Lunar Base & Habitation

The buildout sequence

Four Phases: Robots First, Crew Last

The base is not built by astronauts. Astronauts arrive at a base that robots have already built.

The conceptual error in most lunar-base discussions is imagining astronauts stepping off a lander onto raw terrain and building a habitat by hand. That is not how this works. The buildout is sequenced so that the hardest, dirtiest, most dangerous construction happens before any crew arrives. Robots excavate, grade, trench, deploy, and test. Crew arrives when the power is on, the habitat is pressurized, the communications are verified, and the life-support consumables are pre-positioned. The base is an industrial construction project. The crew is the occupancy permit.

NASA’s Moon Base plan, announced May 26, 2026, formalizes this in three phases. The plan here adds a Phase 0 (pathfinder) and subdivides Phase 2 into crew-robotic MVP and specialist deployment. The table below maps both.

NASA’s official phases are 1 (robotic through 2029), 2 (semi-permanent ~2029), and 3 (continuous 2032+). The RTS decomposition adds Phase 0 and subdivides Phase 2 into the MVP crew-check and the specialist-deployment stage. Building

Phase 0: Pathfinder

Moon Base I, II, and III: Robotic Scouts at Shackleton

Before you build anything, you characterize the ground it sits on.

NASA announced the first three Moon Base missions on May 26, 2026. Moon Base I (NET fall 2026) uses a Blue Origin Blue Moon MK1 Endurance lander to deliver NASA payloads to the south pole near Shackleton crater — its primary objective is measuring how rocket exhaust disturbs the regolith, because every subsequent landing will face the same ejecta problem. Moon Base II (late 2026) delivers more than 500 kg of cargo on an Astrobotic Griffin lander, including Astrolab’s FLIP rover, to demonstrate mobility and cargo handling. Moon Base III (also 2026) flies the first PRISM (Payloads and Research Investigations on the Surface of the Moon) payload on an Intuitive Machines Nova-C Trinity lander, including ESA and Korean instruments.

In parallel, VIPER (late 2027 on Blue Moon MK1) drills into the PSR with its TRIDENT 1-meter auger — the single highest-value measurement for closing the water-extraction case. Chang’e-7 (2027) lands a rover and hopper on Shackleton’s SE ridge to perform the first Chinese in-situ volatile measurement. Phase 0 produces the data that Phase 1 builds on. Without it, every construction decision is an assumption. Building

Phase 1: Cargo-robotic base prep

Heavy Cargo Delivery and Robotic Construction: The Base Before the Crew

Cargo Starships land 100+ tonnes at a time. Nothing returns. Everything stays.

Phase 1 is the construction phase — entirely uncrewed. Cargo Starships deliver the heavy hardware: excavation equipment, Optimus humanoid robots, vertical solar arrays, cable reels, habitat pressure vessels, ISRU demonstration hardware, regolith-processing plants, and pre-positioned crew consumables (food, water, medical supplies, spare parts). Each cargo Starship lands with 100,000–150,000+ kg and stays permanently — the vehicle itself becomes a structural shell.

The construction sequence, executed by autonomous and teleoperated robots before any crew arrives:

Landing pads. The first pad is unprepared — the first Starship lands on raw regolith and its exhaust sculpts a crude depression. Every subsequent landing needs a bermed, compacted, or sintered pad to prevent ejecta from damaging existing equipment. Ejecta from an unbermed Starship landing can travel nearly 100 km at 500 m/s. Pad construction is the first earthmoving task.

Access roads and graded corridors. Autonomous vehicles grade the rim-to-floor corridor, create traversable slopes, and clear boulder fields. The same teach-and-repeat autonomy demonstrated by the CSA cargo rover runs the grading equipment.

Cable trenching and utility corridors. High-voltage DC power cables from the VSAT grid on the rim to the work zone on the inner wall. Communications fiber or waveguide from the habitat to the relay antenna. Propellant and water transfer lines from the extraction site to the processing plant. All trenched and anchored to protect from thermal cycling and micrometeorite impact.

Power grid deployment. Vertical solar arrays erected on the illuminated ridge. LunaGrid nodes connected by 2-km cables. Regenerative fuel cells pre-charged. The grid is powered and tested before any habitat is pressurized.

Habitat preparation. Pressure vessels unloaded from cargo Starships and positioned. Regolith overburden piled or sintered over them for radiation shielding. ECLSS (Environmental Control and Life Support System) activated and run for weeks to verify atmosphere, thermal control, and water recycling before crew arrival.

End-of-phase test: the base is powered, pressurized, communicating, and pre-stocked. A crew can land and move in. Projected

Phase 2: Crew-robotic MVP

First Crew Arrival: Validation, Activation, and Initial ISRU

The crew’s job is not to build the base. It is to verify that the robots built it correctly.

Artemis IV (2028) delivers the first crew to the south pole — 2–4 astronauts on a surface stay of roughly six days. They land on a prepared pad, walk into a pressurized habitat that robots set up months earlier, and spend their time validating systems: testing ECLSS performance under crew load, verifying power delivery, driving the LTV rovers, and conducting initial ISRU demonstrations (oxygen extraction from regolith).

Artemis V (late 2028 or later) targets yearly landings and begins the transition to sustained presence. Specialist crew rotations follow: reactor technicians to commission the fission surface power system, ISRU engineers to scale the water-extraction plant, geologists to direct prospecting, and robotics engineers to maintain and reprogram the autonomous workforce. Each rotation is short — days to weeks, not months — until the habitat, life support, and logistics chain are proven for longer stays. Building

Phase 3: Sustained operations

Continuous Presence: Rotating Crews, Industrial ISRU, and the Operational Analogue

Permanent as a system. Temporary as a population.

Phase 3 (NASA target: 2032+) is continuous human presence — not a permanent settlement in the colonial sense, but a continuously occupied outpost with rotating crews, like Amundsen-Scott South Pole Station. The operational analogue matters: the station is staffed year-round, crews rotate on a schedule, handover costs are measured in lost productive hours, and the system is designed for the institution rather than the individual.

At this phase, the industrial power demand (~490 kW) begins to be met by a combination of VSAT solar grid and fission reactor. The ISRU plant operates at scale. The mass driver enters preliminary construction. The Lunar Cruiser (targeted Artemis VII, ~2032) extends the operational radius to hundreds of kilometers. The base transitions from a research outpost that consumes imported supplies to an industrial node that begins to export. Projected

Cargo Starship as infrastructure

Unpack, Topple, Deconstruct: How a Landed Starship Becomes a Building

1,000 m³ of pressurized volume, 9 meters in diameter, 50 meters tall. It is already a building. The question is how to use it as one.

A cargo Starship that lands permanently is the largest single structure on the lunar surface. NASA’s planning already envisions using early Starship HLS vehicles as initial surface habitats. But a 50-meter vertical cylinder on the surface of a body with 0.165 g presents a center-of-gravity problem — it is tall, narrow, and top-heavy by terrestrial standards.

Three architectural approaches have been discussed for converting a landed cargo Starship into usable infrastructure:

Use it vertical. Access via elevator or external stairway/ramp structure. Internal decks at multiple levels. This is the simplest approach and avoids any structural modification. The drawback is that crew and cargo must move vertically through a 50-meter column, and external EVA access requires climbing or a lift system.

Topple it. After unloading cargo, the vehicle is deliberately toppled onto prepared regolith berms, placing it horizontal. This converts it into a long, low-profile structure with walk-in access along its length. The structural risk is the topple event itself — a controlled fall of a roughly 100-tonne empty vehicle onto prepared terrain, in 0.165 g. The vehicle is not designed for this, and internal fittings would need to survive the impact.

Cut and cover. The hull is cut into sections using thermal or mechanical tools, and the sections are placed in excavated trenches, then covered with regolith for radiation shielding. This produces the most conventional habitat geometry — horizontal modules buried under several meters of regolith — but requires the most construction effort and destroys the vehicle’s pressure integrity, meaning each cut section must be re-sealed.

All three approaches produce usable volume at a fraction of the cost of purpose-built habitat modules launched from Earth. The cheapest habitat on the Moon is the vehicle that delivered the cargo. Projected

Earthmoving and construction

Excavation, Grading, Trenching, and Sintered Construction on the Lunar Surface

Every flat surface, every berm, every trench, and every shielding layer is built from scratch. There is no prepared ground.

Lunar earthmoving is not terrestrial earthmoving. At 0.165 g, excavation reaction forces must come from equipment mass or anchoring rather than from weight. A bulldozer blade that works by pressing down against its own weight on Earth presses with one-sixth that force on the Moon. Either the machine is heavier (more mass to deliver), or it anchors to the surface (more complexity), or it uses different excavation methods (bucket drums, screw augers, pneumatic ejection).

NASA’s IPEx (ISRU Pilot Excavator) demonstrated bucket-drum excavation of 10 tonnes of simulant in five days. ICON’s Olympus system ($57.2M, renewed August 2026) is designed to 3D-print landing pads, blast walls, roads, and habitat shells from regolith using directed energy. Chang’e-8 (~2029) will test autonomous regolith-based 3D-printed construction as an ILRS foundation element.

The construction sequence for a base site: clear and grade the landing pad → berm the pad perimeter to contain ejecta → grade access roads from pad to habitat site → excavate habitat foundation trenches → trench utility corridors (power cable, water/gas transfer, communications) → position and bury habitat modules under regolith overburden → sinter or print blast walls, thermal shields, and walkway surfaces. Every step is performed by autonomous or teleoperated equipment before crew arrival. Projected

The autonomous workforce

Optimus, Teleoperation, and the Lunar Advantage Over Mars

Robots just need electricity, solar power, and a little lubrication in the joints. They do not need air, water, food, or sleep.

The lunar base construction workforce is primarily robotic. Tesla’s Optimus humanoid robot entered pilot production in early 2025, with full-scale production beginning in 2026 — Tesla is converting Fremont Model S/Model X production lines to Optimus manufacturing. SpaceX’s February 2026 lunar pivot explicitly named Optimus as the construction workforce for the lunar factory. Jim Cantrell (who worked with Musk from SpaceX’s founding) stated publicly that the plan is Optimus robots handling the vast majority of the heavy lifting, with Starship autonomously depositing cargo. Building

The Moon has a decisive advantage over Mars for robotic teleoperation: 2.56 seconds of round-trip latency versus 8–44 minutes. A teleoperator on Earth controlling an Optimus on the Moon experiences a noticeable delay but can still perform complex manipulation tasks — roughly equivalent to operating through a sluggish video link. The same operator controlling a robot on Mars must send a command and wait 8–44 minutes for feedback, making real-time teleoperation impossible. Mars robots must be fully autonomous. Moon robots can be teleoperated from Earth as a fallback for any task the autonomy system cannot handle.

The robotic workforce includes at least three classes: humanoid general-purpose (Optimus — manipulation, tool use, inspection, cargo handling), specialized excavation (bucket-drum excavators, grading blades, auger drills — task-specific machines designed for regolith processing), and autonomous cargo crawlers (teach-and-repeat transport between fixed points). The mix evolves as the base matures — Phase 1 is dominated by excavation and construction equipment, Phase 3 by general-purpose humanoids performing maintenance, inspection, and repair alongside crew. Projected

Habitat systems and ECLSS

Pressure, Atmosphere, Thermal Control, Water, and Waste: Keeping Humans Alive

The habitat is not the building. The habitat is the air, the water, the temperature, and the radiation shielding inside the building.

A lunar habitat must provide six functions simultaneously: structural pressure containment (maintaining ~101 kPa of internal atmosphere against hard vacuum), breathable atmosphere (21% O₂, 79% N₂, CO₂ scrubbing, humidity control), thermal control (rejecting metabolic and equipment heat to the vacuum via radiators), water recycling (closed-loop recovery from humidity, urine, and greywater), waste processing, and radiation shielding (2–5 meters of regolith overburden or equivalent mass).

ISS heritage provides the ECLSS baseline, but the Moon adds constraints ISS does not face: no resupply in three hours (the ISS can receive a Progress cargo ship on a same-day launch), dust ingress at every airlock cycle, and no structural protection from moonquakes that ring for an hour. Suitports — rear-entry suit docking systems that keep the suit exterior outside the habitat — are the leading mitigation for the dust problem, because they eliminate the airlock dust-transfer step entirely.

Nitrogen is the critical consumable. Oxygen can be produced locally from regolith. Water can be recycled at >90% recovery rates. But nitrogen exists on the Moon only at parts-per-million concentrations — it is a permanent import. Every airlock cycle loses some nitrogen to the vacuum. Over years of operation, nitrogen resupply becomes a significant logistics cost. Projected

Orbital infrastructure

Communications Relay, Navigation, and Cislunar Depot

The base cannot function without its orbital infrastructure. No relay, no teleoperation. No depot, no affordable resupply.

A south-polar surface base has limited direct Earth visibility — Earth sits near the horizon and is periodically blocked by terrain. Continuous communications require a relay satellite. China’s Queqiao-2 (operational since March 2024) provides relay coverage for ILRS missions. NASA currently plans to rely on direct-to-Earth links supplemented by commercial relay services — Gateway was shelved in March 2026, removing the planned Artemis relay platform.

Navigation on the surface requires a positioning reference. No lunar GPS exists. ESA and NASA have studied lunar navigation satellite constellations (Moonlight), and several commercial providers have proposed lunar PNT (positioning, navigation, and timing) services. Until a constellation is deployed, surface positioning relies on terrain-relative navigation from LRO maps and inertial measurement — adequate for rovers but insufficient for precision landing at prepared pads without supplementary beacons.

A cislunar propellant depot in NRHO or at EML-1 changes the logistics economics by allowing smaller, more frequent resupply flights rather than full Starship-class cargo missions for every delivery. The depot is native to Cislunar, not to the Moon — the Moon’s contribution is water exported upward to fill it. Projected

Repair and maintenance

The Unsolved Problem: Keeping Everything Running at 384,400 km

Every machine on the lunar surface degrades. Dust, thermal cycling, radiation, and micrometeorites never stop.

Maintenance on the Moon is not like maintenance on Earth or even on the ISS. There is no supply chain behind the base — every spare part, every tool, every consumable was either launched from Earth at enormous cost or fabricated locally from limited materials. A pump seal that fails at 2 AM on an Antarctic research station can be replaced by the winter-over mechanic from the on-site spares inventory. A pump seal that fails on the Moon requires either a pre-positioned spare (which took months to deliver) or a local fabrication capability that does not yet exist.

The degradation agents are relentless. Dust abrades seals, coats optics, and embeds in bearings. Thermal cycling fatigues joints and coatings. Radiation degrades electronics and polymers. Micrometeorites puncture thin structures. Every system on the base has a maintenance clock running from the moment it is deployed — and the cost of ignoring that clock is not a repair bill but a mission-critical failure with no backup.

The autonomous workforce partially addresses this. Optimus robots can perform inspection, cleaning, and simple replacement tasks under teleoperator guidance. But complex repairs — welding a pressure-vessel crack, replacing a reactor heat exchanger, debugging a software fault in the ECLSS controller — require human judgment, dexterity, and improvisation that current robotics cannot match. Maintenance is the reason the base needs crew, not just robots. The crew’s primary job in sustained operations is not science or exploration. It is keeping the machines running. Projected

Record

Provenance and Data Sources