Satellite broadband, sometimes referred to as “satellite internet” or “space internet” is a frontier technology that’s worth researching when looking for deep-tech investments. Elon Musk, CEO of SpaceX recently estimated the market potential for Internet connectivity to be as much as $1 trillion. This is set against the global mobile wireless services market of $1 trillion (source: GSMA market data, Sept. 2020) and the global satellite internet market valued at $2.93 billion in 2020, which is projected to reach $18.59 billion by 2030 and is expected to grow at a CAGR of 20.4% from 2021 to 2030 (source: Allied Market Research, July 2021: Global Opportunity Analysis and Industry Forecast, 2020–2030).
Around 40.5% of people on earth don’t have access to an internet connection according to Statista (January 2021), which demonstrates the sheer size of the potential addressable market. Only 10% to 15% of the air surface is currently covered by cellular networks according to Quilty analytics, Sept. 2021.
Satellite broadband offers a range of different opportunities for innovation. By decreasing latency and potentially decreasing capital expenditure (capex) costs for cellular operators / Internet Service Providers (ISPs) in certain areas, it can make internet faster, more widely available and more accessible. Today, 5 billion mobile phones move in and out of coverage and 51% of the global population are without mobile internet (source: GSMA market data, Sept. 2020). Improved connection and coverage can also pave the way for new technologies, like Electric Vehicle (EV) and drone-based innovations which are areas of focus for PSION.
That being said, there are many challenges and hurdles that this industry still has to overcome. What satellite broadband is and what some of the biggest challenges and opportunities are for the companies in the space and the people/funds that invest in them, is what we’re going to cover in this blog post.
This article covers some of the basics of satellite internet technologies, but there’s a lot more you should know to make an informed investment decision. Here at PSION Partners, we have over 20 years of experience in deep-tech investing, both as an investment fund and as advisors to investors, large funds and family offices.
What is satellite broadband?
We’re sure that most of you reading this are already familiar with the concept of satellite broadband. However, in case you’re not, satellite broadband is an internet connection that’s provided by communication satellites in space, as opposed to land-based DSL or cable connections.
Satellite internet traditionally works using radio waves and can therefore provide a large area of coverage per satellite. However, the speed of these connections has traditionally been slower than land-based connections. Today, new market participants are attempting new, faster methods, in a bid to improve internet speeds and accessibility.
SpaceX’s Starlink, for example, is experimenting with lasers instead of radio waves, in a bid to improve the speed of the communication between satellites and back down to earth. They are also planning to work with a larger number of satellites (see our article “Transformative communications is surging with remarkable results” for more details) at a lower altitude (Low Earth Orbit - LEO – see our article on Small Satellites and Microlaunchers for more on this), shortening the distance and therefore the time it takes for information to travel to-and-from the satellite.
This video provides a good explanation of the basics of satellite internet and how it works traditionally. Here is a video for an upcoming SpaceMobile service that will revolutionise the more traditional approach, removing the need for antennas on earth and communicating directly with the phone in your pocket.
There are two main types of internet satellites addressing the space internet market, namely High Throughput Satellites (HTS) and Low Earth Orbit (LEO) satellites.
HTS vs LEO
High Throughput Satellites, also called HTSs, are the type of satellite that’s typically used for space internet today. These satellites provide significantly more throughput than the FSS (Fixed-Satellite Service) satellites that were previously used. HTS capacity can be as much as 2 to 20 times larger than a classic FSS. The net effect is a significant drop in the cost per Gigabit per second (Gbps).
HTSs use Ka- and Ku-bands just like most other communication satellites, and are nearly always in Geostationary or Near- Geostationary Orbit (GEO/NGEO). GEO is just over 35,000 kilometres above the earth (35,786 to be exact), whereas NGEO is usually “only” around 20,000 kilometres above the earth. Examples of these satellites are those of ViaSat, Telesat and Intelsat.
Low Earth Orbit (LEO) satellites are somewhere between 160 to 2,000 kilometres above the earth. LEO satellites cost less energy to place into orbit due to the lower altitude. However, they are more prone to orbital decay, meaning that they generally have a shorter lifespan (typically 5-7 years) or need regular re-boosting (requiring limited fuel) to ensure they stay in orbit.
Due to how high they are above the earth, GEO satellites can serve a large surface area. This means that fewer satellites can cover a large area, whereas, with LEO satellites, it can take many times more satellites to cover the same surface area.
Latency differences between HTS and LEO satellites
Since GEO satellites orbit at altitudes that are typically 30 to 70 times higher than LEO satellites, they have a much higher latency since the radio waves have to travel further.
Latency is the time delay over a communication link, meaning the time it takes for information to travel from point A to point B. It is primarily determined by the distance data must travel between a user and the server, however, other aspects, including the medium used to send the data, can also impact latency speed. Radio waves travel faster in a vacuum, more often than not offsetting the extra distance travelled compared to land-based connections.
Current speeds vs. bandwidth
GEO satellite internet currently takes around 638ms for a round trip from London to New York. The current fastest terrestrial connection on the planet is the privately-owned GTT Express (formerly Hibernian express) cable at 58.95ms – see Figure 1 below.
The Atlantic Cable 1 (AC1), a privately owned terrestrial cable, was the previous fastest connection at just 65 milliseconds. The Hibernian Express project cost an estimated $300m investment and only shaved 5ms off the fastest connection between London and New York before it.
Starlink could likely achieve latency speeds of between 20-40 milliseconds, which, when compared to long-established 76 milliseconds speeds of transatlantic cable/fibre, offers a 77% improvement, making it very attractive for specialist financial services clients.
The vast majority of intercontinental global Internet traffic—upwards of 95 percent (Source: U.N. Env’t Programme World Conserv’n Monitoring Ctr., Submarine Cables and the Oceans: Connecting the World, at 3, UNEP-WCMC Biodiversity Series No. 31 (2009)) —travels through undersea cables that run across the ocean floor. This interactive map gives you an impression of the scale of these submarine cables. This illustration by TeleGeography (2021) depicts 464 cables and 1,245 landing stations; 428 cables are active and 36 are planned. There are over 1.3 million kilometres of submarine cables spanning the globe, which would wrap around Earth more than 30 times end-to-end.
The key change is that having accounted for less than 10% prior to 2012, content providers’ share of total capacity surged to 66% in 2020. Unlike previous booms, content providers like Amazon, Google, Facebook, and Microsoft are taking a more active role in the submarine cable market. However, Vodafone is also muscling into this market as it too has realized the significant shift that has happened recently.
Figure 1 (below) illustrates this shift. The Grace Hopper cable system, owned exclusively by Google, landed in Cornwall & Bilboa, Spain in Sept. 2021, and will be fully operational in 2022. It represents a new generation of trans-Atlantic cables coming to U.K. shores with 16 fibre-pair cables and a significant increase in bandwidth capacity to 352 Terabits per second (Tbps) – a huge step up from earlier cables installed by telecoms players. The Amitié cable achieves a similar bandwidth capacity at 320 Tbps.
Here lies the key to understanding what impact the new small satellite constellations like Starlink, OneWeb, Telesat and Kuiper (Amazon) may have on how internet traffic travels in the near future.
Scientists at Massachusetts Institute of Technology’s (MIT) Aeronautics and Astronautics Department recently updated their system’s performance analysis on high-throughput satellites from SpaceX (Starlink), Amazon (Kuiper), Telesat and OneWeb – see “An Updated Comparison of Four Low Earth Orbit Satellite Constellation Systems to Provide Global Broadband”, January 2021.
The scientists at MIT were able to estimate satellite throughput as seen in Figure 2. Telesat manages to achieve similar throughput to Starlink despite having less than half the satellites, thanks to the dual gateway connection system. Using an Inter-Satellite Link (ISL) of 20 Gbps, currently being tested at Starlink for their 2nd Generation satellites that are soon to be launched, improves the throughput and satellite utilisation and cuts down the need for ground stations. Figure 3 below shows the impact on throughput, with or without a 20 Gbps ISL connection.
All satellite system architectures benefit from using Inter-Satellite Links (ISLs) of 20 Gbps. It is also important to notice that, although OneWeb and Kuiper achieve higher throughput than Telesat and Starlink, they also require a significantly larger ground segment with approximately 50% more gateway antennas.
However, when putting these throughput figures in context with the actual global bandwidth capacity of 2,620 Tbps in 2019, planned to grow to 5,656 Tbps by 2022 (see Figure 4), it becomes evident that even with Starlink having 42,000 satellites in LEO on it’s own it would add up to 897 Tbps, i.e. only 16% of predicted global bandwidth requirements in 2022. When using the maximum data-rate figures from the MIT study including Telesat, OneWeb and Kuiper, it adds up to only 584 Tbps i.e. only 10% of predicted global bandwidth according to statista (May 2021) and according to the latest SpaceX launch data (September 2021)..
However, there is still hope for LEO satellites being able to bridge the global digital divide by decade’s end with the recently patented Dynamic Polarization Spatial Multiplexing and Beamforming (“Dynamic Polarization”, or “DPSMXBF”) technology (see this article “Tech Breakthrough Morphs Gigabit WiFi into Terabit Satellite Internet” by Carlos Rios, 26th July 2021). It suggests that using reformulated radio MIMO technology, Starlink Generation 2 satellites could be upgraded to 1.8 Tbps each (up over 90 fold from currently a maximum throughput of 0.0197 Tbps, see Figure 2 above). Multiple-Input and Multiple-Output (MIMO) is a method for multiplying the capacity of a radio link using multiple transmission and receiving antennas to exploit multipath propagation.
In summary, it’s not only about fast latency speeds which Starlink purports to achieve of 20-40ms. It is also about having sufficient bandwidth capacity to address the mass market. From the estimations presented in the MIT study, all four mega-constellations could offer a total capacity of ca. 584 Tbps i.e. only 10% of world bandwidth requirements in 2022.
With these currently envisaged magnitudes of data rates, they would not be able to compete with the current terrestrial networks, which planned to move around 5,000 Tbps in 2020 (Source: “Submarine Telecoms Industry report”, issue 8, 2019/2020, STF analytics), but could complement the coverage of land infrastructure in regions where a cable connection is ineffective, inappropriate, or just unfeasible (e.g., rural areas, isolated coastal and insular regions, and aerial and maritime mobile users).
It’s also reasonable to expect that LEO satellites such as Starlink will struggle to compete against HTS GEOs on a cost per Gbps basis. However, industry experts believe the future of Satcom will be a hybrid of GEO and non-GEO satellites, with LEO satellites providing mobile backhaul, high-speed consumer broadband and enterprise Very-Small Aperture (VSAT) connections.
SpaceX’s Starlink mega-constellation
SpaceX’s Starlink is one of the main innovators when it comes to LEO satellite internet connectivity. They are planning to launch 42,000 Starlink satellites in the coming years and have already launched over 1,786 to date according to statista (May 2021) and according to the latest SpaceX launch data.
In November 2018, Starlink was granted FCC approval for 7,518 satellites in addition to 4,425 satellites previously authorized. The first batch of Starlink satellites was launched on May 24th 2019 and are now in Low Earth Orbit (LEO) in 5 shells at altitudes between 540km-570km, offering the optimal balance between latency and coverage. On the 15th of October 2019, the United States Federal Communications Commission (FCC) submitted filings to the International Telecommunication Union (ITU) on SpaceX's behalf to arrange spectrum for 30,000 additional Starlink satellites (Gen2) to supplement the 12,000 Starlink satellites already approved by the FCC up to that point. Starlink Gen2 satellites would be deployed at nine altitudes, ranging from 340 kilometres to 614 kilometres.
SpaceX’s Falcon 9 rocket can launch 60 Starlink satellites at a time, but Starship (1st orbital test flight recently estimated in Q4 2021) would be able to “take 400 satellites at a time,” SpaceX president Gwynne Shotwell said in 2019. With this additional launch capability, SpaceX’s 42,000 figure sounds more achievable.
In September 2020, financial analyst firm Cowen & co. has questioned Starlink’s capacity constraints once 12,000 satellites are in LEO, saying Starlink will be unable to service “any more than 485,000 simultaneous data streams at speeds of 100 Mbps in the US.” Additionally, initial customer feedback suggests binary on/off connectivity reliability issues that require resolving before expanding their network further.
Doing the math, this puts in question Musk’s targeted annual revenue figure of $30 billion. Cowen & Co. thinks Starlink will in fact be able to support only 8.28 million customers worldwide (oversubscription of network by a factor of three). This would allow its maximum potential annual revenue to be $9.9 billion based only on residential customer pricing. You would need to add additional revenue from corporate clients out of those 8.28 million customers globally.
The key point Cowen & Co. highlights, however, is that Starlink has clear limitations and therefore cannot solve all of the connectivity issues that exist, which is reinforced by Psion’s analysis of bandwidth capacity issues in the section above – see Figure 2 to 5. In fairness, SpaceX’s owner Elon Musk has already said that “Starlink is not a major threat to telecommunications companies.”
Musk, in June 2021, mentioned his Starlink venture was growing quickly and forecast total investment costs in the business at between $20 billion to $30 billion.
Mass adoption and viability
Apart from the bandwidth issue detailed above, the critical issue for the mass adoption of Starlink / OneWeb is the production of radically cheaper ground station antennas. OneWeb has not been able to achieve it. For their business model to be economically viable, they need to reduce the average cost of ground stations to below $100. When you consider that the current production costs for such a device is well over $1,000, this is a major challenge for this technology.
Starlink will need to produce dishes at around $200 in order to be commercially viable. However, their cost is currently around $1,300 (April 2021, down from $1,500 and $3,000). The current $499 customer price of a Starlink dish, while expensive, is actually being sold at a significant loss. During the Space Symposium event in August 2021, SpaceX President Gwynne Shotwell said: “We’ve made tremendous progress on the user terminal, but they are still expensive. But again, I think they’ll be about a quarter of the cost to us right now in maybe a year.” Whether the new Starlink dishes will still be sold at a loss remains unclear. According to Shotwell, the next version will end up costing closer to $250, and eventually just $125.
Millions of ground stations would be needed to realise the world that Starlink and OneWeb seem to have in mind, and this alone would cost billions of dollars to achieve. One estimate puts the cost of building out the Starlink network at $170 billion. However, some of these costs would be covered by the user in the form of a one-time payment for the ground station.
There’s a UK Start-up that received £9 million of UK Space agency funding that aims to achieve a $600 price point after the full development and testing of their product. This could happen in a few years’ time but is still short of the $200 price point needed for this technology to be viable.
Challenges of this size are generating scepticism about the commercial viability of Low Earth Orbit (LEO) technologies that are dependent on millions of dishes on the ground.
Another issue is that in order for some vehicular based phased array technology company’s satellites to function, they need ground dishes that cost between $30,000 and $250,000. That technology is not fully market-ready due to resonance issues at chip-level, which is a problem that needs several years and tens of millions to resolve. Who knows where Starlink, Amazon’s Kuiper satellites or AST & Science technology - space-based cellular broadband network - will have evolved to by then.
The right business model?
Regardless of the many challenges this technology still faces, it offers many great opportunities. One advantage is the ability to cover a substantial surface area, particularly in less densely populated areas, at a fraction of the cost of land-based equivalent connectivity infrastructure.
It is estimated that moving only the US from a 4G to a 5G wireless network over the next 5 years would cost nearly $300 billion (see Figure 7 below). This is set against a global cumulative mobile operator CAPEX spend of ca. $1.14 trillion worldwide between 2020 and 2025, 78% of which will be in 5G networks. Interestingly, roughly 80% of this will be spent in less densely populated areas. This is where LEO satellite internet may offer an affordable alternative. But what is the right business model for these use cases?
IoT (Internet of Things) will be an integral part of the 5G era. Between 2019 and 2025, the number of global IoT connections will more than double to almost 25 billion, while global IoT revenue will more than triple to $1.1 trillion. Space connectivity infrastructure could offer a capital-efficient alternative in certain use cases to help connect the world.
AST & Science space-based cellular broadband network is gearing up to deploy large ca. 2-ton satellites in LEO to create over 1 million fixed terrestrial cells globally with 4G-5G like broadband capacity. They use High-Throughput Q/V-band feeder links for backhaul. The Mobile Network Operators (MNOs) take on the costs of a very limited number of gateways on earth linking the satellites to terrestrial telecom networks. This enables a direct link to unmodified mobile phones and other cellular devices, eliminating the need for those pricey dishes that Starlink and others need to make their models work.
Additionally, AST uses a business model that foresees in essence a business-to-business (B2B) super-wholesale, 50/50 revenue share model which enables the rapid distribution of their SpaceMobile infrastructure through existing MNOs who currently represent 1.3 billion subscribers. As the MNOs understand the differing & relevant regional regulatory intricacies, and in partnership with AST they ensure a smooth introduction of SpaceMobile across the globe at minimal cost to AST. AST sees its service as complementary to existing terrestrial networks.
According to filed material for their NASDAQ listing in April 2021, they are projected to generate $2.6 billion on a 108 million subscriber base by 2025 and by 2027, ca. $9.7 billion on 372 million subscribers. They are projected to achieve $1bn EBITDA in 2024, $2.6 bn in 2025 and over $9bn by 2027 (90%+ EBITDA margins). AST assume low ARPUs (Average Revenue Per User) relative to market pricing and as low as $1.00 per month in certain markets. They are projected to achieve full global MIMO (Multiple Input Multiple Output) coverage with only 168 LEO satellites vs. 42,000 Starlink satellites and ca. 6,372 OneWeb satellites. This equates to Capital Expenditure of only $3.3 billion for a global coverage space-based cellular broadband network.
SpaceX’s Starlink charges beta residential consumers $99 per month (this is reputedly 4X more for business users but based on an Mbps model), and there’s also an upfront charge of $499 that goes towards the satellite dish and related kit. The company is reporting speeds of between 50 to 150 Megabits per second (Mbps) on the service. In late June 2021, SpaceX’s CEO Elon Musk said that Starlink boosted its subscriber base by nearly 30% in a single month and had passed “the strategically notable number of 69,420 active users,” nearing the 100,000 milestone across 12 countries, but predominantly in the US.
SpaceX was recently awarded close to $900 million in subsidies by the FCC, under the Rural Digital Opportunities Fund, which aims to bring high-speed Internet across the U.S., particularly in rural areas. This could also help the service’s economic viability to a certain extent. That said, competition is also mounting. Wireless carriers have been betting big on 5G technology which can offer gigabit-speed internet over wide areas, competing head-on with the fixed-line internet services. It remains to be seen if SpaceX’s service can offer customers an attractive value proposition as competition increases.
OneWeb also works on a B2B basis — it delivers internet service to existing telecommunications companies who then distribute the internet to homes and businesses. OneWeb will leave the pricing for the telecom firms to set because "they know their customers best," Chris McLaughlin, Chief of Government, Regulation and Engagement at OneWeb, said (April 2021). There are now 322 OneWeb satellites in space (September 2021) at an altitude of 1,200 kilometres.
Like Starlink, OneWeb could be part of the UK's government's $6.9 billion Project Gigabit internet plan, which aims to provide faster broadband to more than 1 million homes and businesses in rural areas of the country. SpaceX reportedly took part in discussions with a UK minister in March 2021.
Meanwhile, Amazon executives have said Kuiper could complement Amazon Web Services’ cloud operations. The FCC approved the launch of 3,236 Kuiper satellites and Amazon is obliged to launch at least half of them by 2026 to retain the operating license. The satellites will operate in the Ka-band (26.5 to 40 gigahertz) with the spacecraft flying at altitudes ranging from 590 kilometres to 630 kilometres. Amazon has committed to invest $10 billion into the program. In April 2021, the company said Project Kuiper’s workforce exceeded 500 employees. In July 2021, Facebook’s satellite connectivity team switched over to Amazon’s Project Kuiper.
Experts suggest that Amazon’s real play will come from its ability to vertically integrate Kuiper into the rest of the Amazon ecosystem—an ability SpaceX cannot match with Starlink. “With Amazon, it’s a whole different ballgame,” says Zac Manchester, an assistant professor of aeronautics and astronautics at Stanford University (August 2020). “The thing that makes Amazon different from SpaceX and OneWeb is they have so much other stuff going for them.” If Kuiper succeeds, Amazon can not only offer global satellite broadband access—it can include that access as part of its Amazon Web Services (AWS), which already offers resources for cloud computing, machine learning, data analytics, and more. AWS also recently launched their AWS Space Accelerator programme with Seraphim Capital in April 2021 to encourage novel use cases for their space internet ambitions.
Facebook has decided to exit the satellite business, with its team moving over to Amazon’s Kuiper team. Instead, Facebook is focusing on terrestrial connectivity programs including Magma (which Amazon Web Services has integrated into its edge computing services), Evenstar, Terragraph and Express Wi-Fi. Facebook has also invested in more than 6,000 miles of terrestrial fibre and 23,000 miles of subsea fibre as part of its campaign to enhance global networking – see its part ownership in transatlantic projects such as MAREA, HAVFRUE/AEC-2 and Amitié Cable in Figure 1 above.
Things to keep an eye on
Although the technology is promising, there are plenty of hurdles still to overcome before economic viability is reached. Investing in satellite internet could potentially offer investors extraordinary returns, but it’s important to tread lightly and take some of the biggest challenges facing this new and exciting industry into account.
Mass adoption may be hard to come by
I would question the assumptions made by OneWebs and similar promoters regarding the mass adoption of their technology. Although improved speed and coverage are important benefits, the issue of bandwidth with currently available technology and the time/effort/costs for all consumers/companies may not be worth it, leaving just consumers/companies in remote locations and those in time-sensitive industries as potential clients.
It seems from Musk’s comments at a satellite conference at the beginning of 2020 that Starlink is not aiming for the mass adoption market and is more focused on serving the "3 or 4 percent hardest-to-reach" customers and those who "have no connectivity" right now. By appealing to those customers, Musk feels a "significant load" will be removed from other Internet Service Providers (ISPs) and by consequence the Mobile Network Operators (MNOs).
This may lead the wholesale distribution via existing operators to be more viable, reducing the 80% of their capex costs spent on the more remote areas they serve.
However, the jury is still out if Starlink will be able to generate Musk’s predicted $30 billion yearly revenue.
Launch and maintenance comes with significant risk
Another issue is physically launching and maintaining over 40,000 Starlink satellites in LEO. LEO satellites typically have an operational life of only 5-7 years. This not only results in potentially high costs for regular launches and maintenance but also means they run the risk of losing their license if any launches don’t meet regulatory requirements.
The emergence of in-space services like those of D-Orbit, Orbit Fab and others may enable the extension of satellite life by the availability of in-space refuelling & maintenance services to enable the satellites to re-boost themselves to their desired altitudes.
Regulatory requirements & politics are uncertain
You also need to bear in mind that regulatory requirements may create other challenges for the companies in this industry. Issues relating to space debris (mapped by the likes of LeoLabs), the reflection of bright light, and the limiting/blocking of launches of other space-bound transports/shuttles are just some of the potential roadblocks that could face regulatory scrutiny in the years to come.
Perhaps there is some new technology that will provide solutions to these important roadblocks. However, currently, the situation seems to favour the sceptics.
The good news…
Regardless of all the hurdles that these technologies still need to overcome, existing companies have already proven that viable business models are within reach. That fact, combined with a total addressable market nearing the one-trillion dollars, has been enough reason for investors to enter into the space at an astounding rate.
Telecom providers are under pressure to reduce CAPEX spend, but with the remote areas encompassing around 80% of their costs, this has proven a difficult task. Combined with the pressure to improve coverage, satellite-based internet offers a unique opportunity.