The gate
One Unflown Maneuver Holds Up the Moon and Mars
Starship trades a large upper stage for a small payload fraction to orbit. The trade only pays off if the tanks can be refilled in space.
Starship lifts about 100 t to low Earth orbit and arrives with essentially dry main tanks. A trip to the Moon or Mars needs those tanks refilled on orbit, which means a fleet of tanker flights, a depot to hold what they deliver, and a way to move cryogenic methane and oxygen from one vehicle to another in microgravity. NASA calls large-scale ship-to-ship cryogenic transfer one of the most significant technical challenges facing the provider, and it has not yet been done.
This is why the HLS lander, the cislunar architecture and the Mars cargo plan all wait on the same milestone. A partial internal transfer flew on Flight 3 in 2024 — propellant moved between the ship’s own header and main tanks — but the vehicle-to-vehicle demonstration is the one that gates the program. Building
Tanker & depot
Two Purpose-Built Ships Whose Payload Is Propellant
Both are Block 3 ship derivatives. Neither carries a conventional payload.
| Vehicle | Role | Detail | Rated |
|---|---|---|---|
| Tanker | Carries propellant up and transfers it | Flies as the “chaser” in the demo. More than 10 tanker flights fill one depot for a lunar mission, launched in quick succession to limit boil-off | Building |
| Depot | Stores aggregated propellant in LEO | Flies as the “target.” No heat shield, flaps or tower-catch hardware — it stays in orbit, so the mass those systems cost is saved. Insulated for long-term cryogenic storage | Building |
| HLS & deep-space ships | The vehicles that get refueled | Fill from the depot in LEO, then depart. The depot is filled starting more than 200 days before crew launch | Projected |
Omitting the heat shield, flaps and catch hardware on the depot is the same weight-saving logic that shapes the HLS lander: a vehicle that never re-enters Earth’s atmosphere does not carry the mass survival there demands.
The transfer demo
Two Ships, One Dock, a Few Tonnes of Oxygen
The demonstration is deliberately small: prove the physics before scaling to a full depot fill.
The plan launches two Starships a few weeks apart. They rendezvous and dock in orbit, and the tanker transfers cryogenic propellant — on the order of several tonnes of liquid oxygen — to the target. Success validates the docking interface, the transfer plumbing and, hardest of all, the fluid behavior: slosh, boil-off and how ultra-cold liquid moves between tanks without gravity to settle it. The Block 3 ship was built for this, adding the docking ports and redesigned quick disconnect the earlier blocks lacked.
Aggregation math
A Lunar Landing Is a Tanker Campaign, Not a Launch
The depot has to be filled faster than it leaks, which sets a punishing flight rate.
| Quantity | Value | Note | Rated |
|---|---|---|---|
| Tanker flights | 10–16, “high teens” per NASA | Each delivers a partial load to the depot | Building |
| Aggregation lead time | 200+ days before crew launch | The depot fills over months, then holds until the crewed vehicle is ready | Building |
| Constraint | Cryogenic boil-off | Methane and oxygen warm and vent over time, so tankers must fly in quick succession — the flight-rate driver | Observed |
The number few outside the program register: a single crewed lunar landing needs a tanker campaign in the high teens, flown fast enough to outrun boil-off. That folds the whole program back onto flight rate and the Raptor production line — refueling does not just need the transfer to work, it needs the factory and pads to sustain a burst of launches per mission.
The slip record
The Date Has Moved Every Year Since 2024
The demonstration is the most-delayed milestone in the program, and the delays cascade into Artemis.
| As of | Target | Rated |
|---|---|---|
| 2024 planning | March 2025 | Observed |
| Early 2026 | March 2026 | Observed |
| September 2026 | Unscheduled — “this year” per SpaceX | Building |
| Downstream | Artemis III recast to a 2027 LEO docking test; first crewed landing to Artemis IV, ~2028 | Projected |
Contested
Deliberately Not Claimed
The load-bearing uncertainties, tagged with the confidence they actually earn.
Tanker count per mission is a range, not a figure. NASA and SpaceX have cited numbers from 10 to the high teens depending on vehicle block and mission. Block 4’s larger tanks would lower it. Treat any single number as provisional until a depot is actually filled.
Boil-off rates in a real depot are unmeasured. The 200-day aggregation window assumes storage losses that have not been demonstrated at depot scale. The demo transfers a few tonnes; a full depot hold is a different regime.
The whole beyond-LEO manifest hangs on this one item. HLS, crewed lunar and Mars cargo all assume refueling works. This page treats the demonstration as the single highest-leverage unknown in the Starship program.
Record
Provenance and Data Sources
- Confidence
- Building throughout — no vehicle-to-vehicle transfer has flown
- Status date
- Q3 2026
- Architecture
- NASA HLS program statements; NASA Office of Inspector General report IG-26-004 (March 2026)
- Demo plan
- SpaceX and NASA concept-of-operations infographics; Flight 3 internal-transfer result
- Weakest link
- The demonstration itself, unscheduled as of this date