Look up into the sky one night and you might spot one of SpaceX’s Starlink satellites sailing overhead. Relatively small in size (they weigh about 227 kg each and are about the size of a table), there are more than 700 of them orbiting the Earth so far, and the company hopes to launch as many as 30,000 more on top of the 12,000 the FCC has permitted. To put it into perspective, there are currently less than 2,800 satellites in space. If SpaceX succeeds, it will provide high-speed broadband internet to a global market of over three billion potential users. It is the most ambitious project of its kind to date, but will it work?
Other companies have tried to capitalize on the untapped pool of consumers before. In the 1990s, Globalstar, Iridium, Odyssey and Teledesic all took swings at the market, and all fell far short of their goals mainly due to high costs and low demand. This time around, SpaceX, Amazon, OneWeb and Telesat are the big players, with SpaceX leading the charge.
Satellite internet has been around for decades, so what makes this new generation so different? The crucial distinction is that they will fly in low Earth orbit (LEO), an orbit much closer to the Earth (less than 1,000 km from the surface) than geostationary orbit (36,000km from the surface), the altitude at which most traditional internet satellites have flown.
The benefit of geostationary satellites is that they travel at the same speed as the Earth’s rotation, so even though they are moving incredibly fast they appear stationary to an observer on the ground. This allows one satellite to consistently cover a wide area of the planet without needing to be tracked by ground antennas. LEO satellites, on the other hand, orbit much closer and can only service a small spot on the Earth at any time. Hence, tens of thousands of them are needed to provide consistent internet service; as one satellite flies out of view, the next one flies in, and so on. SpaceX’s ultimate goal is to create a web encompassing the globe that can provide high-speed service to everyone.
Satellite internet has a reputation for, to put it lightly, sucking. It suffers from limited bandwidth, slow download speeds, data caps, and high latency. Since it takes a long time for signals to travel 36,000km into space and back, the delay for users is typically about half a second, which is fine if you’re shooting off an email but terrible if you’re playing online games.
With Starlink flying a mere 540km off the Earth’s surface, its latency should fall to levels comparable with fiber optic and cable. That’s the theory, at least. Tests from beta users last month seemed promising, but the FCC has expressed serious doubts that it will be able to reach latencies below 100ms.
SpaceX will need to prove it can meet the FCC’s standards if it wants to bid for the Rural Digital Opportunity Fund (RDOF), a reverse auction initiative that will shell out up to $20 billion to broadband providers to extend internet access into rural America during the next 10 years. The first phase of the process is set to begin on October 22. It is not clear if Starlink will make the deadline.
Earlier this month, the FCC released lists of 121 complete applications and 384 incomplete applications for the RDOF. SpaceX was on the incomplete list, which includes ISPs that did not “provide the certifications and basic information required by the Commission’s competitive bidding rules for participation in the auction and ISPs that have not been determined to be financially and/or operationally qualified to bid in all the states or for all the performance tier and latency combinations it selected.”
Without those subsidies, SpaceX may struggle to sustain its Starlink megaconstellation. It needs to build, launch, and maintain thousands of satellites, each of which must generate its own power, propel and maneuver itself, and operate multiple antennas. And since the satellites only last several years, they will need to be constantly replenished. The capital investment required is tremendous.
“The geostationary satellites that provide this service weigh on the order of 20,000 kg,” says Roger Rusch, president of satellite telecommunications research group TelAstra. “The Iridium system, which is a very narrow band system, weighs 70,000 kg. Then if you look at phase one of the Starlink system, those satellites would in total weigh 1.15 million kg. So, you can get some inkling just from the amount of mass they have to put into space that it’s probably going to cost more.”
SpaceX, Amazon, OneWeb and Telesat plan to launch five times more satellites in the next few years than all objects sent to space in the last six decades combined. This shift would introduce an unprecedented degree of complexity to the space environment, bringing with it a panoply of potential complications. “It has implications for space debris mitigation, the stability of the orbital operating environment, and the pace and efficiency of the regulatory system,” says Ian Christensen, director of private sector programs at Secure World Foundation.
The higher frequency of launches and number of objects zooming around the Earth could increase the risk of an accident on the launch pad or in orbit, more advanced technology and collision avoidance maneuvers notwithstanding. Naturally that risk, or even the perception of risk, would prompt insurers to raise premiums, and particularly expensive claims have shocked underwriters in the past. Swiss Re, the second largest reinsurance company in the world, pulled out of the space market last year following the destruction of an Emirati satellite during launch. In March, Assure Space stopped offering policies covering satellite collisions in low Earth orbit, where SpaceX is sending Starlink. Its Managing Director Richard Parker said, “I can’t charge them what I perceive the real risk to be.”
Of course, SpaceX has some counterarguments for the skeptics. It is vertically integrated, so it doesn’t have to worry about the transactions costs of contracting with a satellite manufacturer, launching its satellites on another company’s rocket, or dishing out hefty royalties for crucial intellectual property. Its exclusive claim to the reusable rocket not only cuts down on launch costs but gives it a competitive advantage over its rivals. But it cannot escape the inherent expensiveness of operating in low Earth orbit. And more importantly, even the sexiest rocket in Musk’s imagination won’t address the elephant in the room the whole system depends on — ground infrastructure.
Satellites don’t magically connect you to the internet. They send and receive signals to and from gateway stations that physically tap into the internet backbone. One of these will need to be constructed in every locality SpaceX hopes to service, and that may present a problem in foreign countries whose governments aren’t so keen on letting an eccentric American billionaire establish his own telecommunications network within its borders.
“Each of these gateway stations needs real estate, electrical power, and a trunkline to the fiber optic network,” says Rusch. “And, of course, they will need multiple antennas. Based upon what we have seen in the past, these stations cost $20-30 million or more, maybe even $100 million.”
In addition, each household will need to purchase and install an antenna to communicate with the satellites. This, Musk admits, is Starlink’s greatest challenge. Because the satellites shoot across the sky in mere minutes, the traditional dishes are inadequate for tracking them. Instead, SpaceX plans to use electronically steered phased array antennas, which cost in the thousands of dollars.
If SpaceX doesn’t devise a way to produce the antennas extraordinarily cheaply, it may run face-first into the same demand problem that sank similar projects in the 1990s. Customers in rural areas, whom Starlink is designed to serve, are precisely the people least likely to afford the kind of sophisticated technology needed for the job. And that dilemma will become especially glaring as the company expands its service outside of North America and Europe.
“One thing that isn’t discussed in the literature I’ve read on Starlink is how in the world you are going to get people to pay for it in developing countries,” says Bill Gibson, John Converse professor of economics at the University of Vermont. “How are you going to send a bill to some guy in the middle of the Gobi Desert? It seems like the transaction costs are not going to be insignificant. I think it’s even difficult for telecommunications companies in the United States to keep their payment flows well-managed.”
Of course, our initial question still remains — will it work? Who knows? There are plenty of reasons to doubt Starlink’s practicality, but SpaceX has proved everyone wrong before. Gibson acknowledges, “every time we say Elon Musk can’t do something, he turns around and does it.”