A DC Atlas Frontier
Module 3 of 7
The Launch Problem
That radiator, plus power, plus structure, plus compute. All of it multiplied by dollars per kilogram to orbit. Falcon Heavy's demonstrated cost against Starship's aspirational target, kept honestly apart.
Built on published aerospace engineering figures and DC Atlas's facility data.
Once you accept that an orbital data centre is a radiator field with a power plant attached (Modules 2 and 4), the launch question answers most of the business case on its own. Everything has to go up. The radiators, the solar arrays, the batteries, the structure that holds them rigid, the compute, the propellant to keep the whole thing from falling out of the sky — every kilogram is lifted from the ground at a price per kilogram. So the entire orbital thesis reduces to a single sensitivity: what is the dollar cost of a kilogram to low Earth orbit, and can it fall far enough, fast enough, to matter.
Start with what is real, because the pitch tends to start with what is hoped. The cheapest launch anyone has actually flown to LEO is SpaceX's Falcon Heavy, at roughly $1,400 to $1,500 per kilogram for a dedicated mission carrying up to 63,800 kilograms. Falcon 9, the workhorse, is about $2,700 to $3,000 per kilogram. Every other operational vehicle — Europe's Ariane, China's Long March family, the retired US heavy-lifters — costs far more, anywhere from $4,000 to $40,000 per kilogram. Those are the numbers you can build a plan on today, and at those numbers the orbital data centre is not close to viable.
Cost to launch one kilogram to low Earth orbit
Falcon 9 and Falcon Heavy are demonstrated dedicated-mission prices; Starship is SpaceX's stated target and remains unproven. The Lumen Orbit figure is the proponents' own white-paper assumption, shown for contrast — not a market price.
So the whole thing is a bet on Starship. SpaceX has said it is targeting around $200 per kilogram, a roughly 93 percent cut from Falcon 9, and the more aggressive orbital-data-centre models assume $50 to $100 per kilogram — Lumen Orbit's white paper pencils in a 100-tonne launch for $5 million, which is $50/kg. That price may well come; the long-run trend genuinely points that way, and a fully reusable vehicle flying often is exactly how you would get there. But it is not a fact, it is a forecast, and it rests on flight rates and rapid, full reusability that no vehicle has yet demonstrated at scale. Building a data-centre business on it today is building on a number that does not exist. The honest way to argue about orbit is to keep the demonstrated price and the aspirational price in separate columns and never let the second quietly stand in for the first.
| System | Where it comes from | Mass to orbit | Falcon Heavy launches (≈63 t each) |
|---|---|---|---|
| Radiators | Module 2 (heat) | ~360–1,500 t | ~6–24 |
| Solar array | Module 4 (power) | hundreds–~1,000+ t | ~10–20 |
| Batteries | Module 4 (eclipse) | ~180–250 t | ~3–4 |
| Structure, compute, shielding | not yet counted | additional | additional |
| Total (power + cooling only) | Modules 2 + 4 | ~1,500–3,000+ t | ~24–48 flights |
There is a genuine trend on the proponents' side, and it deserves full acknowledgement rather than a grudging one: launch cost has fallen by more than an order of magnitude in fifteen years, from Space Shuttle prices near $54,000 per kilogram to Falcon Heavy's $1,500, and a fully reusable Starship flying frequently could break the curve again. That is the single most important variable in the whole orbital case, and it is moving in the proponents' direction. If it reaches and holds $100/kg, the arithmetic in this module goes from absurd to merely bad.
But bad is not viable, because launch is only the first invoice, and it is the one most likely to improve. The mass you lift still has to make its own power through the eclipse, still has to survive radiation with no one to repair it, and still has to be replaced on a one-to-three-year cadence it cannot meet in orbit — the subjects of the next two modules — and none of those costs fall when the launch price does. A cheaper rocket makes it cheaper to put a doomed data centre in the sky. It does not make the data centre work.
Falcon Heavy — payload capacity and pricingQuestions this module answers
- What does it cost to launch to orbit today?
- About $1,500 per kilogram on Falcon Heavy, the cheapest demonstrated ride. SpaceX targets roughly $200 per kilogram with Starship, but that is unproven.
- What is the launch bill for an orbital data centre?
- A 40 MW facility must launch roughly 1,500 to 3,000 tonnes of radiators, solar array and batteries: $2.3 to $4.5 billion at Falcon Heavy prices, more than the cost of an entire terrestrial data centre.
- Does cheaper launch fix the problem?
- It helps, but launch is the first and most improvable cost. The heat, power and hardware-replacement problems do not fall when the launch price does.