More than 2,500 active satellites now orbit the Earth, and amateur astronomers and other observers are seeing more every month. 1 Historically, satellite communication involved geosynchronous (GEO) spacecraft—large systems that have become increasingly capable over the years. But now nongeosynchronous-orbit (NGSO) communications constellations, including low-Earth-orbit (LEO) and medium-Earth-orbit (MEO) satellites, are taking to the skies, and their number could soon soar. If current satellite internet proposals become reality, about 50,000 active satellites will orbit overhead within ten years. Even if the most ambitious plans do not come to pass, the satellites will be manufactured and launched on an unprecedented scale.
The ambitions for the large LEO concepts may recall the 1990s, when several companies tried to provide global connectivity. Globalstar, Iridium, Odyssey, and Teledesic had impressive plans. In the end, however, all but Iridium scaled back or canceled their intended constellations because of high costs and limited demand. All suffered financial problems. After that experience, many industry analysts and investors remain skeptical about the viability of large LEO constellations. The recent failures of LeoSat and OneWeb reinforce that impression.
But much has changed over the past 20 years. Satellite technology has advanced; demand for bandwidth has soared, with no slowdown in sight; and companies have developed creative business models to generate profits from connectivity. Moreover, both tech companies and investors now have much larger stores of capital to invest, making it possible to fund large constellations—although this capital clearly does not have infinite patience.
These changes could well make satellite connectivity 2.0 a success. Our analysis, however, indicates that companies planning large LEO satellite internet constellations still need to reduce a range of costs significantly to ensure long-term viability. Lowering launch costs is one part of the equation, but it will be equally or more critical to reduce the cost of manufacturing spacecraft, ground equipment, and user equipment. If suppliers and constellation providers can achieve these cuts, they could unlock enough demand for large LEO constellations to transform both the B2C and B2B communications markets.
The COVID-19 pandemic will also influence the satellite market’s future, but as of the date of this article’s publication it is hard to say how great the impact will be. In the near term, any company that tries to secure funding will face challenges because of economic uncertainty and immediate public-health concerns. These challenges will affect the progress of the remaining licensed concepts—Kuiper, Starlink, and Telesat—differently because their ownership and funding approaches vary.
While physical distancing and work-from-home measures remain in place, the development, manufacture, and launch of large LEO satellites will slow. But the crisis has also caused a spike in demand for internet connectivity and underscored its importance. Investment in any kind of new connectivity infrastructure will be expensive but will almost certainly be needed. Going forward, large LEO concepts could play an important role in meeting this increased demand.
Traditional communications satellites with GEO orbits have proved their worth since the 1960s. Although costly, they are highly capable and have long service lives. Their altitude—more than 35,000 kilometers from Earth—provides them with a wide field of view, allowing operators to cover most of the planet’s surface with three satellites spaced at appropriate intervals. Recent technological advances, including new high-throughput and reconfigurable designs, have improved both efficiency and performance.
The new LEO-satellite concepts, which orbit 500 to 2,000 kilometers from Earth, offer faster communications (they have lower latency) and often provide higher bandwidth per user than GEO satellites do—even more than cable, copper, and pre-5G fixed wireless. Communication occurs through a constellation of LEO satellites; global coverage requires a large number of spacecraft. These concepts will require major changes in satellite operations, including manufacturing and the supply chain, since they ask more of a satellite and shorten its average life span (estimated to be about five years with Starlink, the SpaceX constellation, for example).
With the demise of OneWeb, SpaceX is well ahead in the race to deploy an operational system. For Starlink, 422 satellites were in orbit as of late April 2020, and the company claims that it can begin offering commercial service this year. Telesat, with a proposed initial constellation of 117 spacecraft and the potential to deploy more than 500, appears to be moving forward with its plans.Amazon, which has filed to launch 3,236 spacecraft in its Kuiper constellation, also appears to be proceeding and plans to move its growing team into new facilities this year.
Why the renewed interest in satellite constellations? Our research suggests that it springs from a convergence of forces that make both the development and the market success of large LEO-communication systems more likely now than in the past: technological advances, the emergence of new business models, better funding, and higher demand for low-latency bandwidth (exhibit). Thanks to these developments, the current situation bears little resemblance to the 1990s, when large LEO concepts failed to gain traction.
The most relevant satellite-technology advances fall into four categories. Although these improvements benefit all types of satellite-communications systems, they may be particularly consequential for the new large LEO concepts.
The large LEO concepts are mainly planning to use Ka band. Some propose V band as well. These frequencies enable higher data rates, smaller antennas, narrower beams, and greater security. Higher frequencies are more vulnerable to weather and rain fade, which is the absorption of a radio-frequency signal by atmospheric rain, snow, or ice; frequencies higher than 11 gigahertz are more vulnerable than lower frequencies. Fortunately, expedients such as improved ground-station design, adaptive coding, and signal modulation can reduce this exposure. Improved spectral efficiency and spectrum-reuse rates can also increase the amount of data a system delivers.
Traditionally, satellites have been accessed and tracked via parabolic-dish antennas. This equipment is poorly suited to LEO constellations, which will have numerous satellites all rapidly crossing a ground receiver’s field of view at the same time. Antennas with electronically scanned apertures (ESAs), also called electronically steerable antennas, can shift beams (and track and access large numbers of satellites) without physical movement. ESAs can also be designed for modular assembly, which could allow manufacturers to produce large numbers of basic parts for use in both constellation ground stations and consumer equipment, thereby improving economies of scale. Other important advances in ground equipment include new predictive analytics and network-optimization techniques that use available ground-entry points more effectively.
The operator of a large LEO constellation must monitor and manage the status and functions of thousands of satellites. Recent advances in analytics, combined with improved computing power and artificial-intelligence algorithms, can assist with these functions while reducing response times and operating costs. Likewise, ISL advances that increase throughput also reduce backhaul costs and improve satellite control and network latency. Combining these elements would promote the autonomous and semiautonomous control and management of spacecraft, reducing staffing requirements.
Just as technology has evolved, so have revenue sources from internet connectivity. In the 1990s, communications companies generally followed a business model in which revenues came from service fees for bandwidth and access; rates were often based on usage. With relatively low demand, this model was not viable for the satellite concepts of the 1990s. Today, it would also be risky to charge consumers for usage time, but for a different reason: there would probably be little uptake for such plans. The preference for unlimited access is clear from the mobile-phone industry, where per-text or per-minute billing has given way to unlimited plans.
Fortunately, companies have new options for generating revenue from connectivity:
Adding to the momentum, companies and investors may now be willing to wait longer for profits from large LEO constellations. Instead of expecting an immediate positive cash flow, many are focusing on business models that facilitate the acquisition of customers and the control of ecosystems, so they may initially set low prices for their offerings to attract business, even if that eliminates the possibility of profits. Their goal is to establish themselves as early leaders and to create a foundation for long-term success, following in the footsteps of many high-tech players over the past 20 years. These businesses first concentrated on creating scale and acquiring a critical mass of users and then shifted their focus to making money from the network.
In the 1990s, many companies could not find enough investors to fund their satellite constellations. Teledesic, for instance, initially proposed spending more than $9 billion to launch 840 satellites but then reduced its plan to about 300. 8 Later, the company entirely suspended satellite production after burning through hundreds of millions of dollars in development costs.
Current satellite concepts will initially be as expensive as or more expensive than their predecessors. Although costs continue to evolve and many uncertainties remain, estimates for deploying an operational system generally range from $5 billion to $10 billion. Annual operating costs will be high: the cost of replacing satellites alone will total $1 billion to $2 billion for a large constellation if their life span is about five years. The ground segment, even if largely automated, will require a substantial number of sites and antennas, which entail significant capital and operating costs. All this will require a substantial upfront investment and the ability to sustain expenses until revenue kicks in, especially if providers offer low prices to attract business. Cost levels, including the potential for decreasing them, are discussed in more detail later in this article.
A few companies have encountered financial issues as they sought to develop large LEO constellations. LeoSat recently ceased operations after being unable to secure additional investment, and OneWeb recently filed for Chapter 11 bankruptcy, again reportedly after running out of cash and failing to secure additional financing. The COVID-19 crisis has injected further uncertainty into the investment market.
Despite echoes of the low investment that followed the bursting of the dot-com bubble, the funding picture is different from what it was 20 years ago. Some companies have enough cash available to build and deploy a constellation outright. Amazon, with $55.4 billion on hand, is the only large tech player with an announced constellation, but Facebook ($54.9 billion) has reportedly filed preliminary LEO-satellite plans through a proxy. Other companies are said to be considering similar ventures, including Apple, which has $107.1 billion in reserves. These companies and others may also finance such developments with cheap debt, since interest rates are at historical lows.
In another shift from the 1990s, companies that need outside investment to support their constellation plans have many opportunities. Venture capital has recently been bullish about space projects, investing more than $4 billion in the past two years alone. Funding from that source could diminish, however. The OneWeb example also shows that the capital needed to establish a system may be greater than the amount that venture investors are willing to provide to a single company. OneWeb raised $3.4 billion from a consortium of investors that included Airbus and Softbank, but this was not enough.
Aside from pure self-funding, the remaining players do offer examples of other approaches:
With the exception of Softbank, private investors have focused largely on space projects involving small launches and Earth observation. However, there is a large amount of dry powder on hand—investment firms had more than $2.3 trillion (and growing) to spend in late 2019—so funding for other space projects could soar if additional investors begin to see potential in the market.
These shifts, combined with the new revenue opportunities, have created a very different investment landscape. In the late 1990s, the $5 billion to $10 billion required to deploy a constellation, and the $1 billion to $2 billion required for annual maintenance, were deal breakers for investors. That is no longer always the case.
Bandwidth needs were modest back in the 1990s, given the nascent internet and low e-commerce and social-media activity. Most consumers were venturing online for the first time, usually on desktop computers with dial-up modems. Cellular-phone use was surging, but the need to connect globally—particularly outside normal terrestrial coverage—was relatively low.
Today, consumers not only routinely download high-definition movies but also play games and shop online, consuming vastly more bandwidth. In addition, entirely new demand segments, including in-flight airline connectivity, have emerged. Other markets, such as telecom backhaul, have greatly expanded with increased mobile usage.
In tandem with increased demand for connectivity, service expectations have risen. Both businesses and consumers seek high-bandwidth connections and, for many applications, low latency. Significantly, these expectations have spread beyond technologically sophisticated users to virtually all consumers in developed economies and many in emerging markets. Only people with limited connectivity options accept lower performance.
At present, the vast majority of consumers rely on terrestrial solutions, and the B2B use of satellites is limited to a few end markets where terrestrial solutions don’t work—for example, in-flight internet, long-distance mobile backhaul, maritime internet, remote oil and gas extraction, and certain military applications. That’s true largely because satellite-connectivity options are so expensive. But if constellation providers can offer competitive pricing, demand could soar (see sidebar “How could satellite demand evolve if costs drop?”).
How can large LEO-constellation providers unlock demand by making their prices competitive with terrestrial solutions? The answer is significantly reducing costs, from manufacturing to launch to user equipment—a difficult undertaking that will require close cooperation with suppliers. Of course, there are other obstacles (see sidebar “Other issues facing satellite providers”), but cost is the greatest challenge to profitability and long-term viability.
Satellites have traditionally been more akin to handcrafted items than to mass-produced goods. That kind of customization, combined with long life-span requirements, explains why a typical large communications satellite costs from $50,000 to $60,000 per kilogram. If costs remain at this level, large LEO constellations would be completely unaffordable. Although some recent GEO communications satellites reportedly are less expensive, this information has not yet been detailed publicly. (We use cost per pound for satellites with similar functions and subsystems as a first-order approximation.) Further, other recent reports suggest that satellite costs will remain high.
If large LEO constellations are to be financially viable, their manufacturing costs must fall by more than an order of magnitude from those of traditional satellites. That would probably be at least 75 percent lower than the costs any company has currently claimed it can achieve (for information on our methodology for estimates, see sidebar “Cost calculations for large LEO constellations”). To cut costs in this way, manufacturers must leverage every possible tool, from economies of scale to automation to reduced component costs across the value chain.
Many experts believe that launch costs should be the main target for cost reductions in large LEO constellations, and owners will certainly want to cut them. Launch providers will have to pull every cost-reduction lever available. In addition to reducing the cost of materials and manufacturing, they should lower their operating costs—for instance, by maximizing savings from reusability.
As mentioned in the technology section, large LEO constellations will require many ground stations, even with high-capacity ISLs. By one estimate, the 4,400-satellite version of Starlink will require 123 ground-station locations and about 3,500 gateway antennas to achieve maximum throughput. The gateway antennas must be larger and will require significantly more power than user terminals do.
Current gateways for GEO satellite communications are quite expensive—typically from $1 million to $2 million each. They are not directly comparable to LEO gateways, which have lower power requirements, but the numbers do suggest that gateway costs must be much lower than those of current approaches to make ground-segment costs manageable. Modular antenna designs could help, since they would enable equally critical cost reductions in user-equipment antennas, but owners of large LEO constellations will also look for other efficiencies.
To be most effective, user equipment for a large LEO internet network must incorporate advanced ESAs. These devices currently cost several thousand dollars, though manufacturing costs may be substantially lower; some analysts suggest that they are in the $300 to $500 range. At current levels, ESA prices would be too high for residential customers—the largest source of potential growth—who now pay about $100 to $200 to purchase customer-premise equipment (CPE) or pay monthly rental fees of $10 to $20. Even if constellation providers made their satellite-access and bandwidth fees comparable to those of terrestrial solutions, the high cost of CPE would severely limit their success in the consumer market. If, alternatively, providers slashed CPE prices to compete, they would incur acquisition costs of several thousand dollars for each customer. Under this model, even a premium product would be unprofitable.
To unlock the consumer market—the one with the most potential—the cost of ESA antennas must drop by an order of magnitude or more. While some companies have recently claimed breakthrough reductions in manufacturing costs, none has yet brought a low-cost design to market, nor have any produced ESAs at scale. Companies that do create less expensive ESA concepts will have to preserve their quality: for instance, ESAs will still need to provide high data rates, reliable beam steering, smooth satellite handoff, and other features that ensure a good customer experience.
For satellite-component suppliers, the large LEO market has significant potential. Although many companies now working on large LEO constellations may produce satellites or even ground equipment in house, they will still require external components and service support. Suppliers could find huge opportunities by helping to reduce costs, and the eventual market could support a large amount of second- or third-party user equipment.
The growth of large LEO constellations will create an unprecedented demand for spacecraft—in particular, high-performance satellites at lower cost. This shift could open the door to specialty providers in a number of areas, including space-qualified solar arrays, power- and thermal-management systems, satellite guidance, navigation and control, on-board processing, and antennas (both to transmit and receive signals). Suppliers that can reduce component costs could be rewarded with contracts for thousands of spacecraft a year.
With the cost of ESAs now prohibitively high, constellation providers will struggle to capture the consumer market. At present, however, they remain undeterred. Starlink and OneWeb have filed blanket license requests with FCC for 1.0 million and 1.5 million user terminals, respectively—all in the United States. Amazon’s filing proposes to connect “tens of millions” of users across the globe. A supplier that can design a reliable but much less expensive unit could see a market for several million devices.
Large LEO constellations, even in their initial form, will require hundreds of ground stations and thousands of gateways to maximize throughput. But the ground stations will not resemble current versions, which have teleports covering acres of ground with dozens of large dishes. Instead, they may be placed in multiple locations (somewhat like cell-phone towers today), with a large number in remote areas. This configuration will require highly automated management systems. The ground stations will also need antennas—almost certainly flat-panel ESAs that have the same design and technology used for consumer equipment, only in scaled-up or modular form.
A single large LEO constellation will require anywhere from three to 40 launches a year (depending on the size of the constellation and rocket type), both initially and during maintenance. For constellation operators—even those that build their own rockets—these launch costs will be significant. To ensure a viable business, launch providers will probably need to reduce the cost to orbit below $2,000 per kilogram.
Companies must have end-of-life plans for constellations. Many, however, lack strategies for addressing anticipated or unexpected on-orbit failures. Concerns about such issues could create demand for a completely new market to find satellites and take them out of orbit.
For the stars to align so that large LEO constellations are deployable, one critical factor stands out: cost reductions across the value chain, from satellite manufacturing through launch and operations. Space operations have never previously occurred on this scale, and the manufacturers and suppliers of both space and ground equipment may find it challenging to meet ambitious cost-reduction and performance targets. If these companies succeed, however, they could serve a burgeoning market. The new constellations would add tens of billions of dollars of economic activity to satellite manufacturing, operations, launch, and consumer equipment. Simultaneously, more consumers will have access to internet connectivity. Together, these benefits should encourage providers of satellite components to persevere.