Telcos are dead? Will Starlink take over?
stein@opensky
Being an old telco guy, I keep noticing a large number of commentaries these days about the end of telcos. I have even written about it myself e.g. on “What’s left of opportunities for MNOs?” – not too optimistic. One concern relates to Low Earth Orbit satellites and Starlink in particular – so I thought I would add to the hype about Starlink myself as well.
Non-Terrestrial Networks (NTN)
After a few decades not too concerned about Mobile Satellite Services. I wrote this article in 2020 (Will satellite comms happen this time?). The background was this: Having been part of the development of mobile communications since 2G and 3G, I clearly remember the satellite crowd in the early days claiming to be “better” than terrestrial mobile services – or at least competitive. Therefore, mobile satellite services were ”hot” around the time of 2G hitting the market, then it went quiet for a while – and then they were “hot” again around the time of 3G and so on. With 5G (and 6G on the horizon) mobile satellite services seem to be “hot” again. It seems there has been a hype around mobile satellite services around every “G”. At the time of writing that article I concluded that, despite the history, there might be light in the tunnel for LEOs and Mobile Satellite Services.
As there finally seems to be some traction, about a year ago I looked at the topic again and wrote the article “Mobile Satellite Services on the roll)”. In my opinion, there are a few pre-requisites for mobile satellite services to take off and make a true global impact, i.e.:
Avoidance of proprietary and expensive mobile devices
Standards based global solutions
Integration with terrestrial mobile networks
Cooperation between mobile and satellite network operators
Functioning commercial agreements between them
The technical aspects of these prerequisites all seem to be met these days. The main remaining challenges are most likely on the commercial side.
Some physics
This was elaborated in my previous article on “Mobile Satellite Services on the roll)” and I will only repeat some of the highlights:
Coverage: The coverage provided by MSS services using LEO satellites will be great. Satellite beams will cover large rural as well as urban areas. MSS coverage is thus a potential great complement to terrestrial coverage from terrestrial MNOs – specifically in rural areas.
Indoor coverage: Indoor coverage from MSS will be limited or non-existent. The distance from space for LEOs is up to 2000 km depending on LEO constellation – and the signal attenuation from space is excessive (unless you have directive antennas – which do not apply to standard mobile devices). Also, trees, leaves and buildings will attenuate the signal further. Combined with the fact that mobile users spend 80%-90% of their time indoors, MSS will only work in 10%-20% of cases – and even less in urban areas (where most users are) due to building penetration and shadows. MSS might thus only be relevant for 1-5% of the population or so.
Capacity: The capacity of a terrestrial mobile network is fundamentally based on frequency re-use per cell and that cells in urban are extremely small, down to a few hundred meters. Compared with cells from a satellite the cells are several magnitudes smaller, thus meaning that the capacity for MSS is also several magnitudes lower. This could, of course, be compensated for by excessive use of spectrum, which would be considered wasteful and unacceptable.
Spectrum: As commented in my previous article, there is “IMT spectrum” and also “MSS spectrum”, which are globally and regionally identified for MNOs and MSS players (MSOs), respectively – but the regulatory situation in Europe and the US is different. In all cases, however, the most relevant spectrum is 1-4 GHz – so it should be comparable from a propagation perspective – and the fact is that the lower spectrum the longer the range and the better the building penetration. In a terrestrial network, the larger is also the cell size – so lower frequencies are more attractive. Therefore, even the 800 MHz band is on the wish-list for all parties.
Latency: Due to the long distance from the earth, satellite communication will result in latency in signal reception – and for a LEO satellite at e.g. 2000 km distance the latency will be 13.3 ms (and for a Starlink LEO satellite at 480 km distance from earth the latency will be around 3 ms). Long latency may create challenges for e.g. voice communication, but 3 or 13 ms should be acceptable. It should be noted, however, that the design criterion for 5G was to keep the radio latency below 1 ms for time-critical communication. In the western world, it should also be commented that Gen Z youngsters almost do not use voice anymore.
Doppler: For LEO satellite constellations, there will also be a Doppler effect, since the satellites seem to be moving as seen from the earth – which may jeopardize communication. In fact, e.g. Starlink satellites will travel around the earth at 28,000 kmph speeds. This is handled by digital signal processing, however.
To summarize, MSS will have physical constraints, most notably in capacity and indoor coverage – and in urban areas – so the most obvious use case will be in rural areas and in other areas where terrestrial coverage will have challenges, like e.g on ships and in aircraft. There will be challenges in urban areas and specifically indoors.
In the case of Starlink, Starlink as a stand-alone MSS service will therefore not be viable as a full-blown competitor to terrestrial MNOs, except for some peripheral use cases. For global scale and high volumes, integration and collaboration with terrestrial MNOs and networks will be required.
Starlink specifically
These days, my LinkedIn profile is currently being bombarded with news and commentaries about Elon Musk, Starlink and how Starlink is planning to take over the mobile telco world. There are also a few commentaries from other parts of the world, but most of it (almost all) are related to the US. Commentaries to note include very many from e.g. Sebastian Barros, Amanda Newman and many others. These are mostly focused on the US market and what is happening in the USA these days.
First, let it be clear, Starlink is a very good and useful service – and is being used in many countries across the world already. A recent high-profile example is the agreement with Ukrainian operator Kyivstar which is used to support Ukraine for rural D2D coverage for operation of drones in the defensive war against Russia (using “Starshield”, the military version of Starlink). More generally, however, Starlink has agreements with 60 or more MNOs across the world – most notably for MNOs needing complementary rural coverage, e.g. countries like Australia, New Zealand, Canada and Switzerland – but also Japan, Germany, the UK and USA.
I will therefore not comment on whether Starlink is good or bad in itself but will rather make some comments on the larger picture of whether Starlink will take over the mobile telco world and make “cell towers obsolete”.
Competition or complement?
As commented in my article from last year, “all in all, if all pre-requisites are in place, with good cooperation between the camps, be it locally or at industry level, and acceptable business models for all are developed, then satellite integration with terrestrial mobile will happen. For the end customer this will ultimately provide an extended and much better coverage – and for the world, it could improve public safety and connect the unconnected.”
The key word here is business models – and some form of commercial arrangement between MSOs and the MNOs will need to be in place.
Terrestrial MNOs today build their own coverage in their country of operation (in competition with other domestic operators) and they make agreements with operators in other countries for international roaming. In some selected cases, MNOs also make national roaming agreements with competitors in the same country – but this is generally an exception.
Domestically, MNOs more often use MVNO (Mobile Virtual Network Operator) agreements e.g. for “competitors” without a radio access network allowing wholesale access to their radio networks in order to offer services to customers in the same market. Such MVNOs generally have a wholesale agreement with one of the MNOs in the market only – and they pay a lot of money for this wholesale access (as they don’t have to build an expensive radio network on their own – and, after all, they provide competing services). The MVNO will have its own core network.
If e.g. Starlink had one such agreement with a major MNO, Starlink would have its own core network (plus the satellite radio network) – and would use the MVNO agreement to get access also to the MNOs radio access network – thus having the satellite RAN plus one terrestrial RAN in the country of operation, in combination offering more or less ubiquitous coverage in the country. For Starlink, this would be great, however, this is also what MNOs try to achieve through various NTN agreements with MSOs – with NTN coverage as fallback (refer also my article from last year and also various statistics, like e.g. GSA in Q2 2026 referring to 275 existing partnerships, spanning 101 countries and territories – with Starlink having around one third of these).
Then, for Starlink to achieve a global “coverage” without actually covering the whole globe by satellites, they would also need international roaming. One approach to roaming agreements would be for Starlink to make direct roaming agreements with every terrestrial MNO (if possible) – but roaming agreements are a very long and tedious process.
The second and better approach would be via an MVNO arrangement, if they could inherit roaming agreements from the host MNO (which may or may not be included in the deal) – and being a US player, the most likely MNO candidates would be the American MNOs like AT&T, Verizon or T-Mobile.
The large American MNOs are, however, not too enthusiastic about Starlink (see also AT&T, T-Mobile, Verizon Fight FCC Satellite Proposal That Helps SpaceX and T-Mobile wants some space from SpaceX) – so may be not? If not, some form of roaming broker setup might also be as option for Starlink.
A third approach (which might seem to be the actual approach at the moment) is for Starlink to have its own terrestrial network, through acquisition of an existing mobile operator (or cooperation with towercos). This Starlink MNO would still need to have sufficient international roaming and interconnect agreements though. It would still not be certain that the roaming partners would include NTN coverage as part of the roaming agreements.
Technically, it will also be desirable to have automatic handover between IMT and MSS spectrum – but this should be defined in 3GPP standards.
Who “owns” the customer?
The whole purpose of a roaming agreement is for the end customer to only have one business relationship with an operator (not needing a commercial relationship with every single network they are on) – and this is probably an important area of contention, i.e. who “owns” the end customer. Most mobile users today have a subscription with their local MNO, however, with time it could be foreseen that some customers will prefer Starlink as their primary operator – and with a mutual roaming agreement, the customer would only need one subscription for combined coverage (terrestrial and satellite).
On a personal level, I would prefer only to have one subscription with an operator (typically my existing one). However, if you are open to have more subscriptions, even without roaming it would be quite easy with eSIM to have more than one subscription and using the same mobile device with satellite and terrestrial coverage – although not seamlessly. A pre-requisite is that there is a 3GPP- and GSMA-compatible SIM in the mobile devices and that one is associated with Starlink.
What is happening these days?
There are various comments in the media and social media these days about Starlink replacing terrestrial networks – and Elon Musk himself is also quite bullish about it. He recently stated during a SpaceX earnings call that the company "will be building out our terrestrial networks so there's broad-based connectivity anywhere that SpaceX and Starlink are allowed to operate." He added that "satellite launches plus spectrum acquisition will pave the way for Starlink Mobile to become a major mobile carrier in the US."
There are also many examples with reality checks about the Starlink hype (e.g. CCS Insight, Strand Consult and more) – and from various individual commentators.
Some facts include that the FCC in the US has approved up to 15,000 next-generation satellites for Starlink Mobile, “creating a regulatory path for SpaceX to turn Starlink Mobile into a competitor to AT&T, T-Mobile and Verizon”, using also the previously acquired IMT spectrum from EchoStar and Grain Management (i.e. 65 MHz from EchoStar and 14 MHz from Grain in the 800 MHz-band). See also “SpaceX Eyes 15,000 More Satellites for Cellular Starlink, Hints at Carrier Plans”.
Recently, it is also commented that SpaceX / Starlink plans to get access to 80,000 cell towers through working with towercos like e.g. Crown Castle. Bloomberg reports that SpaceX wants to secure “a minimum of 80,000 cell sites across the US, and is already in talks with one major cellular infrastructure provider”. Crown Castle said it estimates SpaceX could provide an “additional catalyst” for its business, which covers 40,000 towers. It should be noted that 80,000 towers more or less correspond to what the major MNOs like AT&T, Verizon and T-Mobile each use in terms of towers.
It seems very clear that Starlink is planning to compete directly with the major US MNOs – and to build a fourth terrestrial MNO tightly integrated with the NTN part. See also “SpaceX wants to be America's fourth phone carrier — tower owners are cashing in”. SpaceX plans to use their own equipment in the towers and hopes to roll out the service in the next few years up to 2028.
Co-existence challenges between cellular and satellite
A classical challenge which is generally handled by local regulators world-wide is interference between different services using frequencies – and this is also an issue between different countries with different frequency allocations. The international radio regulations have thus identified frequency bands for IMT (cellular) or MSS or other types of radio services. There are also different regulated bands for Frequency Division Duplex (FDD) or Time Division Duplex (TDD) – and spectrum allocations are different in regions of the world and also in different countries.
One challenge which has been brought up in the US is co-existence between satellite and cellular use in the TDD band used by 5G (T-Mobile Slams SpaceX Efforts to Secure More Spectrum, Citing 5G Problems).
Without getting deeper into this, it is part of a more general challenge of coexistence between cellular and satellite when frequencies are close or the same. Refer also GSMA’s report from 2025 saying that “spectrum and regulatory complexity are a major hurdle: Effective deployment of NTN depends heavily on spectrum coordination, coexistence management, and regulatory approvals, particularly where MSS and IMT bands overlap.”
The Business case for launching and operating satellites
I have no direct insight in the cost of launching and operating Starlink satellites (i.e. the satellite part) – but I can imagine a few cost elements:
The cost of the satellites themselves
The cost of launching the satellites
The cost of operating and maintaining the satellites
The cost of replacing or replenishing the satellites at end of life
Spectrum and / or license fees
General Opex including personnel, facilities, etc
Other
I asked Claude what would cover the first two elements - and the answer was around 1 - 1.5 MUSD per satellite (obviously with a lot of uncertainty) – and with 15,000 satellites, this would be of the order 15 – 20 billion USD in Capex only.
It is also known that Starlink paid 17 billion USD for spectrum from Echostar and 8 billion from Grain Management – on top of what they previously paid for their initial MSS spectrum. On spectrum fees to FCC, the official statement is that “FCC commercial satellite application fees range from 30,000 to 45,000 USD per major commercial license filing, while large commercial constellation annual regulatory fees scale between 20,000 and 50,000+ USD per year.”
In addition, however, there will be annual O&M costs, personnel costs, cost of facilities etc - and every 10 years or so, each satellite must be replaced or replenished – either ending up as space junk or with a costly repair / replenishment mission. As far as I know, the FSS has required a plan from Starlink on how they will manage this, however.
Without any specific insight on it or getting into Starlink’s business case (for the NTN part), even without the annual O&M costs, personnel costs, cost of facilities etc, with e.g. 100 million customers, I would estimate that Starlink would need a monthly ARPU of at least 10 USD simply to recover costs. With the additional cost elements and profitability requirements, I would estimate that this might be tripled.
Starlink typically charges a monthly service subscription (30–130+ USD per month for residential users), plus a one-time hardware kit cost (199–599 USD) – and for the moment, Starlink’s reported ARPU is 69 USD in Q3 2026. My assumption is that Starlink so far is priced as a premium service, focused on enterprise customers and customers in rural areas needing coverage, so I expect that this ARPU will go significantly down. With time, I would therefore assume that a stand-alone Starlink satellite network would barely be profitable – thus the terrestrial extension Starlink is currently pursuing, hoping that the combined business case will work out.
The terrestrial extension therefore makes commercial sense – provided they will manage to build a competitive terrestrial network in terms of price, quality and time. Having been part of building a few terrestrial mobile networks and having seen business cases for the operations of these, this is not an easy task (see more below).
Practicalities of rolling out Starlink’s terrestrial network
As stated, I have built a few terrestrial networks, and it takes a lot of planning, time and cost. Assuming that Starlink acquires access to a lot of cell towers (80,000 as their ambition is), this will remove a significant hurdle in terms of site acquisition. It should be noted, however, that these towers will need a transport network to them, radio equipment in the towers which transmit and receive signals in the correct directions (i.e. radio planning).
Starlink refers to “building a terrestrial “small cell” network using these towers. “Small cells” may be correct compared to satellite beams, however, the majority of cells in a terrestrial network are in urban areas and are the proper small cells – and these are needed to offer the required capacity in the network – not coverage. In terrestrial terms, towers provide large cells. Site acquisition in urban areas is a whole different ball game and is where a terrestrial MNO actually competes.
Building a terrestrial network is one aspect. It typically takes 3-5 years to build and it never stops. For the MNO to make money, it takes even longer. This is confirmed by Verizon’s CEO, “estimating that a full network build would take at least four-to-six years, with spending of $300 billion to $500 billion over seven or eight years, which he argued could leave SpaceX with broadly the same capabilities and cost structure Verizon has today.”
Competition in the US only or everywhere?
As indicated above, it seems very clear that Starlink is planning to compete directly with the major MNOs in the USA – and to build a fourth terrestrial MNO tightly integrated with the NTN part. Most commentaries referenced above are also focused on the USA (which also seems to be the case for Elon Musk himself).
In other parts of the world, MNOs and regulators have embraced Starlink as a very useful satellite service, providing coverage where it is not viable to do it with terrestrial base stations. The larger competitive issue versus terrestrial MNOs is not as pronounced, however, it can also be noted that Europe Doesn't Want Elon Musk's SpaceX to Hold Its Phone Networks Hostage, where “four of Europe’s biggest telecom companies are joining forces to challenge Elon Musk’s SpaceX as European leaders push to reduce the continent’s reliance on American tech.” Deutsche Telekom, Orange, Vodafone and Telefónica announced that they signed a letter of intent to explore a joint bid for satellite spectrum. Here it should be noted that this is not only a pure competitive measure against Starlink – but also a general drive for European sovereignty, i.e. being less dependent on American tech and the whims of Elin Musk and Donald Trump. Refer also my earlier article on “Sovereignty - what about it?”. See also the statement from Canada here and here.
So, if Elon Musk and Starlink want to “take over” the European mobile market, what can they do? The easiest and lowest cost alternative would be to use roaming between the US hybrid (satellite and terrestrial) operator and the local European operators - provided that the European MNOs are interested. They would probably be interested in extending the coverage for their own customers – but they would probably want to keep “owning” these customers themselves. European MNOs would probably also be interested in allowing existing Starlink customers (terrestrial or satellite) to roam on their networks – increasing traffic, however, they would not like Starlink to fully compete for customers within their country.
Starlink already sells satellite services in Europe directly to consumers, although at a whole different scale than terrestrial mass market volumes. For Starlink to become a real challenger to terrestrial MNOs in the retail market in Europe, they would have to up the volume significantly and make it a price and branding issue. Maybe they could make a strategic alliance with e.g. Apple or Samsung? … but this might be far out?
Without any terrestrial Starlink customers in Europe, there might not be any big challenge for MNOs. For it to be more challenging, Starlink would need to use a local MVNO approach – or to acquire or build a terrestrial MNO – which has a quite different cost and time perspective. There are no signs of this (yet) – but CCS Insight has speculated on whether SpaceX next year would acquire VEON, Starlink’s largest European customer – which has operations in Russia, Ukraine and other ex-Soviet republics. I expect that a similar competitive analysis would equally apply to any region / country in the world.
What is Elon’s real motivation?
Elon Musk has a lot of money, so he has a lot of money to “waste” – or to make big bets and up-front investments that most others cannot do. History has shown in many cases that a high-profile CEO can talk up the share price of his company – and in some cases it results in a “bubble” like e.g. the dot.com bubble in the 90s.
The ambition of ubiquitous mobile coverage in the world is a great one – but rolling out 15,000 satellites and 80,000 towers for terrestrial coverage in the US is a lot of work and takes a long time. If the same terrestrial approach were repeated outside the US, it would take even longer and cost even more money.
Starlink is a daughter company of SpaceX – and with Starlink satellites being launched by the mother company, there is a potential opportunity to do cross-subsidies and to hide the business case for Starlink under its mother brand (thus a lot of uncertainty about my previous comments on the business case).
SpaceX was in June 2026 valued at 1.77 trillion USD – and by the recent quarterly financial report in October, the share price went up 18% - so the current value of the company is huge – but the investments to be made are also huge.
I am in no position to know the real ambitions of Elon Musk, however, a lot of the money to be spent will come from investors. Although SpaceX is doing well at the moment with positive Q3 net results of 2.8 billion USD, the Capex to be spent is significantly higher and to be increased to 35-40 billion going forward.
The question at the end here will be, considering that the rollout of the satellite and terrestrial networks will take time, whether or not investors will have the patience to wait for the returns – or if there will be another “bubble”? At the moment, Elon Musk is in a good political and regulatory position in the US, closely linked to Donald Trump, but this might change – and the situation is not the same elsewhere in the world. In the meantime, Elon Musk could cash out though.
Summary
To start, I would like to repeat that Starlink is a very good and useful service – and is being rolled out in many countries across the world already – and they already have 10,000 – 11,000 satellites in orbit today. This article has mostly been about how and if it will “take over the world” from Mobile Network Operators regionally or globally.
If I can summarize a little, it seems very clear that Elon Musk has a stated ambition to compete strongly against (at least) American MNOs, “providing a fourth mobile operator with ubiquitous coverage in the US”. The questions are really about the realities of this. Incumbent American MNOs have all questioned practicalities and realities of this – even T-Mobile, one of Starlink’s initial strategic partners.
Although Starlink is planning an extensive rollout of 15,000 satellites or so, any mobile satellite service will obviously have physical constraints in terms of capacity and indoor coverage, which clearly will have challenges in urban areas (where most of the MNOs’ customers and traffic are). Even with Starlink complementing its satellite network with 80,000 towers, these are mostly in rural areas – so, although it will help with national geographical coverage, the urban and capacity challenge is still there. To build a competitive terrestrial MNO, it will take 4-6 years – and it never stops.
As I commented already a year ago, there are (finally) a few pre-requisites for mobile satellite services to take off and make a true global impact, i.e.:
Avoidance of proprietary and expensive mobile devices
Standards based global solutions
Integration with terrestrial mobile networks
Cooperation between mobile and satellite network operators
Functioning commercial agreements between them
The technical aspects of these prerequisites all seem to be met these days. The main remaining challenges are on the commercial side – and the area is a big battlefield these days. Not only is Starlink (the biggest MSO at the moment) challenging the American MNOs, but there is also a tech war within the US with also other MSOs challenging Starlink (seen e.g. here: US clears $11.5B Amazon-Globalstar deal - Mobile World Live) – and many MNOs are also partnering with other MSOs than Starlink (ref. GSA on MNO-MSO partnerships – with Starlink having around only one third of these). In Europe, several of the largest MNOs have also created a Joint Venture bidding of MSS spectrum (Europe Doesn't Want Elon Musk's SpaceX to Hold Its Phone Networks Hostage). See also the following statement from Canada here and here.
All in all, Starlink could provide a very useful service complementing the terrestrial mobile services across the world, which could be well appreciated by consumers for extended and / or ubiquitous coverage. To get to this goal, good cooperation between MNOs and Starlink is required – and with acceptable commercial agreements between them. This could be via roaming or one or several MVNO agreements, however, with head-on competition for consumers and customer ownership, the ambitions will be severely hampered.
Now, such head-on competition is happening in the US, but not in other countries. I could imagine that Elon Musk has enough to do in the US for the moment (“America first”) and may be that might be enough forever as well. Building proper small cells for capacity in urban areas is a challenge I haven’t seen much of in the media – but it will be there!
In other parts of the world (e.g. Europe), the regulatory situation is different (many different countries - and also more restrictive) – so the challenge might be bigger. In my mind, the better approach would be some form of international roaming arrangement. Starlink could still sell satellite services in Europe directly to consumers – without competing head-on with MNOs. Starlink could still decide to make a price and branding war in Europe though – but this is what is already happening between MNOs.
Finally, I don’t need to get into this communication battle about Starlink, but I will keep on watching the situation – and it will be interesting to see how it plays out. Will Starlink really take over the mobile telco world – or will it become another bubble with a lot of space junk flying around the globe?
Comments