AndrueC]

Re: And end to geofencing ?

starlink users are experiencing every kind of anomaly with geoip

I have experienced a fair few over the years. Depending on the site I'm either located at my home town in Northants (rare), London (guess that's the default for UK), Hitchin (my ISP's offices) and on one weird occasion somewhere up in the wilds of Scotland - and I do mean the wilds.

[Terje]

While they may function well enough and do provide a valuable service, my concern is that these very large leo constellations are just the latest attempt to 100% speedrun Kessler syndrome and a nicely aimed high energy event from the sun will knock out enough of the satellites that a collision becomes inevitable.

[Charlie Clark]

That, and the risk of privately run monopoly.

[Kurgan]

You mean "ANOTHER" privately run US-based monopoly? Like after MS, Google, and AWS?

[doublelayer]

A counterargument that breaks down as soon as you provide a list of three competitors in your sentence, competitors who don't own the cloud market all to themselves even if we grouped them. Cloud servers are hard to monopolize and there are a lot of providers even though those big three, all of which dislike and try to win over one another, do have more customers. Satellite internet has fewer competitors, especially if we eliminate geosync providers and only consider those that can offer better latency. In most cases, there are no competitors, because OneWeb doesn't serve individuals and nothing else has a constellation established yet.

[MachDiamond]

"Cloud servers are hard to monopolize and there are a lot of providers even though those big three, all of which dislike and try to win over one another, do have more customers."

And when any of those smaller competitors get above a certain size, they're made an offer they can't refuse.

[doublelayer]

Such as? Which bought out Oracle Cloud? How much was IBM or Alibaba paid when AWS swallowed theirs? But then again, maybe those still exist because they're tied to existing big companies. I mean, we couldn't point to any large companies whose only business is cloud, especially not European ones as there's a conspiracy to rob Europe of them as options. Companies like OVH, IONOS, Hetzner, Contabo, all of these were bought just last week by Google, weren't they?

Your theory is wrong. The big clouds don't have the resources to buy all of those. They try to compete by having tons more services than any individual smaller cloud could offer and maintain, and that sometimes works for them, but it doesn't need to affect a buyer of services.

[fishman]

It's a monopoly for now. China, Amazon, and others are slowly deploying theirs.

[Kurgan]

Yes, and since now every rich enough nation wants one (for good reasons, I'd say, you don't want to depend on a private US firm) there will be so many sats up there that the risk will increase a lot. Also, as soon as someone decides to wage "space wars" and actually shoot enemy satellites, the collapse will be inevitable.

[Roland6]

Given what we are seeing with respect to cyber attacks, I suggest actually well before people get around to shooting individual satellites they will simply toss a rouge satellite into the relevant orbit.

https://www.youtube.com/watch?v=XIvHd76EdQ4

[Anonymous Coward]

> toss a rouge satellite into the relevant orbit

Rouge?

Guess so, that applies to China, Russia and Maga.

[MachDiamond]

"I suggest actually well before people get around to shooting individual satellites they will simply toss a rouge satellite into the relevant orbit."

A bottle of vinegar, some solid heating fuel and a trip to the nearest ground-link station is easier and cheaper. Put the vinegar in a pot over the flame and close up the cabinet doors. The vapors will go everywhere and find every exposed bit of copper and then some.

[Jou (Mxyzptlk)]

"nearly always drops packets during that process"

Last part of the article, that is the important information for me.

I hoped or guessed it might be without that, but actually losing pakets (and relying on, for example, the TCP layer to fix that) it is then...

[Roland6]

Re: "nearly always drops packets during that process"

That does seem to imply Starlink is using conventional “after the event” IP routing rather than predictive routing, which perhaps fine for static device-router connections but ill-suited to mobile connections.

[Anonymous Coward]

Re: "nearly always drops packets during that process"

Starlink isn't paying close enough attention to be able to predict which satellite will be the next to pick up a link and pass that information across the satellite-satellite laser link.

[Mike Pellatt]

Re: "nearly always drops packets during that process"

Given that the UDP services I use, which ofc do any necessary retransmission at the app layer, run without missing a beat these days, I'm pretty sure that assertion isn't still the case. POTS VoIP - perfect. WhatsApp - perfect. Zoom - perfect.

2+ years ago was a different story, but Starlink have clearly put massive effort into seamless handover. As well as terninal-base propagation times, reflected in RTT, and max speed.

There were multiple terminal firmware updates in a week in 2024. It's dropped to less than one a week now .

[Prst. V.Jeltz]

I feel this article and/or comment section is missing the el'Regs traditional "IN Spaaaaaaaaaace" motif

or did I imagine that ?

https://youtu.be/-GFgwkLrwPk

[Fruit and Nutcase]

Tin Hat

Given Musk's disdain for regulators and regulations, I'd start wearing a tin hat do prevent the high powered radio waves from his satellites frying my brain

[Anonymous Coward]

Re: Tin Hat

Mark Steele: 5G is a weapons system!

Musk: Look what I can do with 4G!!!*

* Starlink "Direct to Cell" 4G service, turned up to 11, beamforming the full power of every satellite with line of sight straight at your tinfoil hat. Frying tonight!

[DrewPH]

Hobson's choice

I don't like having to use Starlink. I don't like Musk, and I don't like the pricing. I worry about the crowded skies and the long term effects and dangers.

However, what I like is living on a beautiful piece of land in the hills of the southern Philippines with almost no neighbours and where no usable terrestrial internet is available.

So I suck it up. I work around any geoip issues as best I can, and I honestly couldn't care less about the regulators. It's my lifeline, and one that generally works well.

[MachDiamond]

Re: Hobson's choice

"However, what I like is living on a beautiful piece of land in the hills of the southern Philippines"

Have you looked at other options? I see so many media articles that state that Starlink is the first and only satellite internet provider, which they aren't. Maybe in the Philippines they are but, in the US there's also Hughesnet and Viasaat. There's more for commercial applications.

[doublelayer]

Re: Hobson's choice

Compared to LEO satellite like Starlink, geosync satellite like both the services you name has a lot more latency. That comes with benefits like not having packet loss all the time and having four satellites instead of tens of thousands, but it's the classic tragedy of the commons; those satellites are already up there and if I had to buy one of those services, I'd want the lower latency too. Often, I wouldn't mind the increase, but I do things on remote machines and adding 250 ms to each interaction would become annoying when I did. The prices are often similar, so those tend not to win even for those who don't care about the latency.

[Anonymous Coward]

Re: Hobson's choice

Nobody's that wealthy. Do you know what it would cost to watch youtube over HughesNet????? It used to be so expensive that it was more cost effective, and less time consuming to drive 30 miles into the nearest town, park outside of a library, and download files there, and the drive 30 miles home.

[samsungfreud]

Not a fan

My employer is toying with the idea of using starlink as a backup circuit for 5 warehouse locations.

The test network is the local priority subscription with the performance Gen 3 dish.

I can't see this as being a stable platform for commerce.

I've run a workstation directly off the router and through a dream machine pro and a ubiquiti LR AP.

Results aren't consistent.

IPv4 was invoked, sometimes the router would revert back to IPv6.

Numerous vendor / client portals would become unreachable since they weren't running IPv6, including the starlink portal itself!!

At times the speed would drop to around 5 down / 10 up.

Reboots didn't help and tech support always suggested it was a local equipment issue.

I found that leaving the starlink gear turned off for about an hour would clear the issue.

This is just my experience at this point in time.

I'm recommending against this but my boss is a strong fan.

[Oneman2Many]

Re: Not a fan

Been pretty stable whenever i have had need to use it. Don't know if you are in a crowded area or you have LAN issues.

[Marty McFly]

Re: Not a fan

I was a huge skeptic. I kept paying my land based DSL for over two years until I felt I could trust Starlink. The only Starlink outages I have experienced have been system-wide - which means it gets fixed immediately. Meanwhile the legacy DSL would go out and not come back on for a couple days until some tech finally drove out to the country and replaced some piece of kit in the big phone box.

My Starlink is in bridge mode (I NEVER let my ISP own the connection AND my network). Unifi Cloud Gateway Max owns the edge and the Starlink IP address.

The only thing I miss is having a routable IPv4 address. Yeah, I know it is a business product offering, but it is priced beyond viability for a geeky home user. Static IP DSL was only an extra $10/mo.

[Rahbut]

Re: Not a fan

I've been mightily impressed with starlink. It has been quick and reliable, if a little pricey. But seeing as noone can readily give me similar speed/latency, that's the price I will pay.

The lack of IPv4 made me look at IPv6, but even that isn't for everyone. Tailscale, Cloudflare and Mullvad come in handy.

I had more problems with Plusnet thinking I was in America than I've had with Starlink.

I've noticed the recent release of the lower price tiers has dropped the speed at peak times.

[MachDiamond]

Re: Not a fan

"I'm recommending against this but my boss is a strong fan."

Bosses don't read the article, just the headlines so they buy into the hype

[awomanmanhasaname]

Re: Not a fan

Is it bridged?

Speedcast may have solutions for you.

[xyz]

Look....

I run the following...

Starlink in the countryside @40€pcm

Multiple 4G mobile SIMs in the countryside @ between 1€-8€ pcm.

1Gb symmetric fibre in town @20€pcm.

All work, but Starlink is the most stable. My only Starlink issue is power draw as I'm on solar in the countryside, but apart from that after nearly 3 years, I can't fault the service..