Starship stands roughly as tall as a 40-storey building and carries more than 5,000 tonnes of liquid methane and oxygen. Methane was chosen partly because both propellants could theoretically be manufactured on Mars from atmospheric carbon dioxide and water ice—meaning its engines are designed to burn a return ticket whose ingredients may already be waiting there.

Starship’s stainless-steel silhouette makes its scale easy to see, but the less visible measure is inside its tanks. SpaceX lists the current full stack at 124 metres tall, with capacity for 1,600 tonnes of propellant in the upper ship and 3,650 tonnes in Super Heavy. Together that is 5,250 tonnes of liquid methane and liquid oxygen at full capacity.

A 40-storey building is a fair visual comparison, though an imprecise one because storey heights vary. At roughly three metres per floor, a 124-metre rocket comes out close to 41 storeys. The propellant figure carries a more important qualification: it belongs to the entire Earth-launch stack, not to the spacecraft that would sit on Mars.

Super Heavy supplies the first part of the climb and then separates. The 52-metre upper Starship continues alone. If that ship is ever to land on Mars and fly away again, it needs a new supply of fuel and oxidizer. Methane makes that return conceivable because Mars appears to offer the raw carbon, hydrogen and oxygen needed to manufacture both liquids.

The 5,000 tonnes belong to the launch stack

SpaceX’s current vehicle specifications split the tower into a 72-metre Super Heavy booster and a 52-metre ship, each nine metres in diameter. Their listed propellant capacities add to 5,250 tonnes. Thirty-three Raptor engines power the booster, while six power the ship in the configuration shown by SpaceX.

Raptor is a staged-combustion engine burning methane with oxygen. The two cryogenic liquids are stored separately, not as one combustible mixture, and meet inside the engines. Most of the mass is oxygen because methane cannot burn without an oxidizer and there is no useful oxygen supply in the air at launch altitude or in space.

The Mars return problem concerns the upper ship, whose current stated capacity is 1,600 tonnes. Even that number is not a required return load. The actual amount would depend on the vehicle version, payload, trajectory, reserves and mission design. Mars’s weaker gravity reduces the energy needed to leave its surface, so the ship would not need a Super Heavy beneath it.

One ship faces two refuelling problems

A Mars-bound Starship would burn much of its initial load simply reaching low Earth orbit. SpaceX therefore proposes refilling it there from tanker Starships before departure. SpaceDaily has examined the still-unproven chain of rendezvous, docking and cryogenic transfer required for that step.

A return trip creates a separate problem. Sending all the departure propellant from Earth would require landing enormous extra mass on Mars and keeping it cold through the mission. Producing it locally moves the burden to power plants, excavators, compressors, chemical reactors and storage tanks, but it avoids carrying the finished product across interplanetary space.

That trade shaped the fuel choice. SpaceX’s current Mars programme page frames surface operations as a system of enabling technologies, while the company’s published architecture has long described propellant production from atmospheric carbon dioxide and water ice, followed by refilling the ship for Earth. Methane’s local manufacturability is a design factor, not evidence that any working Mars factory exists.

The chemistry closes on paper

The proposed chain starts by turning ice into clean water. Electrolysis then splits that water: 2H₂O becomes 2H₂ plus O₂. The hydrogen flows to a Sabatier reactor, where CO₂ plus 4H₂ produces CH₄ plus 2H₂O. Recycling the water leaves a net result of CO₂ plus 2H₂O becoming CH₄ plus 2O₂.

The chemistry is almost the reverse of what happens in the engine, where methane and oxygen release energy and become carbon dioxide and water. NASA has investigated reactor designs that combine water electrolysis with methane production. The reactions are well understood; running them reliably at rocket scale on another planet is not.

Oxygen is the larger product stream. NASA concept studies often use an oxygen-to-methane mass ratio near 3.5 to one for Mars ascent propellant. The popular phrase “make fuel on Mars” can hide that requirement. A useful plant must manufacture, purify, liquefy and store both the fuel and its much heavier oxidizer supply.

Carbon dioxide is abundant but thin

Mars’s atmosphere is about 95 per cent carbon dioxide, an excellent fraction for a carbon source. Yet surface pressure is less than one per cent of Earth’s at sea level. A plant cannot treat that air like a dense industrial feed flowing from a terrestrial pipeline.

It must pull in a great volume of atmosphere, remove dust, compress the gas and deliver it steadily to reactors. Pumps and filters consume power and wear. Seals, valves and lubricants face large temperature swings. Dust can coat radiators and solar panels. The atmosphere is globally available, but extracting tonnes of carbon from it is still an industrial operation.

Power may become the plant’s central constraint. It is required for excavation, heating ice, electrolysis, compression, chemical processing, purification and refrigeration. Solar generation must cope with night, winter and dust; nuclear systems bring different mass, deployment and reliability questions. Neither option turns the process into a simple extension cord.

Water ice still has to become water

Orbiters have detected widespread buried ice, especially outside the equatorial region. NASA’s Subsurface Water Ice Mapping project combines observations from several missions to identify promising deposits across the northern mid-latitudes. Those maps help planners compare landing zones where ice may be relatively shallow.

Remote evidence is not the same as a proven mine. A specific site must contain enough accessible ice, in soil that robotic machinery can handle, at a rate compatible with the production schedule. The water may carry dust and dissolved salts that must be removed before it reaches electrolysers and reactors.

Landing-site choice becomes a compromise. Higher latitudes may offer more ice but harsher cold and less favourable sunlight. Lower latitudes are warmer and often better for solar power, yet usable ice may be deeper or absent. Flat terrain safe for a large vehicle must also sit close enough to the resource and the plant.

MOXIE proved one link, not the factory

Humans have already manufactured a propellant ingredient on Mars. Perseverance carried MOXIE, a toaster-sized experiment that drew in carbon dioxide and electrochemically stripped away oxygen atoms. Over its mission it operated 16 times and produced 122 grams of oxygen in total, reaching 12 grams per hour at its peak, according to NASA’s Jet Propulsion Laboratory.

MOXIE was a successful proof of principle. It showed that oxygen extraction can work through different Martian seasons and times of day. It did not mine water, produce hydrogen or methane, liquefy either propellant, keep large tanks cold or transfer anything into a rocket.

The scale gap is the lesson. MOXIE’s lifetime output would fit inside a small bottle. A crewed vehicle needs tonnes, potentially hundreds of tonnes, made repeatedly by hardware that can detect faults and recover without a technician standing beside it. Scaling a chemical reaction means multiplying pumps, heat exchangers, power and failure points as well as output.

A Mars fuel plant is an industrial system

NASA engineers have modelled plants sized to produce either 30 or 300 tonnes of methane and oxygen over one Martian year, about 687 Earth days. The 2025 feasibility assessment divides the task into water supply, electrolysis, methane production and storage. It also compares solar and fission power.

This is useful scale context, not a published design for filling Starship. Even the 300-tonne case is less than one fifth of the upper ship’s listed full capacity. A Mars departure may require far less than 1,600 tonnes, but SpaceX has not published a fixed operational load for a mature return mission, and the vehicle itself continues to evolve.

A functioning site also needs redundancy. A buried cable, frozen valve or failed excavator could stop every downstream process. Spare parts and repair robots add landing mass. The plant has to measure product purity, recycle water, vent safely, prevent leaks and keep methane and oxygen isolated until they enter the engine.

The hardest part may be keeping the tanks full

Both products must be cooled into liquids before they become compact rocket propellants. Liquid oxygen boils at about minus 183 degrees Celsius at Earth sea-level pressure, while methane boils near minus 162 degrees. Local operating pressures change the precise temperatures, but the engineering fact remains: the tanks must stay extremely cold.

Heat leaking through insulation boils liquid into gas, raising tank pressure and wasting product if it must be vented. NASA’s zero-boil-off research explores active cooling that removes incoming heat instead of accepting gradual propellant loss. On Mars, refrigeration would compete for the same power required to keep production running.

Storage time matters because a plant may fill tanks slowly over many months. The ship could arrive only after a surface depot is ready, or the production system could land with it. Each architecture changes how long the propellant waits, which equipment must survive first and whether a crew has a verified way home before leaving Earth.

The return ticket has to be printed before departure

The safest logic is to deliver the power and propellant plant ahead of people, start it autonomously and confirm a sufficient stock before committing a crew. That requires communications, health monitoring and repair strategies that work despite signal delays. It also requires a landing precise enough for the arriving ship to reach the fuel site.

None of these obstacles makes the chemistry imaginary. Carbon dioxide is present in the air. Water ice is present beneath parts of the surface. Electrolysis and Sabatier reactors are established technologies, and MOXIE showed that resource processing can work on Mars. The unresolved question is whether all the steps can be integrated at the mass, reliability and cadence a large reusable ship demands.

Methane gives Starship an unusually elegant loop. Its engines consume methane and oxygen, while the exhaust molecules point back to the carbon dioxide and water from which fresh propellant can be made. Mars may hold the ingredients for a return ticket, but robots, reactors and refrigerators would still have to print it, one cold tonne at a time.