Lunar Transportation Architecture

The vehicles

Five Vehicles, Two Roles: Crew Ferry vs. Cargo Lander

A crew vehicle returns from the surface. A cargo vehicle stays and becomes infrastructure.

Every lunar mission architecture assembles a stack of vehicles, each performing one leg of the journey. The distinction that matters most is not which company builds the vehicle — it is whether the vehicle returns or stays. Crew ferries ascend from the surface and must carry ascent propellant (subject to boiloff, subject to refueling questions). Cargo landers land heavy and stay permanently — they become habitats, warehouses, power stations, or scrap. No return propellant, no human rating, no ascent stage.

Stack A

Crew Mission: SLS/Orion + Starship HLS (Artemis IV Era)

~13–19 Earth launches per crew mission. The tanker campaign is the mission.

Step 1 — Build the propellant supply in LEO. Launch a Starship propellant depot to LEO. Then launch 10–16 Starship tankers, each docking with the depot and transferring methalox. Every tanker’s Super Heavy booster returns to the launch site (caught or landed). The tanker ship reenters and is reused. The depot stays in orbit.

Step 2 — Send Starship HLS to the Moon. Launch Starship HLS (uncrewed) to LEO. It docks with the full depot, takes on a complete propellant load. It then performs trans-lunar injection (TLI), coasts to the Moon, performs lunar orbit insertion (LOI), and enters NRHO. It is now waiting in lunar orbit with descent and ascent propellant aboard.

Step 3 — Send the crew. Launch SLS with Orion from Kennedy Space Center. Orion carries 2–4 crew to NRHO (~3–5 days transit). SLS is expendable — the core stage and boosters are not recovered.

Step 4 — Surface mission. Crew transfers from Orion to Starship HLS in NRHO. Starship HLS descends to the lunar surface (powered descent using high-mounted RCS thrusters for final approach to avoid regolith excavation). Crew works on the surface for days to weeks.

Step 5 — Return. Starship HLS ascends from the surface back to NRHO. Crew transfers from Starship HLS back to Orion. Orion flies the crew home — trans-Earth injection, reentry, splashdown in the Pacific. Recovery ships retrieve the capsule.

Step 6 — What stays. Starship HLS remains in NRHO. On early missions (Artemis IV, V) it is effectively expendable — parked or eventually disposed. On later missions, it waits in NRHO for the next crew arrival, but it needs propellant for its next descent, which means either a new tanker campaign sent all the way to NRHO, a return to LEO for refueling (requiring propellant it doesn’t have), or lunar-surface ISRU. SLS is destroyed. Orion capsule is recovered but not reused for lunar missions.

Launch count per mission: 1 depot + 10–16 tankers + 1 Starship HLS + 1 SLS = 13–19 launches. After the first mission, the depot may remain in LEO, reducing the count by one. Building

Stack B

Crew Mission: SLS/Orion + Blue Moon MK2 (Artemis V Era)

2 Earth launches per crew mission. No tanker chain. But the lander needs propellant from somewhere.

Step 1 — Send the lander to the Moon. Launch New Glenn from Cape Canaveral with Blue Moon MK2 and the Lockheed Martin Cislunar Transporter tug. New Glenn’s first stage lands on a drone ship in the Atlantic (reusable). The second stage is expended. The Cislunar Transporter delivers MK2 from Earth orbit to NRHO — New Glenn is an Earth-to-orbit vehicle only and does not go to the Moon.

Step 2 — Send the crew. Launch SLS with Orion. Crew flies to NRHO. Orion docks with Blue Moon MK2.

Step 3 — Surface mission. Crew (2–4) transfers from Orion to MK2. MK2 descends to the surface on three BE-7 engines burning LOX/LH₂. Surface stay up to 30 days.

Step 4 — Return. MK2 ascends to NRHO. Crew transfers back to Orion. Orion flies the crew home.

Step 5 — What stays. Blue Moon MK2 remains in NRHO, designed for reuse. The Cislunar Transporter tug stays in cislunar space and can reposition MK2 or deliver propellant between orbits. For MK2’s next mission, it needs descent propellant. Two paths: the Cislunar Transporter delivers propellant from an Earth-orbit or cislunar depot, or MK2 refuels on the lunar surface with ISRU-produced LOX/LH₂ (both are products of water electrolysis — MK2’s propellant choice is not an accident). New Glenn first stage is reused. Second stage is destroyed. SLS is destroyed.

Launch count per mission: 1 New Glenn + 1 SLS = 2 launches. If MK2 is already in NRHO from a previous mission, only the SLS crew launch is new — plus whatever delivers propellant to MK2. Building

Stack C

Cargo Mission: Starship Cargo Direct to the Surface (Uncrewed)

100+ tonnes to the surface. No crew, no return, no human rating. The vehicle becomes the base.

Step 1 — Tanker campaign. Same as Stack A: depot + 10–16 tanker flights to LEO.

Step 2 — Send the cargo Starship. Launch a cargo-variant Starship (not human-rated — no life support, no crew cabin, no emergency systems). It docks with the depot, fills its tanks, performs TLI, LOI, and powered descent to the lunar surface. It carries 100,000–150,000+ kg of cargo: construction equipment, reactor components, ISRU hardware, solar arrays, Optimus robots, habitat modules, regolith-processing plants.

Step 3 — What stays. Everything. The cargo Starship lands permanently on the lunar surface. It does not ascend. It has no ascent propellant and no need for any. The vehicle itself — 1,000 m³ of pressurized volume, 9 m diameter — becomes a habitat, warehouse, or structural shell. NASA’s planning already envisions using early Starship vehicles as initial surface habitats. A cargo variant designated by NASA as the Human-class Delivery Lander is expected in service by Artemis VII.

Launch count per mission: 1 depot (if not already in LEO) + 10–16 tankers + 1 cargo Starship = 12–18 launches. No SLS, no Orion, no crew. Building

Stack D

Future Crew Mission: Starship Direct — Earth to Surface and Back

The logical endpoint. No SLS, no Orion, no vehicle transfer in orbit. One vehicle does the whole mission.

Step 1 — Tanker campaign. Same depot + tanker chain in LEO.

Step 2 — The whole mission in one vehicle. A human-rated Starship launches crew from Earth, docks with the depot, fills its tanks, performs TLI, LOI, and powered descent to the lunar surface. The crew works on the surface. When the mission ends, Starship ascends from the surface, performs trans-Earth injection, reenters Earth’s atmosphere, and lands — returning the crew directly.

Step 3 — What stays. Nothing. The vehicle is fully reusable. After refurbishment on Earth, it flies again.

Why this doesn’t exist yet. Starship has not demonstrated human-rated Earth reentry and landing from a lunar-return trajectory. The thermal protection, life support, and abort-mode certification required for crew reentry are beyond what cargo or HLS variants need. The institutional constraints are equally real: SLS and Orion are congressionally mandated for crew flights in the current Artemis authorization. NASA’s February 2026 announcement that SLS will be replaced by commercial vehicles after Artemis V points toward this architecture, but no timeline has been set.

ISRU becomes most important here. In this architecture, the vehicle needs propellant for the return trip to Earth — not just ascent to NRHO. Methalox is roughly 78% oxygen by mass. Surface-produced LOX replacing even the oxidizer fraction of the return load cuts the LEO tanker campaign by a large fraction. The architecture that looks like it eliminates the lunar propellant customer is the one that needs it most. Projected

Launch count per mission: 1 depot + 10–16 tankers + 1 Starship = 12–18 launches without ISRU. With lunar LOX for the return, the tanker count drops significantly. No SLS. No Orion. No vehicle transfer.

Stack E

China Crewed Mission: Long March 10 + Mengzhou + Lanyue

Two launches, no tanker chain, no orbital refueling. Smaller payloads, simpler logistics.

Launch 1. Long March 10 launches the Lanyue lander to lunar orbit. The vehicle is expendable.

Launch 2. A second Long March 10 launches the Mengzhou crew spacecraft with astronauts. Mengzhou enters lunar orbit and rendezvous with Lanyue.

Surface mission. Crew transfers from Mengzhou to Lanyue. Lanyue descends to the surface. Short-duration surface stay. Lanyue’s ascent stage returns crew to lunar orbit. The descent stage stays on the surface permanently.

Return. Crew transfers from Lanyue ascent stage back to Mengzhou. Mengzhou flies the crew home — trans-Earth injection, reentry, capsule recovery. The Lanyue ascent stage is disposed in lunar orbit.

What stays. Lanyue descent stage on the surface. Lanyue ascent stage in lunar orbit (disposed). Both Long March 10 vehicles expended. Mengzhou capsule recovered.

Launch count per mission: 2 launches. No tanker chain, no orbital refueling, no depot. The trade-off is payload — Long March 10 delivers roughly 27 tonnes to TLI, versus Starship’s 100+ tonnes to the surface after refueling. China’s architecture is logistically simpler but carries far less to the Moon per mission. It is institutionally separate from the ILRS robotic program (Chang’e-7, Chang’e-8). Target: first crewed landing before 2030. Building

The tanker chain

Orbital Refueling: The Hardest Unsolved Step

Stacks A, C, and D do not work without it. Stack B and E avoid it entirely.

Every Starship-based lunar mission — crew or cargo — requires orbital refueling in LEO. Starship reaches orbit with its tanks nearly empty because its 275-tonne dry mass consumes the propellant just getting there. The mission profile: launch a propellant depot, fly 10–16 tanker Starships to fill it, then launch the mission Starship to dock and fuel up.

Estimates of the tanker count vary: SpaceX has suggested 4–8 (assuming partial fill or higher tanker capacity on the V3 vehicle with ~200-tonne propellant loads); NASA’s GAO and OIG have estimated 11–19. The discrepancy rests on assumptions about actual payload-to-orbit performance, transfer efficiency, and boiloff during the depot loitering period while waiting for the full campaign to complete.

Depot boiloff. The depot itself is a cryogenic vessel parked in LEO — in full sun for roughly half of each 90-minute orbit — absorbing heat while waiting for the tanker campaign to finish. If 16 tanker launches at a six-day turnaround take roughly three months to complete, the depot is losing propellant the entire time. The first tanker loads are boiling off while the last tankers have not yet launched. This is not a rounding error. At typical LEO boiloff rates for passively insulated methalox, losses of 0.1–0.5% per day compound over weeks into tonnes of propellant that were launched, transferred, and then lost to heat before the mission Starship ever docks. The depot does not hold what you put into it. It holds what you put in minus what boiled off while you were still filling it. Active cryo-cooling (zero-boiloff technology using cryocoolers) can reduce losses dramatically but adds mass, power demand, and complexity to the depot — and has not been demonstrated at this propellant volume in LEO. SpaceX’s depot variant omits flaps and heat shield to maximize insulated tank volume, but passive insulation alone does not solve the problem at campaign timescales of months. Building

Compound probability. If each tanker launch plus transfer has 98% success, five flights yield 90% mission success; 16 flights yield 72%. Every additional flight multiplies the probability of a delay that cascades into the crew timeline. A March 2026 NASA OIG report called large-scale vehicle-to-vehicle cryogenic transfer one of the most significant technical challenges facing HLS.

Status as of August 2026: SpaceX internally transferred roughly five tonnes of liquid oxygen between tanks on the same vehicle during the third test flight. Ship-to-ship transfer between two separate Starships — the actual operational requirement — has not yet been demonstrated. The tanker variant hardware has been spotted at Starbase but has not flown. Building

Boiloff

Surface Boiloff: The Tax on Every Lunar Stay

The depot loses propellant while being filled. The lander loses propellant while the crew works. Both ends of the chain leak.

Any vehicle running on cryogenic propellants that parks on the sunlit lunar surface is losing propellant continuously. LOX boils at 90 K. Liquid methane boils at 112 K. Liquid hydrogen boils at 20 K. The illuminated south-polar rim sits at 170–220 K under near-continuous sun. Over a multi-week surface stay, the ascent or return propellant reserve absorbs heat and boils off.

Starship HLS (methalox, Stacks A and D) is affected for any stay longer than a few days. Blue Moon MK2 (LOX/LH₂, Stack B) is affected more severely because liquid hydrogen has the lowest boiling point and the highest boiloff rate of any common propellant — but MK2’s four large thermal panels are designed specifically to manage this. Cargo Starship (Stack C) has no ascent propellant and is unaffected.

Every kilogram of propellant lost to boiloff is a kilogram that was launched from Earth through the tanker chain at enormous cost. The vehicle can carry margin, but margin means more tanker flights. A local top-off capability — even modest lunar-oxygen production to replace boiled-off LOX — creates ISRU demand in any architecture where a crewed vehicle sits on the surface. Projected

Where ISRU enters

Propellant Demand by Architecture: Where the Water Mine Matters

Every stack creates propellant demand somewhere. The question is whether it comes from Earth or from the Moon.

The architectural irony is clear. The simplest near-term stacks (A, E) need ISRU least. The architectures that scale to sustained operations (B, D) need it most. Blue Moon MK2’s LOX/LH₂ propellant choice aligns with water electrolysis by design. Starship’s methalox requires both oxygen (available from regolith or ice) and methane (requires carbon, which the Moon has only at parts-per-million) — making full methalox ISRU on the Moon far harder than LOX-only ISRU. The likeliest near-term ISRU contribution to a Starship architecture is LOX top-off only, with methane shipped from Earth. See the Lunar Polar Water page for the full extraction economics. Projected

Record

Provenance and Data Sources