Amazon files to beam phone service from space, taking on Starlink in a market that tends to crown one winner

Space 5 min 17 sources

Amazon asked US regulators for a 5,105-satellite network to reach phones directly, joining a race that has cost three companies about $36 billion in a single year just to get in.

Key takeaways

Amazon asked US regulators for a 5,105-satellite network to beam phone service directly from space, taking on SpaceX's Starlink.
Three companies spent about $36 billion in a single year just for the airwaves to compete - a sign of a market where only a few giants can afford to enter.
Satellite internet has huge fixed costs and tiny per-customer costs, so the biggest player is the cheapest, and the market tends to tip toward one dominant provider.

Amazon has asked US regulators to let it build a constellation of more than 5,100 satellites that would beam voice, text, and data straight to ordinary smartphones [1][2]. The filing was lodged with the Federal Communications Commission on July 24. It is Amazon’s bid to challenge SpaceX’s Starlink in a business the industry calls direct-to-device, or D2D. That means satellites reaching the phone already in your pocket, with no special dish needed [3][4].

If approved, Amazon would start launching the 5,105-satellite network in 2028 [3]. It runs on radio spectrum - the airwaves a signal travels on. Amazon is buying that spectrum from a company called Globalstar for more than $11 billion, in a deal not due to close until 2027 [4]. Amazon is asking regulators to sign off before it even owns the airwaves, the same move SpaceX made a year ago [4].

The price of a ticket

What stands out is not the technology. It is the cost of getting in the door. Over the past year, three companies each spent billions just to acquire the spectrum a D2D service needs. SpaceX bought a firm called EchoStar for about $17 billion. Amazon is buying Globalstar for over $11 billion. Rocket Lab is buying Iridium for roughly $8 billion [4]. Add it up and that is about $36 billion in one year, spent by three companies, simply to reach the starting line [4].

That is before a single one of Amazon’s new satellites flies. The 5,105 spacecraft, the rockets to launch them, the ground stations to talk to them - all of that comes on top. This is a business where the fixed cost of showing up is measured in tens of billions of dollars.

Why the sky fills with a few giants, not many

The whole space economy was worth about $686 billion last year [10], and the satellite-internet corner of it has a particular shape. The cost to build the network is enormous and mostly fixed - thousands of satellites, the airwaves, the launches. The cost to add one more customer, once the network exists, is almost nothing. In a business built like that, the biggest player is also the cheapest, because it spreads that huge fixed cost over the most customers.

Starlink is that player today. SpaceX launches fresh batches of Starlink satellites almost every few days - its West Coast pad alone flew seven Starlink missions in July, with an eighth planned [5][6]. Its new mega-rocket, Starship, carried the first upgraded Starlink satellites on a test flight last week, a sign of how cheaply SpaceX intends to keep adding to its lead [7]. The company already partners with phone carriers to reach handsets directly; Amazon’s plan is to do the same [3].

So the field is not filling with dozens of competitors. It is filling with a handful of the richest firms on Earth: Amazon, SpaceX, and Europe’s Eutelsat. Eutelsat is quietly working on a new 528-satellite constellation of its own [8]. Everyone else is priced out before they begin. That is the mark of this kind of market: the barrier is not a patent or a law. It is the sheer scale you need to be worth trying at all.

The launch bottleneck

Even the giants are straining. Amazon is behind. It has more than 390 satellites in orbit but had planned to have roughly 1,600 up by this month [4]. The holdup is not building the satellites - it is finding rockets to launch them, and there are not enough to go around [4]. Amazon is drawing on a shared pool of vehicles from several launch companies to catch up.

The scramble for launch capacity is its own story this week. Rocket Lab, one of the three D2D contenders, won a $266 million US Space Force deal to open a launch site in Alaska [11]. When the buyers of rockets outnumber the rockets, whoever can get to orbit reliably holds real leverage.

The crowded sky above everyone

This all lands over your head, whether or not you ever buy the service. Tens of thousands of new satellites are heading for low orbit, and astronomers warn they are already streaking through telescope images and brightening the night sky worldwide [9]. The trouble, researchers say, is that no international law can stop it: “the technology is outpacing the regulatory environment” [9]. A shared sky is being reshaped by a few private companies. Stargazers, dark-sky communities, and poorer nations with no seat at the table inherit the result without a vote.

Elsewhere this week

A US-Russian crew came home. NASA astronaut Chris Williams and two cosmonauts landed in their Soyuz capsule on Sunday after 241 days aboard the International Space Station [12].

NASA’s Roman Space Telescope, a wide-view infrared observatory built to survey huge patches of sky, is set to launch on August 30 [13]. Scientific American, meanwhile, revisited an old question: whether NASA has already found signs of life on Mars. It framed the answer as an open debate over ambiguous data, not a discovery [14]. And NASA picked a successor to CAPSTONE, a small probe that scouted the orbit its planned Moon-orbiting station will use [15].

The under-covered one: Africa aims for the Moon

Quietly, China agreed to launch Africa’s first lunar science mission in 2029 [16]. The project, called Africa2Moon, would set small solar-powered probes on the lunar surface. “If this is successful, it will ignite the reality of space on the African continent,” one organiser said [16]. It is a reminder that while a few firms race to own the sky above Earth, whole continents are still working to reach it for the first time. Astronomers this week also flagged a candidate for the first moon ever spotted outside our solar system, though they stress it may yet turn out to be something else [17].

02 · Lesson · why it matters

Why some markets can only hold one giant

When getting bigger keeps making each unit cheaper, a market doesn't settle into many rivals - it leans toward a single winner, and only the very largest can afford to try being it.

The strange shape of a $36 billion race

Amazon just asked permission to put 5,105 satellites in the sky to reach phones directly. To do it, it is paying more than $11 billion for the airwaves alone. It is not the only one. In the past year, three companies spent about $36 billion between them - not on satellites, not on rockets, just on the radio spectrum to compete. Then come the thousands of satellites, the launches, the ground stations.

Notice who is in this race: Amazon, SpaceX, and Europe’s Eutelsat. Three of the largest players on Earth. Not thirty firms, not three hundred. When the price of showing up is measured in tens of billions, the field is decided before it starts. That is not an accident. It is the shape of the business.

Big makes cheap, and cheap makes bigger

Some things get cheaper the more of them you make. That much is ordinary. What makes satellite internet different is the second half. The cost of building the network is enormous and nearly fixed - the constellation, the airwaves, the launches. The cost of adding one more customer, once it all exists, is almost nothing.

Put those two facts together and something tips. The biggest network is also the cheapest, because it spreads that vast fixed cost over the most customers. Being cheapest wins more customers. More customers make it cheaper still. The lead doesn’t just hold - it compounds. In a business built this way, competition doesn’t spread out into many steady rivals. It slides toward one.

This is the piece that a cost curve alone misses. It is one thing to say the hundredth satellite costs less than the first. It is another to see what that does to the market. It doesn’t produce ten balanced competitors. It produces one giant and a graveyard of firms that never reached the scale where the numbers work.

The barrier no one built on purpose

There is no law saying only Amazon and SpaceX may reach your phone from orbit. No patent locks the others out. The barrier is quieter and harder to argue with: the sheer size you need before the economics even make sense. You cannot enter this market small and grow. You enter it enormous or you do not enter at all.

That poses as a plain fact of engineering, and mostly it is. But choices sit on top of it. Regulators recently locked the best airwaves to the companies already using them, so newcomers had only one way in - buy a firm that already holds spectrum. That is why the year’s story is $36 billion in acquisitions rather than new entrants. A rule meant to keep the system orderly also raised the wall around the giants. It serves the public’s wish for reliable service, and it serves the incumbents. Both are true at once.

Everyone is standing underneath it

You do not have to buy any of this to be inside it. The winner of this race will one day be the thing that connects your phone when there is no tower. Think of a mountain, the open sea, or the hours after a storm knocks the cell network flat. Whoever owns the sky owns that lifeline.

And the sky is shared whether you consent or not. Tens of thousands of these satellites are heading for low orbit. Astronomers say they are already streaking through telescope images and brightening the night for everyone, and no international law can stop it. The stargazer, the dark-sky town, the country with no seat at the regulator’s table - all inherit a sky reshaped by a handful of firms they never voted for.

Even the giants don’t command it. Amazon is far behind its own schedule, not because it can’t build satellites but because there aren’t enough rockets to launch them. The people racing to own the sky are themselves stuck waiting for a ride to it.

What the shape leaves us with

A market that tips toward one winner is not a story of a villain. No one chose for satellite internet to have huge fixed costs and near-zero costs per customer. That shape was there in the physics and the money before anyone acted, and it quietly sorted the players down to a few. The most useful thing to hold is not who to root for, but the shape itself - because it recurs. Think of railroads, power grids, the pipes under a city, the network your messages travel on. Wherever building is dear and serving is cheap, expect a market that leans toward one. And the fight will be over who that one gets to be.

Seeing that shape doesn’t tell you what should happen next. It only makes the race legible. And it reminds you how much of the world overhead is being settled by a handful of seats, none of which can see the whole sky either.

03 · Lab · your turn

Race to Own the Sky

Rehearse how a market with huge fixed costs and tiny serving costs tips toward one giant, and feel why only the largest can even try.

04 · Hope · carry this

The same scale that crowns a few giants is also why an ordinary phone can now reach for help from a mountaintop, or after a storm flattens the towers. And while a handful of firms fight over the sky, a whole continent is quietly working out how to reach it for the first time.