Electricity Costs of Starlink’s UK LEO Broadband Satellite Service
The end-user electricity costs of consumer broadband ISP connectivity isn’t something that we’ve touch on much because it’s usually fairly insignificant. But those planning to adopt Starlink‘s (SpaceX) new service may need to pay a little more attention.
A few years ago we examined the electricity costs of home broadband routers (here) and found that most devices only used a handful of Watts, which meant they generally wouldn’t cost more than around £15 per year to run (assuming 24/7 usage and multiple connected devices etc.). At some point we may need to update this for FTTP ISPs, which mix both a router and Optical Network Terminal (ONT), but the impact will be small.
One other area that we haven’t really looked at before is the power consumption of Satellite based broadband platforms, particularly with respect to Starlink’s new mega constellation of Low Earth Orbit (LEO) based ultrafast broadband platforms. Just to be clear here, we’re only interested in the end-user’s side (i.e. router, dish etc.).
Advertisement
Thankfully the Director of full fibre ISP Broadband for Rural Kent (B4RK), Tim Higgs, has been doing some testing on this, and it turns out that running Starlink might well be something that you’ll notice on your electricity bills. Admittedly, Tim has a bit of a vested interest in pushing FTTP as the better long-term solution, but his figures are similar to others that we’ve seen online.
Starlink’s Electricity Consumption
At present SpaceX has already launched c.1,300 LEOs into orbit (space) around the earth and their initial ambition is to deploy a total of 4,425 by 2024, which could potentially be followed by up to 12,000 at a later date (possibly late 2026). The service has already gone live in the USA, Canada, UK and is now extending into Europe, albeit mostly still in the beta and early pre-order phase.
Customers typically pay £89 per month for the service itself, plus £54 for shipping and £439 for the kit (dish, router etc.). For that you can expect unlimited usage, good latency times to 20-40ms, downloads of between 50-150Mbps (speeds will rise as the network grows) and uploads of c.20Mbps. During the beta phase there’s still the potential for some connection drops.
However, in terms of power consumption, Tim says Starlink’s system tends to use about 89 Watts when idle and then bumps up to around 110w during normal usage (average of c.100w). But he did also see the occasional peak, reaching at most 175.7w (the Power Supply Unit is rated for 180w). “The motors on the dish seem to only be used during initial alignment, then it’s all beam forming,” said Tim.
Advertisement
The below reflects Tim’s rough workings and his comparison between technologies, which he kindly shared with ISPreview.co.uk (note: his setup cost should actually be £493).
| Starlink | FWA | 4G | 5G | FTTP | |
| Speed (down/up) | 150/50Mbps | 60/30Mbps | 90/40Mbps | 300/50Mbps | 300/100Mbps |
| Set-up | £489.00 | £300.00 | £420.00 | £600.00 | FREE* |
| Monthly | £89.00 | £30.00 | £35.00 | £35.00 | £40.00 |
| Yearly | £1,068.00 | £360.00 | £420.00 | £420.00 | £480.00 |
| Watts (average) | 100.0 | 9.7 | 7.0 | 10.8 | 9.5 |
| Annual cost of electricity | £162.41 | £15.75 | £11.37 | £15.43 | £17.54 |
| 1st year cost | £1,719.41 | £675.75 | £851.37 | £1,035.43 | £497.54 |
| following years cost | £1,330.41 | £385.45 | £438.37 | £446.23 | £507.04 |
* When DCMS gigabit voucher funding pays for the infrastructure
** Calculated using the UK Average rate of 18.54p/kWh (December 2019 tariff)
Obviously, there will be some caveats to consider above. For example, the use of different routers on the other networks can increase or decrease power consumption. Likewise, some people may be paying less than the old UK average rate of 18.54p per kWh (our own tariff is 13.65p per kWh) and there are also those with tariffs that vary by time of day (e.g. cheaper at night). We won’t touch on the standing charge as that’s less relevant.
Nevertheless, a system that is consistently gobbling up c.100watts is something you will notice on your bill. On the other hand, anybody willing to spend £89 per month on such a system probably won’t be too bothered about that, and you can always turn it off when you go to sleep.
Advertisement
At the end of the day the energy costs involved probably won’t be considered a major obstacle by those who really need such a connection, but it is still something to be mindful of, particularly given Starlink’s plans to introduce a budget tariff for those on lower incomes in the future (here). If you’re on a lower income then any extra costs are more of a concern.
Mark is a professional technology writer, IT consultant and computer engineer from Dorset (England), he also founded ISPreview in 1999 and enjoys analysing the latest telecoms and broadband developments. Find me on X (Twitter), Mastodon, Facebook and Linkedin.
« SIM Swap Causes Annoyance for UK ASDA Mobile Customers
Latest UK ISP News
- FTTP (5849)
- BT (3605)
- Politics (2656)
- Business (2372)
- Openreach (2372)
- Building Digital UK (2302)
- Mobile Broadband (2088)
- FTTC (2070)
- Statistics (1856)
- 4G (1763)
- Virgin Media (1710)
- Ofcom Regulation (1537)
- Fibre Optic (1444)
- Wireless Internet (1436)
- FTTH (1383)
With those types of power requirements, it kind of goes in the face of what Elon is selling his starlink system for. I was under the impression it was designed at least initially to provide free access to remote locations with low incomes across Africa?
Saying that. looking at the cost of this thing currently I’m almost certainly covering both those categories………
I can’t see how this thing can ever be cost effective looking at its base power consumption, especially when its over ten times that of a regular router.
If the other options aren’t available then you’ve not much choice, go without or pony up.
For a lot of the places it will be used a solar panel & battery will power it nicely.
Why would it of been built for Africa when its initial target was the US.
It’s not meant to be cost affective, in North America the monthly cost is not much more than the cost of Internet services available anyway and its not even meant to be a competitive product.
Personally I’m happier to have a broadband service provided via space than some clapped out taped together networks like we do in the UK.
If the UK operators actually did a from nothing to working service roll-out I can bet the pricing would be up there with these guys,
But no we’ve spent 15+ years staggering from pillar to post adding bits that are already out of date here and there, and now that some of the drunks have made it home everybody has forgotten about the shambles of a journey that many still gotta do and not considering the issues still to come from ageing infrastructure.
89W idle power has to be prototype hardware, surely? Mobile phones are well below 1W in idle mode.
Solar panels in UK vary from 2% to 20% capacity factor, so am guessing you’d need a 1-2kw panel with battery storage to power this, probably with winter backup. Is that 5-10m squared of panels? Or is my maths wonky?
This new article is BS, can we have some comparisons from existing satellite broadband kit?
Trying to compare Apples to Oranges,
Not really. The end product is the same, this is a comparison of the delivery methods.
That’s true :S
The starlink terminal has heaters in it to melt ice and snow. It also has a radio transceiver in it that has to have enough power to talk to satellites miles away. Of course it’s going to use a bit of juice to do that, I’m not really sure how you can compare it with your average piece of junk ISP router. Still £160 a year (I calculated it much lower, but then again im not paying 18p per kWh) isn’t exactly nothing either.
As for one of the comments above about giving internet to all and in poor areas etc etc, a 12v battery, 200W solar panel should do the trick. I mean if we’re going to complain about 100W of power, what are they planning to use with their starlink terminal? An abacus? A paper chromebook?
Well said, I wonder with what wattage the transceiver emits RF.
The comparison is valid – because both impact the annual electricity cost of broadband consumers. A typical chrome book consumes about 45w, so at face value this triples someone in power poverty’s energy needs.
Most rural African deployments however share a connection in a community, so that 100w needs to be divided by the number of people it’s shared with.
I agree that this is fairly easily met by solar end storage, but the initial comparison is entirely valid to make. A community may need to upgrade their energy supply on top of the setup costs for starlink.
125us
Why are we pretending that people in Africa are going to be using this any time soon?
They’re not. It’s not a valid comparison. Because you say it is, doesn’t make it true either.
It’s overpriced, expensive, not really valid for developing nations. Not for a long time. And your little story of sharing 100W between villages is laughable. Oh no, people in Africa can’t manage to make 100W of energy. Right. When they have solar/wind and plenty of countries and companies helping them to generate power.
The bleeding heart stuff is a joke
If they can afford a starlink terminal, they can afford a solar panel and a battery which would be cheaper than the terminal. I would imagine if they are installing this stuff in communities then it will be sponsored anyway.
I find it a bit strange that people are arguing about the power requirements when the terminal costs nearly £500 and £90 a month
Apparently it doesn’t have a heater. It produces enough heat during regular operation that it doesn’t really need one.
Since consumer use of this antenna tech is still quite new, I suspect there’ll be reductions in power usage over time which will help with running it from solar/battery (at which point it might need a dedicated heater… but perhaps not in Africa).
Alex you’re right, there’s no dedicated heating element.
There are however tons of resistors all over the board that don’t appear to have any valid function and one could assume perhaps they are there to provide heat? But I think your hypothesis that normal operating temperatures for the ICs on the board (there are lots of them) would probably provide sufficient heating in and of itself.
Landauer’s principle tells us that pretty much any computation turns into heat. They probably don’t need a dedicated heater when there are so many ICs to do it. But the additional resistors that do not apparently serve any purpose are puzzling to say the least.
Suffice it to say, nobody has proven it either way at the moment.
I also found this while researching, from a supposed starlink employee:
Though, the Starlink dish “does have self-heating capabilities to deal with a variety of weather conditions. In fact, we’ll be deploying a software update in a few weeks to upgrade our snow melting ability with continued improvements planned for the months ahead
@Sceptic.
I sponsored a family in Nigeria for a number of years and amongst the things they bought with the sponsorship was a small solar and storage system enough to provide lighting for their home in the evening. Amongst other things were shoes for their son to go to school, repairs to their roof and some livestock so that they could be more self sufficient.
It wasn’t me who raised using starlink in Africa, I was responding to someone who did, claiming that if they couldn’t power starlink they couldn’t power devices. That’s clearly not true. The power requirements of a chrome book or older devices such as an OLPC are below 50w.
I have no idea what you mean by your ‘bleeding heart’ comment.
125us
I also have no idea what:
“Most rural African deployments however share a connection in a community, so that 100w needs to be divided by the number of people it’s shared with.”
means.
I really don’t think that a community sharing a 100W power consumption starlink terminal is going to be a problem when they need your help to keep the lights on. I think they’ve got bigger problems. I imagine that if they were to get a starlink terminal they are going to get funded with the power system required to run it as well. People who can’t keep the lights on aren’t going to be too bothered about having starlink internet terminals are they? I don’t think your comment about diving up 100W worth of power (pretty insignificant) is going to be a problem. It seems you and others are worried about the consumption of a fairly insignificant amount of power. In the western world, that’s less than your refrigerators use, but with starlink that’s somehow bad and has to be shared amongst people who don’t have power or shoes.
It’s a non argument. If they get a starlink terminal, in conditions as you describe, then they probably won’t be paying for it, or the power required to run it. They will also most likely need a BTS (mobile network) to give them access to it since most don’t have access to laptops etc and mobile use is the predominant form of internet access in poorer African nations.
Starlink isn’t going to be giving people living in huts internet access, and even if it does, it’s not going to be a power problem that stops them. Bleeding heart means, look at the power consumption that has to apparently be shared between lots and lots of people, when it’s most likely going to be provided for free, for a service they probably couldn’t care less about
JP is harsh, the article is good. It’s a good point, whether you live in the UK or Africa. I depend on an always-on connection and I have slow FTTC. I’ve ordered Starlink and when it comes I’ll be leaving it on continuously and that cost is not negligible. People like me who need the speed and can’t get it via ancient terrestrial infrastructure will pay the premium. I have studied the orbits and the polar regions are still not covered though, so you’re still stuffed if you’re Santa Claus and live up in Lapland or Antarctica. Of course initially it will always be a premium service as the new LEO infrastructure costs beeellions. But in 20 years it’ll be commonplace and there will be fiber everywhere so competition will drive the price down. Eventually if you live in Africa you will get a good connection one way or another. But if you’re still poor then 100w still isn’t trivial as it’ll take a few big solar panels and a few good batteries to keep it running day and night. Perhaps in 20 years technology will also drive down the power consumption.
Horses for courses. I have FTTC thru Community Fibre and my Adtran ONT pulls 6 watts which implies annual electricity costs of £6.82. Of course I then have the router as well.
I suspect over time this issue will be addressed with more efficient hardware/software as Starlink looks to move into poorer areas.
Actually this is pretty relevant.
Bear with me.
Our house has an average power consumption of just under 800W continuously.
Our vampire load consumption, from having things idling, is 450W.
The majority of our electricity bill is from the vampire load.
So adding a constant vampire load of 85W to consumption is actually pretty significant. Whilst I appreciate that it can be powered from a solar panel and a Li battery 24/7 and I do have quite a bit bigger solar panels than that – it will still add unpleasantly to people’s electricity usage.
It’s called “Parasitic load” btw.
If you can shell out £500 knicker for a starlink terminal, i think you can afford the 100W to run it constantly.
Some call it parasitic some call it vampire – I have seen both used.
The issue is the cost/CO2 footprint over the lifetime.
I tend to look at power wastage over a 10 year period to see it if it worth replacing an item.
Things like halogen -> LED lightbulbs pay back in about 1yr so well worth doing.
Certain white goods the payback is greater than 10 years so no point until they break.
I appreciate that some people are so desperate for a decent or even semi decent connection that they won’t care. But these things do need to be called out; particularly if you have thousands of them around all using too much juice.
IMO it is likely that future versions of Starlink will have a power saving mode to kick in and out according to traffic volumes.
Could you link to the article/source writing so we can see it?
Sorry to say but this article is complete BS. If 100 watt power consumption is such a big deal then you better give up most of your electronics kit and don’t even think about using an electric cooker, kettle or immersion heater. My PC alone can pull up to 850 watts and my laptop 300 watts.
As usual the media will attack Elon’s ventures at every opportunity simply due to fear.
I don’t run my electric cooker, kettle, immersion heater, PC or laptop 24/7. I do run my router 24/7. I think I’m fairly typical.
You’ve not really thought this through, have you?
I don’t see the point in comparing this product. If you want this you clearly live in a rural area without or with very very! slow connection. So you are clearly going to need a service like this. If you shell out for this over your regular bb service then it would be considered a luxury and you can afford the extra power its clearly going to use to communicate and reposition itself with moving sats???. There are easily available dual link routers or services available today to merge bb with mobile signals for cheaper and negligible usage. I can plug my phone into my router and have instant coupling if I needed to. Also if your parasitic (powered off) load is 500w you seriously need to look at your power consumption usage. I have hue bulbs everywhere google speakers, fish tank, multiple appliances and a gaming pc all left on standby and my base load is 145 – 160w, Im all electric in a two bed house.
Well out house is a bit bigger and Victorian with some things fitted 15+ years ago that are far from efficient in today’s world.
For instance bathroom fans – just changed them from ones that used 25W to ones that use 4.5W.
Do I Chuck all my 1st gen Sonos in the skip because it uses too much when plugged in? Yes, I could take that approach and get a really low number but also a headache sized bill so I prefer to approach it incrementally.
I also believe that it is necessary to think of the carbon footprint of the item and it’s remaining lifespan before replacing for the sake of replacing.
We have various other bits of kit such as backup servers which are about 1/3 of the vampire load in standby.
Yes, I am all up
what “all up” means?
Sorry English is not my first language
Ahh…disruption in an industry is fun to watch.
I’m near Cambridge and I have Starlink. I’m consistently getting speeds between 250-400 Mbps and 20-40 Mbps Upload
The dish using 100W average (approx 72KWH/month) is not the end of the world but consider the following. Before we purchased our Tesla, the power usage for the entire house for a month was normally under 300KWH; likewise, our Tesla can travel 300 miles/483 KM on the 72KWH the dish uses in a month.
According to reports the Starlink aerial energy consumption rises to 100W when it is cold to try to melt the snow & ice that forms on the dish and preventing loss of signal. All this is outside of course so unlike other parasitic loads does not help to heat the house in winter.