SpaceX has achieved much since it launched its Starlink low earth orbit satellite broadband service in 2020.
Commercial service started in February 2021. At the last count, Starlink said it has 1.5 million subscribers1.
The latest stats from APNIC, the Asia-Pacific regional Internet address registry, estimates Starlink has 85k New Zealand users. That’s a shade under 2 per cent of the total market.
Given 13 per cent of New Zealanders are outside the fibre footprint and accepting there will be few Starlink users inside the fibre footprint, that’s about 15 per cent of the rural market.
Starlink huge impact on rural comms
It didn’t take long for Starlink to have an impact on rural telecommunications here. It is popular with rural businesses and with well-off consumers who live beyond the fibre footprint.
Note the word ‘well-off’ in that last sentence. A basic consumer grade Starlink connection costs NZ$160 a month and customers need to buy a dish up-front before they can connect.
The dish costs NZ$730 at the time of writing although it has been discounted in the past. Earlier this year it cost NZ$300.
Either way, Starlink subsidises residential dish hardware at least some of the time.
During Starlink’s early days it cost the company much more to manufacture dishes than they sold for. There were reports in the US media that a dish cost Starlink US$3000.
Today the economies of scale and improved processes will have reduced that, but not to the point where it costs the company less than NZ$300 per unit to get dishes to customers.
Affordable means different things to different people
Starlink’s residential prices put satellite broadband in reach of middle class rural New Zealand families. The monthly subscription is about double what urban consumers pay for a faster, more reliable fibre connection.
Business grade connections are faster, but still slower than everyday fibre broadband speeds. Prices start at NZ$426 a month for a service with a terabyte of data and climb to NZ$2507 for a service with 6 TB. The higher performance business class dish costs $4,567 up front.
Residential users can buy business class dishes. Starlink recommends this in cases where a customer lives in a hot, cold or rainy location or where the terrain is rugged. It gets to see more of the sky, which means a more consistent connection. Meanwhile its larger size means better speeds, although Starlink does not guarantee that.
The normal path of a new technology is that it starts out expensive, then becomes more affordable over time as providers reap economies of scale and competition kicks in.
That could happen to Starlink. It has found ways to trim costs and competing services are preparing to open for business. One rival, OneWeb, is selling enterprise satellite broadband in Australia and New Zealand.
Yet, a closer look at Starlink’s economics suggests the opposite is more likely. It is more likely that Starlink prices will climb.
A satellite network is a complex business with many parts. Yet you can boil down its basic economics to a few simple ideas.
Satellite broadband uses radio spectrum. This means it has to deal with the same laws of physics as other wireless communications technologies.
The key issue is that customers share spectrum.
What this means in practice is there is a hard limit to the amount of available bandwidth in any given area at any given moment. Starlink can choose to serve more customers by giving each customer a slower connection, but the total bandwidth is fixed.
Starlink does this today in regions where the service is congested. At the time of writing, this doesn’t appear to apply in New Zealand. But it could over time.
There is a point where Starlink has to halt signing up new customers in order to make sure existing customers have a worthwhile service. Managing traffic this way is how mobile phone companies deal with fixed wireless broadband customers.
You could argue that with 1.5 million customers worldwide, Starlink is bumping up against capacity constraints today - at least in certain markets.
SpaceX has options to deal with this. It can launch more satellites. More satellites give it more capacity to serve customers.
The other possibility is to upgrade the hardware on each satellite to increase its capacity.
There is a physical limit to how much extra capacity each satellite can handle. We’re not there yet. Remember satellites use radio spectrum.
Radio spectrum frequency bands are regulated by governments. The precise rules vary from country to country. But there is a limit to the bandwidth each satellite can carry.
SpaceX has made this capacity constraint clear in its marketing. It tells potential customers to order early because, in effect, the supply of available connections is running out.
It’s worth pointing out that not everyone gets a decent service from Starlink. There are many happy customers, but there are disappointed ones too.
Starlink has around 5000 satellites in low earth orbit today (reference). It says it plans to have 12k with a possible extension to 42k.
More satellites means more capacity. They also mean more crowded skies. Remember there are rival LEO constellations. Satellites already have to manoeuvre to avoid each other. This will become more complicated as the number of satellites increases.
When mobile phone operators need to upgrade their hardware, they send engineers and trucks to the tower. Once there, the job can be as simple as swapping circuit boards. Either way, the upgrade is trivial compared to upgrading satellites.
A LEO satellite typically stays in orbit for five or six years. Starlink is launching satellites all the time, but, in effect, it takes the company years to upgrade the entire constellation.
A costly business
Back-of-an-envelope calculations suggest each satellite costs Starlink around US$250k. The company has 5000 in orbit. That’s more than a billion. Each lasts five or six years. Present technology means you can’t recycle satellites.
Not every satellite makes it, but let’s forget that for now.
Starlink is owned by SpaceX which is in the business of launching rockets. It gets to the front of the queue and, presumably, the sharpest prices. Even so, at today’s launch prices, the cost of putting a satellite in space, in effect, doubles its price. That’s US$500k per active satellite - perhaps a little more if we allow for wastage.
Two years ago Starlink’s owner Elon Musk went public on what was then still called Twitter saying bankruptcy was a possibility. He has gone on the record saying it will cost between US$20 and 30 billion to launch the entire constellation.
Business is essential
Given overheads, marketing, support and everything else, it looks like Starlink’s business model would be stretched if the company relied on residential broadband customers. There’s a margin, but not a margin that usually satisfies US-based technology billionaires.
The big money is likely to come from other business. The higher price business-grade Starlink service has a hefty monthly subscription for not much more broadband than the residential plan.
Starlink has higher priced roaming plans, plans suitable for vehicle owners, maritime and aviation plans. All these cost considerably more than a residential broadband plan and don’t use proportionately more resources. There’s also an IoT option.
This is where the arrival of OneWeb is interesting.
While Starlink has a direct-to-customer model, OneWeb is pitching its LEO service at the wholesale level. It works with telcos, ISPs and organisations like airlines or cruise boat companies. In the US, OneWeb has a deal with AT&T.
There is a clear focus on government and enterprise customers. In other words, the more lucrative business.
Starlink has its own wholesale business. It has cut satellite deals with One New Zealand while 2degrees resells Starlink’s business grade communications. Yet, the direct-to-consumer model means it competes with its wholesale customers. That can lead to friction.
Not priced for everyone
At the top of this post it says that Starlink is “popular with rural businesses and with well-off consumers who live beyond the fibre footprint”.
The price is too high for many rural New Zealand families.
This is even more the case in poor countries. The US$100 a month subscription, NZ$160 here, is far more than most people in Central America or Africa can afford.
While middle class rural New Zealand customers are pleased they can use Starlink, a lot would prefer the reliability and lower cost of a fibre broadband connection. Having country people pay twice the price their urban cousins pay for an inferior option is far from ideal.
Even if we learn to love Starlink and accept it can solve rural connectivity issues for many New Zealanders, we will still need to find a way to get the rest of the rural population online.
A company tweet sent on May 6, 2023 says Starlink has 1.5+ million customers. ↩︎