Starship's Orbital Refueling Test and Artemis

Last Thursday, a NASA safety panel put a number on it. To fuel one Starship for one trip to the Moon, you'd need about twenty-two tanker launches. Twenty-two.

Twenty-two launches to enable one landing? That's absurd.

Elon Musk said the estimate was wrong. His number is six to eight. And on Friday, when NASA officials were asked directly at a press conference to settle it, they declined to give a number at all.

They just... wouldn't say?

They wouldn't say. Because nobody actually knows. And here's the thing that got me, Theo. The reason nobody knows is that the test that would tell us has never been performed. By anyone. In the history of spaceflight.

Hang on. We've been flying rockets for seventy years. Nobody's ever passed fuel from one ship to another?

Not cryogenic propellant. Not ship to ship. NASA's Human Landing System program calls it, quote, one of the most significant technical challenges facing the provider. And every single plan for getting humans to Mars runs through it.

Okay, explain why. Starship reaches orbit. Why can't it just keep going?

Because getting to orbit eats almost everything. Starship arrives in low Earth orbit essentially empty. It has the engines and the structure to go to the Moon or Mars, it just doesn't have the gas. So the plan is a gas station in orbit.

A literal one.

NASA's own description uses that phrase. You launch a series of tanker Starships. They each fill up a depot Starship parked in orbit. The depot tops off the lander, and only then does it leave. The entire Artemis Moon landing architecture is built on it.

So how do you pour liquid oxygen between two spacecraft when there's no down? Doesn't it just float around in blobs?

That's exactly the problem. And the solution is beautifully dumb. You fire the thrusters.

Fire the... what?

You make your own gravity. The two ships dock tail to tail, using the same mating interface and the same propellant lines Starship uses when it's sitting on the booster. Then you gently accelerate. The propellant in the tanker wants to stay put, so it sloshes backward, settles against the wall, and flows into the receiving ship. Musk described it as milli-g settling. A thousandth of Earth gravity is enough.

That's it? That's the whole trick?

That's the principle. The execution is brutal. These fluids are hundreds of degrees below zero, they're boiling constantly, you're managing two-phase flow, liquid and vapor at once, through a transfer line. Suck in vapor instead of liquid and your numbers fall apart.

Has any of this been tested at all?

Once. March fourteenth, twenty twenty-four. Starship Flight Three. Under a NASA Tipping Point contract, SpaceX moved liquid oxygen from a header tank in the nose down to the main tank in the tail. It started twenty-four minutes and thirty-five seconds into the flight and finished at twenty-six minutes and ten seconds.

Under a hundred seconds.

Under a hundred seconds, inside one vehicle. NASA modelled it in advance and confirmed afterward that milli-g was enough to keep the liquid settled. That's genuine progress. But it's one ship, moving fuel between its own tanks.

And the two-ship version? Where is that?

It keeps sliding. Per a NASA Inspector General report from March of this year, the ship-to-ship demo was originally scheduled for March twenty twenty-five. It moved to March twenty twenty-six. As of September, it still hadn't flown. Flight Thirteen went up July twenty-fourth. No rendezvous, no docking.

So where does that leave us right now, in October?

Flight Fourteen flew about three weeks ago. Some Raptor engine issues, but it largely went well. Flight Fifteen is targeting November, and Musk wants to catch BOTH stages on the tower. He says success is more likely than not. Gwynne Shotwell said they'll refly an upper stage early next year, and the refueling demonstration comes early in twenty twenty-seven.

Which means the Artemis landing schedule is waiting on a test that hasn't been scheduled firmly yet.

And look what Blue Origin did. For its twenty twenty-eight landing attempt, Blue deferred propellant transfer entirely. Four New Glenn launches, three transfer stages that drop away like stages of a rocket. No ship-to-ship refueling at all.

They looked at the hard problem and walked around it.

For now. And that's the real answer to twenty-two versus six to eight. The number depends on how much Starship can lift, how efficient the transfer is, and how fast the propellant boils off in orbit. Three unknowns. One flight test resolves all of them.

Early twenty twenty-seven.

NASA Administrator Jared Isaacman said Musk's six-to-eight was possible, but represented the ultimate potential end state of Starship. Which is a polite way of saying: not on day one.

And Mars is much further than the Moon.

Much. Every Mars architecture on the table assumes you can fill a ship in orbit. Until two Starships dock, open a valve, and nudge themselves forward... the Moon is a stretch and Mars is a drawing.