Cislunar Stations

The no-rescue architecture

Designing for Unreachability: The No-Rescue Constraint

LEO has an ambulance. Cislunar space does not.

A crew aboard the ISS or a commercial LEO station can be home in hours. A Soyuz or Crew Dragon deorbits, enters, and lands on the same continent in under four hours from undocking. That timeline shapes everything: medical emergencies can be evacuated, consumables can be resupplied on short notice, and critical failures can be resolved by getting the crew off the vehicle.

No equivalent exists beyond LEO. A crew at EML1, EML2, or in NRHO is days from Earth by any return trajectory. Orion cannot simply descend into low lunar orbit and climb back out — the propulsion budget does not allow it. That constraint was one reason NASA’s Gateway architecture existed in the first place: the station would sit in an orbit Orion could reach directly, avoiding the need for a separate lunar-orbit-insertion vehicle.

A crewed cislunar facility inherits the full no-rescue architecture:

Consumables must be sized against an abort clock — the longest credible time from the decision to leave until the crew is safe. At NRHO, that clock runs days, not hours. Oxygen, water, food, CO₂ scrubbing, and thermal control all size to that interval with margin.

Medical capability cannot assume rapid evacuation. Any condition that would be handled by getting to an emergency room on Earth must be handled on board or endured for the duration of the return.

Redundancy must be designed around unreachability. A second pump, a second computer, a second pressure shell — the backup is not the ground crew. It is the hardware that was launched with the station.

Gateway

NASA’s Lunar Gateway: Paused in March 2026

A decade of architecture, billions in hardware, paused in one announcement.

On 24 March 2026, NASA Administrator Jared Isaacman announced that Gateway is “paused in its current form” and the agency will redirect resources toward a $20 billion permanent lunar surface base, to be developed between 2029 and 2036. The decision reshapes the entire Artemis architecture.

Gateway was designed as a small crew-tended station in a southern L2 near-rectilinear halo orbit (NRHO). It was meant to serve three roles simultaneously: a transfer station where Orion crews would board HLS landers before descending to the surface, a science platform for cislunar and deep-space research, and an international partnership structure binding Canada, ESA, Japan, and the UAE into the Artemis program.

The three most advanced modules — the Power and Propulsion Element (PPE, built by Maxar/Vantor), the Habitation and Logistics Outpost (HALO, built by Northrop Grumman), and ESA’s International Habitation Module (I-Hab) — face uncertain futures. Isaacman indicated the hardware would be “repurposed” for surface objectives or other programs, but the engineering challenges of adapting orbital modules for surface use are substantial.

The hardware that exists

Gateway’s Built Modules and Partner Commitments

The station was largely built when it was paused.

CAPSTONE — the 25 kg CubeSat that validated the NRHO from November 2022 through mid-2026 — was a Gateway pathfinder. NASA ended its CAPSTONE involvement in July 2026, though the spacecraft itself continues to operate under Advanced Space. The NRHO data it gathered remains the most complete operational characterization of the orbit, regardless of whether Gateway ever occupies it.

ILRS orbital segment

China’s International Lunar Research Station: The Orbital Component

ILRS is primarily a surface station, but its architecture includes a lunar-orbit segment.

China’s International Lunar Research Station (ILRS) — led by CNSA with Roscosmos and over 40 partner institutions — plans a basic surface station at the lunar south pole by 2035 and an expanded station with a moon-orbiting hub by 2045. The orbital segment is intended to serve as a logistics and communications hub connecting surface facilities, cislunar transit vehicles, and Earth ground stations.

The precursor missions are already in motion. Chang’e-7, scheduled for around 2026, will survey the south-pole environment and resources, carrying six international scientific instruments. Chang’e-8, targeting 2028, will demonstrate in-situ resource utilization technologies and work alongside Chang’e-7 and the Queqiao-2 relay to establish the basic ILRS model.

The ILRS orbital component has no equivalent to Gateway’s detailed module manifest yet. What is visible: China intends a crewed lunar landing by 2030 using Long March 10 and the Mengzhou crew vehicle, and the architecture includes a lunar-orbit rendezvous. Whether that rendezvous involves a permanent orbital station or a series of temporary crew vehicles is not yet publicly settled. ILRS Phase 2 (2036–2045) explicitly includes a lunar-orbiting space station as the system hub, with surface, orbital, and Earth-based segments operating as a network.

Programs at a glance

Cislunar Station Programs

Two programs, two architectures, two timelines.

Why the orbit is the design

Orbit Selection as System Architecture

The orbit is not where the station goes. The orbit is what the station does.

Gateway’s NRHO was chosen not for science or convenience but for propulsion budget, eclipse avoidance, and surface access. The 9:2 synodic resonance (period ≈ 6.56 days) keeps the orbit eclipse-free — eliminating the deepest battery-sizing driver. Perilune around 3,000–3,500 km provides long dwell over the lunar south pole for surface operations. Apolune near 70,000 km keeps the station visible from Earth for near-continuous communications. And the near-marginal stability means station-keeping is manageable but departure manifolds exist for outbound vehicles.

The lesson generalizes beyond Gateway. For solar-powered infrastructure in cislunar space, the correct design variable is not average eclipse time — it is the worst contiguous outage. The cheapest battery is an orbit chosen so the outage never occurs.

An EML1 or EML2 halo orbit would offer better corridor access and depot logistics but at the cost of less polar coverage and higher station-keeping. A DRO would provide stability but poor surface access and no departure manifold. Every cislunar station is a trade among these variables. The orbit is the architecture.

Radiation and shielding

Radiation Exposure Beyond the Magnetosphere

LEO sits inside Earth’s magnetic shield. Cislunar space does not.

The ISS orbits within Earth’s magnetosphere, which deflects much of the galactic cosmic ray (GCR) flux and nearly all solar energetic particles (SEPs) during normal conditions. A cislunar station sits outside this shield entirely. Crew radiation exposure rises correspondingly — roughly doubled for GCR, and catastrophically higher during a major solar particle event without shelter.

The architectural response is a storm shelter — a small, heavily shielded volume the crew retreats to during an SEP event. Gateway’s HALO module included a storm-shelter zone with additional polyethylene or water shielding. The shelter does not need to be large or comfortable. It needs to reduce dose rates below career-limit thresholds for the duration of the event, which can last hours to days.

The upstream warning network at Sun-Earth L1 — SOHO, ACE, IMAP, SOLAR-1 — provides the early detection that gives a crew time to reach the shelter. Without L1 monitoring, a cislunar station has no advance notice of an incoming particle storm. The two pieces of infrastructure are coupled.

Lunar-sourced shielding mass — regolith, water, structural slag — could eventually replace launched shielding. This is one reason a lunar mass driver exporting bulk material to EML1 has value beyond propellant: the cheapest radiation shield is the one that does not ride a rocket from Earth.

What pausing reveals

What Pausing Gateway Reveals About the Regime

The orbit is still there. The physics has not changed. The policy question was always which surface to prioritize.

Gateway’s pause does not invalidate the cislunar regime. It clarifies a priority: with constrained budgets and a competitor building a surface station, NASA chose the surface. The NRHO remains the most operationally characterized cislunar orbit. The PPE’s solar-electric propulsion system remains the most advanced high-power SEP system ever built. The international agreements remain in force, redirected rather than cancelled.

The deeper lesson is that a cislunar orbital station is infrastructure for a regime, not a destination in itself. Gateway made sense as a transfer node in an architecture that included regular crew rotations to the surface, propellant depots, and outbound deep-space missions. Without those downstream users, the station becomes overhead.

China’s ILRS plans the orbital segment later — after the surface infrastructure exists to justify it. That sequencing may prove more durable than Gateway’s orbit-first approach. The junctions and warehouses will be occupied when the traffic justifies them. The traffic starts on the surface.