Tranquility · White paper · August 2026
Heat Management: Sintering, and Why Orbit Will Not Work
Every watt of training compute becomes heat. In free space that heat has nowhere useful to go. On the Moon it can be spent making the ground.
Heat is not a facilities footnote. It is the reason Tranquility is a surface program.
A training cluster turns almost all of its electrical power into waste heat. The reactors add more. On Earth you throw that heat into air and water. On the Moon there is no air and no river. You still have to get rid of it, or you throttle, then you stop. That is the kill switch. The question is not whether heat exists. The question is whether you have a place to put it that does work for you.
Why Lagrange and orbit fail
Put the same plant at a Lagrange point, or in free orbit, and you have no ground. You have sunlight, vacuum, and the area of whatever radiator you were willing to fly. Heat can only leave as infrared. The radiator must face cold sky, stay clean, and stay large enough for the load. There is no soil to bury in, no night-side berm, no loose rock you can spend the energy on. There is also no cheap way to grow the dump. Every extra square metre of radiator is launched mass.
That is why “just put the data centre in space” keeps dying in the same place. Power is solvable in principle. Heat rejection at gigawatt scale, with no planetary body, is not a spreadsheet tweak. It is a wall. You can reject kilowatts on a spacecraft. You cannot casually reject the waste heat of a training foundry on a station that has to carry its own sky.
The Moon is different because it is a place. It has a surface, a thermal mass, a night, permanently shadowed cold, and a material that is already there: regolith. Tranquility is not “compute in space.” It is compute on a body that can take the heat.
The first answer was aluminum
The early papers sized deployable radiator panels — heritage from the ISS and JWST — and treated cooling as imported hardware. Fold it, fly it, unfold it, radiate to 3 K. That arithmetic still matters as a backup and as a commissioning tool. It is also the expensive answer. You are paying lift for sheet that does one job.
Those papers also used an optimistic rejection rate. If the radiator skin has to run hundreds of degrees hot to hit a high kilowatts-per-square-metre figure, silicon on the other end of the heat pipe cannot live there. So the aluminum story, taken as the whole story, still has a hole. We did not ignore that. We changed what the heat is for.
Sintering: spend the heat, make the floor
Sintering is not singeing. Singeing burns a surface. Sintering heats loose powder until the grains fuse into a solid — brick, paver, road, pad — without needing water. Lunar regolith does this. Microwave or thermal sintering to roughly 1,000°C will bind it. No Earth cement. No imported aggregate if you accept a rougher product.
That is a heat sink with a purpose. Reactor waste heat, and some of the compute waste heat during build-out, goes into making pavement. You are not only throwing infrared at the sky. You are buying landing pads, roads, berms, and a harder yard. The thing that wrecks seals and coats radiators — electrostatic dust — becomes the floor.
Several problems move at once:
- Heat has a job during construction instead of only a radiator bill.
- Landing plumes have somewhere hard to hit, so less of the site is re-aerosolized.
- Robots get paths instead of tilling powder on every traverse.
- The early aluminum field can be smaller, because you are not asking shipped panels to be the only dump.
This is why the Moon wins and orbit does not. Orbit has nothing to sinter. The liability and the solution are the same material, and that material only exists on a surface.
A Lagrange plant would still need the aluminum field, still need to keep it clean, and still have no construction product for the energy. You would fly mass to reject heat and fly more mass when the dump was short. On the Moon the same energy can leave as infrared and as a road. That is not a slogan. It is the difference between a station that only spends and a yard that keeps a receipt.
What sintering is, and what it is not
Sintering uses heat while you are making pavement. A finished road does not keep eating gigawatts. Once a slab is done, the energy that fused it is already spent. If you stopped there, you would still need a steady dump — radiators, or a sintered surface that is the radiator: a large, sky-facing, high-emissivity plate made of the Moon.
So the operating picture is two layers. First, a build campaign that spends heat expanding pavement, pads, and berms. Second, a standing dump: either remaining panels, or the paved field itself radiating to space, or both. We are not claiming opex of zero and no physics. We are claiming the imported-aluminum plant is the wrong default, and that the surface lets you couple construction to rejection in a way Lagrange cannot copy.
Keep paving and you keep a construction-rate dump. Stop paving and the pavement has to work as a radiator or you are back to panels. That sentence should stay in the decision log. It is the honest limit, and it is still better than a station with no ground.
There is also night, and there are permanently shadowed sinks. A surface can hide a radiator from the sun, couple to cold ground, or work in the dark. An orbital platform is often lit, always mass-limited, and never finished paving. If you remember one contrast, remember that: orbit has a view. The Moon has a job for the heat.
Why this belongs on the masthead
The 2025 cooling notes led with panels and treated sintering as a dust fix on the side. That buried the actual design move. Heat management is not “add radiators.” Heat management is: use the Moon so the waste does work, then radiate what is left. If that coupling fails, the reactors and the lift do not save you. If it holds, you get a yard you could not have afforded to fly as finished decking.
Orbit remains useful for comms and staging. It is a poor place to hide a foundry. The foundry needs a floor, and the floor is how you live with the heat.