A five-month-old startup just asked to put 100,000 data centers in orbit

Space 3 min 80 sources

Companies are racing to file plans for enormous orbital data-center swarms — hundreds of thousands of satellites that don't exist yet. Plus a spacewalk arm repair, a nearby maybe-habitable planet, and China's push into very low orbit.

Key takeaways

A five-month-old startup asked U.S. regulators to launch 100,000 orbital data-center satellites, joining a crowd of companies filing for hundreds of thousands of machines that don't exist yet.
The pitch is real — space offers constant sunlight and free room to cool — but every plan is waiting years on SpaceX's Starship rocket to actually get anything to orbit.
Elsewhere: astronauts repaired the space station's aging robot arm, astronomers spotted a maybe-habitable planet 25 light-years off, and China is racing to claim very low orbit.

The strangest number in space this week isn’t a distance or a speed. It’s 100,000 — the number of satellites a company called Orbital just asked permission to launch.

A paper land-rush above your head

Orbital is five months old. It emerged from stealth earlier in June with $5 million in early funding [4]. On June 24 it filed with the U.S. Federal Communications Commission — the agency that hands out the rights to broadcast from orbit — for permission to deploy up to 100,000 “data center satellites,” aiming to put 10 gigawatts of computing power in space to feed the demand from artificial intelligence [4].

None of these satellites exist. The company’s first real one is slated for 2028, and it says the full swarm likely wouldn’t fly until “well into the next decade” [4]. Its founder, Euwyn Poon, previously ran an electric-scooter company [4].

Orbital is not alone, and that’s the story. Starcloud has filed for an 88,000-satellite constellation [4]. Cowboy Space filed for up to 20,000 [4]. SpaceX has outlined plans for up to a million orbital data centers [4]. Blue Origin, Jeff Bezos’s rocket company, is chasing the same idea [4]. A crowd of companies is filing claims for hundreds of thousands of machines, almost none of which have been built.

The pitch is real enough. Data centers on the ground are hitting walls: power, cooling, and land [4]. Up in orbit, sunlight is constant and space is free — you can spread out a 100-meter solar array and radiate heat into the dark [4]. The proposed satellites are big — 100-kilowatt-class, weighing up to 2.5 metric tons each, the mass of a large car [4].

But the gap between the filing and the flying is enormous. Poon put it plainly about the hard part: “The complexity is all launch” [4]. Every one of these swarms is waiting on SpaceX’s giant Starship rocket to work reliably before anything at scale can reach orbit [4]. The paper race is years ahead of the real one.

Astronauts operate on a broken robot arm

Two NASA astronauts went outside the International Space Station on June 30 to repair Canadarm2, the station’s 17-meter Canadian-built robotic arm — the crane that catches arriving cargo ships and moves equipment [9][16]. NASA billed the spacewalk as astronauts “operating” on the arm, and the repair work continued across follow-up outings [38][42].

It’s a reminder of what the station is now: a 25-year-old machine that needs constant patching. The arm is one of its most-used tools, and fixing it in a vacuum, in bulky gloves, is delicate work. Aboard the station, the crew also wrapped up a week of cartilage and cardiac research [19].

A maybe-habitable world, 25 light-years away

Astronomers reported a rocky planet, GJ 3378b, orbiting a faint red dwarf star 25 light-years away in the constellation Camelopardalis, the Giraffe [40]. Twenty-five light-years is close by cosmic standards — but still about 240 trillion kilometers, far beyond any reach we have [40].

“Potentially habitable” is doing careful work in that sentence. It means the planet sits at a distance from its star where liquid water could, in principle, exist — not that anyone has seen water, air, or life [40]. Red dwarfs are dim and prone to violent flares, so a planet in the right temperature zone may still be a hostile place. It’s a candidate worth a closer look, not a second Earth.

China stakes out the very low orbit

China set up a national industry alliance on June 27 for “very low Earth orbit” — the band below about 300 kilometers, where the air is thin but not gone, so satellites there need constant thrust just to avoid dragging down and burning up [35]. Several Chinese satellites are now demonstrating sustained operations at those altitudes [35]. Flying lower means sharper images and faster signals; the trade is fuel and engineering to fight the drag. It’s a quieter move than a giant swarm — but it’s the same instinct now running through the whole industry: claim the new territory before the rules for it are written.

02 · Lesson · why it matters

Why a hundred thousand satellites can exist on paper and nowhere else

Claiming a thing is cheap and fast; building it is dear and slow — so the announced world always runs years ahead of the real one.

Start with the number that can’t be true

A company five months old, with six people and $5 million, asked the U.S. government for permission to launch 100,000 satellites. Say that slowly. The entire history of spaceflight, across every nation and company since 1957, has put roughly 20,000 objects into orbit. One startup just requested five times that, for itself, before it has flown a single unit.

The instinct is to laugh or to gasp. Both miss the point. The filing is not a lie and it is not a plan to launch 100,000 satellites next year. It is a different kind of object entirely — and once you see what kind, the whole crowded race behind it starts to make sense.

A claim is not a thing

There are two worlds here, and they run on different clocks.

In one world you build a satellite: you design it, source the parts, test it, put it on a rocket, and get it to orbit. That world is slow, expensive, and unforgiving. Orbital’s own founder named the hard part in four words — “the complexity is all launch.” Every one of these swarms is waiting years for a rocket big enough to carry it, SpaceX’s Starship, to work reliably first.

In the other world you file a claim. You write a document, pay a fee, and ask a regulator to reserve the right to do something later. That world is fast and cheap. A filing costs a rounding error next to a single launched satellite. And crucially, filing does not commit you to building anything. It buys you a place in line.

The 100,000 is a number from the second world. It is not a forecast of hardware. It is the size of the claim.

Why the claim is always bigger than the plan

Once you know a claim is cheap, you can predict its size: it will be large, and larger than anyone actually intends to build.

Two forces push it up. The first is that regulators hand out scarce things — slices of orbit, radio frequencies — roughly in the order people ask. Get your filing in early and big, and you sit near the front when the real allocation happens. Ask small, and you may find the good altitudes and frequencies already spoken for by someone who dreamed bigger. So everyone dreams bigger.

The second is signalling. A number like 100,000 is not really addressed to the regulator. It is addressed to investors, rivals, and reporters. It says: we are serious, we are enormous, take us seriously. Starcloud filed for 88,000. SpaceX outlined up to a million. Each number is partly a genuine ambition and partly a flag planted to look unignorable. When the cost of a bigger flag is near zero, the flags climb.

This is why you can’t read the size of a claim as the size of a coming thing. The number is inflated by design — not by dishonesty, but by the plain incentives of getting in line and being seen.

The map is drawn before the territory exists

Here is the part that binds you in, wherever you sit.

Almost every field has these two clocks — the fast clock of announcements and the slow clock of delivery — and almost every field lets the fast one set the mood. A drug company files a patent; a country announces a target; a firm reveals a roadmap; a founder tweets a plan. Each of these is a claim, cheap and quick. Each gets reported as if it were the thing itself. And so the world we hear about is the announced world, drawn on the fast clock, while the world we live in arrives on the slow one, quietly, years later, usually smaller.

You are not above this. When you read “100,000 satellites,” some part of you files it as a fact about the near future. When a company says it will change an industry, part of you starts living in the changed industry before a single customer has been served. The gap between what has been claimed and what has been built is where most disappointment — and most quiet bad decisions — are made. It is also where the money moves first, and where the people far from the headline get sorted into winners and losers before anything real has happened.

What the gap actually holds

None of this makes the orbital data centers fake. The pitch has real physics behind it: sunlight in space is constant, and the dark is a free place to dump heat, while data centers on the ground are choking on power and cooling and land. Some version of this may well fly. That is the honest tension — a real idea and an unreal number, arriving together, and our habit of hearing only the number.

The skill is not cynicism. It is holding the two clocks apart in your head: noticing when you have been handed a claim and quietly asking what it would take to make it a thing, and how long, and who has actually done the slow part yet. Almost no one can see the whole distance between the filing and the flying — not the founder, not the regulator, not the reader. The wise move is to know which world a number came from, and to hold it as loosely as its clock deserves.

03 · Lab · your turn

File Your Claim

Rehearse setting the size of a paper claim and watch how little of any filing ever becomes a real, flown thing.

04 · Hope · carry this

The wild numbers are ambition getting ahead of itself, and that has always been the first thing progress does. Behind every inflated filing, someone quiet is doing the slow, real work — and in the end, that is the part that flies.