The next billion people who get connected to the Internet may be looking to the heavens. That’s where a race is on to provide online access from fleets of satellites, led by a who’s who of tech and several deep-pocketed startups.
The aim is to help connect people in developing countries, provide speedier online access to mainly rural users who depend on today’s slower and more expensive satellite Internet services, and cater to business customers that want real-time data from their equipment, like oil rigs and ocean buoys.
The biggest names in the race include Facebook, Elon Musk’s SpaceX, and OneWeb, backed by Japanese billionaire Masayoshi Son’s SoftBank Group. They’re pitted against dozens of upstarts like Swarm Technologies, Astrocast, and Sky and Space Global that want to send cheap, toaster-size satellites called cubesats into orbit.
Wary of the competition, existing satellite-based Internet providers such as Viasat and EchoStar’s Hughes Network Systems are moving to defend their businesses. They plan to launch new satellites that are far more powerful than their predecessors.
For now, the satellite-based broadband industry is relatively small. It will account for only $4 billion in revenue this year, according to Morgan Stanley. But with all the planned networks, and a shift in consumer Internet habits, sales are expected to reach the stratosphere. By 2024, revenue is expected to rise to $22 billion and then to $41 billion in 2029.
The current space race harks back to the 1990s, when several satellite companies launched with similar aspirations. The Bill Gates–backed Teledesic, as well as Iridium and Globalstar, bragged about their big plans but ended up disbanding or filing for bankruptcy after costs soared and money from investors dried up.
It’s a sometimes dangerous adage, but this time really should be different. New rocket-launching companies, including SpaceX and Jeff Bezos’s Blue Origin, are driving down the cost of launching satellites, while the miniaturization of computers has enabled smaller and cheaper spacecraft.
“Each kilogram that you can put at 300 miles above the earth’s surface is many orders of magnitude superior to what we had in the ’90s,” explains Morgan Stanley analyst Adam Jonas.
Companies pushing into the satellite Internet business fall into different categories based on the size of the satellites they plan to use and how high into orbit they hope to send them. They also differ in whom they’re targeting as customers and the kind of Internet access they want to provide.
For example, SpaceX’s initiative, called Starlink, involves sending 12,000 satellites aloft over the next few years, or 50% more than the total number of objects blasted into space since Sputnik. The mini-refrigerator-size satellites would orbit at as low as 340 miles, well below typical communications satellites that are parked in geostationary orbit at 22,300 miles.
By launching so many satellites, Starlink says, one will always be overhead for customers. Though it may be difficult for Starlink to match the speed of ground-based broadband from the likes of AT&T and Comcast, the aim is to provide affordable high-speed service over vast areas of the globe that lack top-notch wired access, anywhere from rural Arkansas to Africa.
Last year, Starlink sent up its first two test craft—dubbed Tintin A and Tintin B after the Belgian comic book character—and received U.S. government approval for a full 12,000-bird fleet.
That puts it far ahead of Facebook, which just filed a request with the federal government to fly a single experimental satellite in low orbit. Facebook hasn’t committed to deploy a full-fledged network or revealed much about its strategy.
Some startups are trying to take advantage of declining satellite costs by using tiny cubesats, which can cost just a few hundred thousand dollars each. That’s a big savings from typical satellites, which can cost hundreds of millions of dollars.
Swarm CEO Sara Spangelo, the aerospace engineer who previously worked at Google’s experimental incubator X and at NASA, says her smaller, cheaper fleet will be operational years ahead of the bigger players. The company already has seven satellites aloft out of 150 planned.
Of course, the network won’t provide high-speed Internet access to millions of subscribers, either. Many initial customers will be businesses that want to collect data or need Internet connections only occasionally. In January, for example, Swarm partnered with Ford to use its satellite network to provide emergency communications in cars—much like GM’s better-known OnStar network, which relies on regular wireless networks.
Compare those plans with Viasat’s $1.4 billion initiative to send two satellites aloft over the next two years. The satellites, each bigger than a school bus, would have twice the combined capacity of all 400 communications satellites currently in space, Viasat says.
Analysts note that small players may be able to succeed in a crowded market with their cheaper networks. Getting a foothold no longer requires a huge upfront investment.
Still, Tim Farrar, president of satellite consulting firm TMF Associates, doubts that all the challengers will survive, particularly the upstarts. Even with today’s cheaper technology, they’ll need big money to pay for their networks and to maintain them.
Says Farrar: “There will definitely be a shakeout.”
A version of this article appears in the February 2019 issue of Fortune.
(Update: This story was updated on Nov. 27, 2019, to correct that Teledesic never filed for bankruptcy.)