Internet Giants Are All Eyeing Space. What Has Changed About Launching Satellites?

Startups, not government agencies, are the main players in the new space race

Over the past year, the tech giants have kept turning up in stories about satellites.

This year, Google shut down its satellite project after the executive running it left, and Facebook has just dropped its plan to spread internet access by satellite because of the cost, even though the social network is worth as much as US$250 billion...

Yet the satellite business still draws a steady stream of new competitors.

More companies keep joining in. Blue Origin, owned by Amazon founder Jeff Bezos, has just finished testing its spacecraft. After leaving Google, Greg Wyler founded OneWeb, which plans to launch 648 low-orbit satellites; he has brought in Airbus as a partner and raised US$500 million. SpaceX is winning rocket and satellite launch work by charging less. And after its own project fell apart, Google invested in SpaceX and bought Skybox, a satellite imaging company...

This is not the tech industry's first satellite craze. Microsoft and Motorola both once poured money into satellites, and both failed.

So what turned a market that governments once ran into ground that tech companies are fighting over?

More and more small satellite companies

Here is a curious fact: most of the satellites launched last year were sent up not by traditional space companies but by venture-backed startups.

The satellite images we see on Google Earth are usually bought from DigitalGlobe. Founded in 1992, the company sets the standard in commercial satellite imaging. Its large remote-sensing satellite, WorldView-3, works about 616 kilometers above the Earth, can deliver images with a resolution as fine as 0.3 meters, and cost US$400 million to build. But its photos are at least six months, and sometimes several years, old.

DigitalGlobe's WorldView-3 satellite, which weighs 2.8 tons

The startups that have entered the industry in recent years offer something different. They use swarms of small satellites in low Earth orbit. Their images are not as sharp as DigitalGlobe's, but they can launch more satellites for less money and supply more recent photos.

An artist's impression of a Skybox satellite, which weighs about 100 kilograms

Skybox Imaging was founded in 2009 and launched its first low-orbit satellite in 2013. The satellite weighed about 220 pounds, was roughly the size of a small refrigerator, and took pictures at a resolution of about 1 meter. That falls short of DigitalGlobe's ultra-high-definition photos, but each Skybox satellite costs less than US$20 million to launch and lasts four years, while each DigitalGlobe satellite costs as much as US$400 million.

Skybox now relies mainly on six low-orbit satellites, which can photograph the same spot up to four times a day and deliver the images in real time. In 2014, Google bought the company for US$500 million.

A Planet Labs satellite, which weighs about 4 kilograms

Planet Labs was founded at the end of 2010. Its CEO, William Marshall, said at the World Geospatial Developers Conference (WGDC) last month that the company now has 87 satellites in orbit, and will soon pass 100.

Planet Labs' images are not very sharp, only 3-5 meters in resolution, but it makes up for that with numbers. To keep costs down, the company flies flocks of small satellites; each of its Dove spacecraft measures just 10x10x30 centimeters. It works with launch providers such as NASA and SpaceX to get the Doves into space, and sells the 3-5-meter-resolution images they capture to companies in different industries, such as mapmakers and construction firms.

BlackSky was founded less than three years ago. Its parent company, Spaceflight Industries, has raised a US$20 million Series B round and plans to launch six satellites next year and build a constellation of 60 by 2019. The satellites will sit in specific orbits and photograph major cities at a resolution of about 1 meter, 40-70 times a day.

Unlike Skybox and Planet Labs, BlackSky is building a satellite imagery platform, with users ranging from individuals to governments and data analytics firms.

There are others too: Canada's UrtheCast; PlanetiQ, which forecasts space weather; Teledyne, which specializes in Earth imagery taken from the International Space Station; and more...

At first, satellites were the governments' game

Satellites, and the rockets that carried them into orbit, began as research projects that the American and Russian governments pursued for military ends.

In 1957, the Soviet Union launched the first artificial satellite, Sputnik 1. By the 1960s, with the Cold War and the space arms race under way, rocket technology took off, and payload capacity rose from 20 tons to as much as 120 tons. Militaries used satellite imagery to show off their countries' technical strength and to threaten other nations, because rockets could now carry not just satellites but intercontinental missiles.

Sputnik 1

By the 1960s, the United States had already spent US$25 billion on its space program. Because rocket launches were so expensive, the government, besides launching its own satellites and missiles, rented out space inside its rockets to private satellite companies. Many of those companies' first customers were in television, which needed to relay broadcasters' analog signals to TV sets.

Carrying satellites into space on rockets gradually became a government business. In the 1970s, the United States launched the Space Shuttle program, which made a shuttle that could be reused 100 times the main vehicle for carrying cargo.

NASA's Space Transportation System User Handbook, published in 1977, put the cost of renting a shuttle at US$20 million per flight. Several companies could also share one shuttle to split the bill. At the time, launching a satellite 1 meter long and weighing 1 ton cost about US$3 million, in 1970s dollars.

Later, private companies also began offering rocket and shuttle launches for civilian customers. Most of them had government ties and deep pockets, and their total costs were lower than what a government would spend developing and launching satellites itself. So when governments had civilian needs, they became customers of these companies instead, shifting the cost of further rocket and shuttle development onto them.

Because of the high costs, and because working with governments was slow, both Motorola and Bill Gates stumbled in the satellite business.

In the late 1990s, a cash-rich Motorola had already dreamed of linking the whole world by satellite, in a project called Iridium. It spent US$3.4 billion launching 66 low-orbit communications satellites (77 were originally planned) to offer satellite phone service worldwide. But as ground-based networks improved, too few people used the expensive satellite phones. A year after Iridium began service, it filed for bankruptcy protection in 1999.

An Iridium 9575 satellite phone

Teledesic, the company Bill Gates founded with Craig McCaw, the "father of the cellular phone," fared even worse. Its grand plan was to launch 840 low-orbit satellites to form a communications network. It launched just one before declaring bankruptcy in 2002.

Why companies worth less than one old satellite want to launch hundreds

OneWeb has raised US$500 million so far and plans to launch more than 600 satellites. That sum would have paid for just one launch of the Space Shuttle Endeavour in the early 1990s. So what has changed, that satellite startups without deep pockets can now make money from cheap satellites?

There are three main reasons. First, the way satellites are built, or more precisely developed, has changed. Second, demand for satellite services keeps growing. Third, the launch industry itself has changed: there is less reliance on the big launch consortiums, and launch costs have fallen.

Satellites used to be large, long-lived, high-orbit machines developed with government money. The Upper Atmosphere Research Satellite (UARS), which the United States launched in 1991, weighed 6 tons.

Startups, by contrast, build satellites weighing around 1 kilogram that fly in low orbit, 300-2,000 kilometers above the Earth. The growth of consumer electronics has supplied plenty of cheap, capable components on which small satellites can be built. Skybox and Planet Labs both build their satellites from parts already on the commercial market: radio transmitters from mobile phones, reaction wheels from dental tools, processors from cars and medical devices. For the most important part, the camera system, they combine cameras from the film industry with open-source software from the internet.

In both software and hardware, Skybox and Planet Labs follow the "lean startup" thinking common in Silicon Valley.

Skybox's satellites cannot match the ultra-sharp images taken by DigitalGlobe's expensive satellites, so the company puts its effort into software. Its algorithms combine dozens of images taken from different angles to produce clear images at 1-meter resolution. Better algorithms place far fewer demands on the satellite itself and cut the power it needs, which lowers costs further.

Planet Labs' Dove satellites last only 1-3 years, which makes them all but disposable. But being able to throw them away means the company can put its newest technology into each new Dove. Combined with steadily improving image analysis software, this fast cycle of replacement suits a startup far better than traditional satellites that take years to develop.

At the same time, many companies have sprung up to supply these satellite startups. Pumpkin, in San Francisco, has sold microsatellite components since 2000. Its microsatellites consist of an aluminum frame, a motherboard, software, memory and a processor. The result looks like a hardware Rubik's cube, 0.9 kilograms in weight and 0.1 meters tall.

A Pumpkin microsatellite, which looks like an ordinary circuit board

In Britain, PocketQubes makes pocket-sized cube satellites only 5 cubic centimeters in volume. For as little as US$5,999, students, researchers and space enthusiasts can build a cheap, fully working small satellite.

In 2013, the space business brought in a total of US$256.2 billion. Satellite launch services made up 33% of that, and services for consumers on the ground made up 58%. The latter include crop monitoring, city management and weather forecasting based on remote-sensing images, as well as satellite communications.

Peter Wegner, BlackSky's chief technology officer, said in an interview that many algorithm and analytics companies have a strong appetite for images of the Earth. And Iridium's revival and the rise of satellite internet show that people's demand for internet access has never weakened.

Meanwhile, the launch industry is changing too. Alongside the satellite companies, the business of launching rockets to carry satellites into low orbit has also boomed. Both share the same goal: launch more often, so that building infrastructure in space costs less.

Until recently, the satellite launch market had no more than eight competitors, most of them backed by wealthy conglomerates or with deep government ties. Examples include France's Arianespace, an arm of the European Space Agency; China Great Wall Industry Corporation, which is authorized by the Chinese government; and America's United Launch Alliance, a joint venture of the aerospace giants Boeing and Lockheed Martin.

In 2004, United Launch Alliance quoted between US$70 million and US$90 million per launch. In 2012, it launched four times for a total of US$1.74 billion, an average of US$435 million per launch. International Launch Services charged about US$100 million per launch, and Arianespace averaged US$200 million.

And SpaceX, the best-known rocket startup? In 2014 it charged about US$54 million per launch, and it has kept testing reusable rockets, which would cut launch costs by reusing the most expensive part, the booster. A Falcon 9 launch currently costs about US$61 million. If SpaceX's reusable rockets one day succeed, each trip to space could cost just US$3-5 million.

According to the consulting firm Teal Group, 303 satellites went into space in 2014, and half of them, about 150, were small satellites. So a niche market for launching small satellites has grown up as well.

NanoRacks, in San Francisco, has developed Kaber, a launch adapter for microsatellites that can carry up to 100 kilograms of payload at a time (about two microsatellites). A SpaceX rocket will take it to the International Space Station at the end of 2015.

In the past, no rocket was dedicated to small satellites. Small customers usually had to wait several months to a year, depending on rocket launch schedules, before their satellites could hitch a ride into space alongside a commercial satellite. Remember that in 2012, United Launch Alliance launched only four times.

Some companies launch rockets from aircraft instead, a technically less demanding approach. Generation Orbit, a launch company founded in 2011, cuts costs by mounting a rocket on a Gulfstream business jet, and offers launches for US$2 million. It has already received 10 letters of intent, with customers asking for 200 to 250 launches a year.

OneWeb and the plan for satellite internet

In this round of the space race, one small company stands out: OneWeb, which has investment and partnership ties with Airbus, Coca-Cola, Qualcomm and the Virgin Group. Its founder also led Google's earlier satellite project.

OneWeb's founder, Greg Wyler, made his first fortune at 25 by designing and selling computer parts. He went on to found the telecom startup Terracom, the satellite company O3b Networks, and now his third company, OneWeb.

In 1992, after finishing an undergraduate degree in finance and computer science, Wyler founded his first company, No Overhead, which built custom desktop PCs. Its most important product, though, was a new heat sink he designed from a large block of metal, to replace the noisy, bulky fan coolers then in common use.

In 1994, Wyler founded Silent Systems to sell his heat sinks. To win orders, he worked long hours at a computer in a cubicle while also getting through law school at the Illinois Institute of Technology in Chicago. Five years later, Silent Systems had made the best heat sink in the desktop industry, at a cost of only 50 cents. It won him customers such as Dell and Hewlett-Packard, and it also made Wyler his first fortune: the electronics maker Molex bought Silent Systems for US$100 million.

Wyler's dream of connecting the world by satellite began in 2003. After a chance meeting with Rwanda's president, Paul Kagame, Wyler went to Rwanda and set up a telecom company called Terracom, which provided mobile phone and internet service to Rwandans.

Wyler laid fiber-optic cable there and built Africa's first 3G mobile network. Although it signed up many subscribers, he eventually sold Terracom, because a fiber network has limits and could not connect well to the global internet. So he created O3b Networks.

As the name suggests, O3b stands for the "other 3 billion." Wyler's goal was to bring internet access to the regions around the equator using low-orbit satellites. Because they orbit only 5,000 miles above the ground, latency fell from the 500 milliseconds of geostationary satellites to 150 milliseconds, almost as good as fiber-optic cable.

The service area O3b is targeting

O3b has now launched 12 satellites and become the largest internet service provider in the Pacific, winning orders from network operators on Pacific islands, cruise lines and oil rig companies. It expects revenue to reach US$100 million this year.

In 2010, Google acquired O3b Networks. Wyler himself spent less than a year at Google before quitting, saying that Google could not build satellites.

And so OneWeb was born.

This time, a few dozen satellites will not satisfy Wyler's ambition. OneWeb plans to launch 648 satellites, and they will orbit closer to the ground, only about 750 miles up. Wyler hopes to cut the time it takes signals to travel between satellite and ground to 20 milliseconds, giving people high-quality internet access.

O3b mainly serves telecom companies with large receiving antennas. OneWeb, by contrast, mainly targets individuals and public facilities such as schools and hospitals. Its terminals broadcast a hotspot, and nearby devices can connect to it over Wi-Fi or LTE.

His persistence has won him many partners. After the Virgin Group and Qualcomm each invested, Airbus announced that it would build the satellites, and Virgin Galactic will handle the launches.

OneWeb has now reached agreements with some carriers to abandon fiber-optic lines on the ground and use satellites instead to carry signals between their base stations. Add the Ku-band spectrum rights Wyler holds, and this may be the closest he has come to his dream.

If the last century's space race began in 1957 with the Soviet Union's first artificial satellite, Sputnik, then today's new space race is once again starting with satellites.

There are already 3,000 artificial satellites at work above the Earth, yet last year there were still 132 recorded satellite launches. If SpaceX's and OneWeb's plans go well, we can expect a very crowded sky.

The space race is back. This time, we no longer have to wait for governments.

Cover image from Bloomberg. Images in the text from PlanetLabs, CubesatShop, RemoteSatellite and SatelliteToday