A two-stage, stainless-steel launch system built to fly, be caught by its own tower, refuel and fly again. Explore the hardware, fly a mission and catch both stages yourself.
Space Exploration Technologies Corp. was founded by Elon Musk in 2002 with one stated aim: cut the cost of reaching space far enough to make a settlement on Mars possible. Its method is reuse. Falcon 9 boosters land and refly; Dragon capsules carry crew to the International Space Station; Starlink satellites fund the work. Starship is the attempt to make the whole rocket reusable, both stages. Select a year.
Two stages, both burning liquid methane and liquid oxygen. The Super Heavy booster does the first two and a half minutes, then flies back. The upper stage, called Ship or simply Starship, carries the payload on to orbit. Select a numbered marker, step through with the arrows, or switch to the cutaway to see the tanks.
Schematic side elevation, Block 3. Proportions follow published figures; internal layout is simplified and the header-tank positions are indicative.
Figures from the public record; SpaceX revises them between flights.
Liftoff thrust, meganewtons
Height, metres
Seven years of hardware, drawn to one scale. Select a vehicle to see what changed and what it proved.
Raptor is a full-flow staged-combustion engine. All of the methane and all of the oxygen pass through a preburner and a turbine before reaching the main chamber, so nothing is dumped overboard and the chamber can run near 350 bar. Higher chamber pressure means more thrust from a smaller engine, which is how 33 of them fit under a 9 m booster.
| Quantity | Value |
|---|---|
| Propellants | CH₄ + O₂ (methalox) |
| Cycle | Full-flow staged combustion |
| Chamber pressure | ≈ 350 bar |
| Thrust, Raptor 2 | ≈ 2.3 MN |
| Thrust, Raptor 3 (80.8 MN ÷ 33) | ≈ 2.45 MN |
| Specific impulse, sea level | 327 s |
| Specific impulse, vacuum | 350 s |
| Specific impulse, Raptor Vacuum | 380 s |
| Booster / Ship count | 33 / 6 (3 SL + 3 vac) |
Why methane: it burns cleaner than kerosene, which matters for reuse, it is denser and easier to store than hydrogen, and it can in principle be made on Mars from CO₂ and water ice.
On 13 October 2024 the booster came back to the pad for the first time. Drag the slider or press play to follow both stages. Times are approximate and the drawing is not to scale.
Fit a payload, launch, reach orbit, release the cargo and bring both stages home. Use the level switch in the top bar: Simple flies itself and explains each moment, Normal adds the flight plan and in-flight controls, Scientific adds every setting, manual steering and live charts.
A flatter pitch gives the Ship more speed but sends the booster further from home. Reserves are propellant held back for the return, so they cost payload.
The board has flown your mission in the background. Green means it expects to reach orbit with this payload.
These unlock when the ascent ends. Each return starts with the propellant your flight left over.
A round, rotating Earth; thrust and specific impulse that rise as the air thins; drag; mass falling as propellant burns; a 4 g limit on the Ship; and break-up above 60 kPa of dynamic pressure. Vehicle masses are estimates, so treat results as indicative.
Mechazilla is SpaceX's name for the launch tower and its two steel arms, the chopsticks. The booster has no landing legs. It flies itself to a hover between the open arms, the arms close, and it settles onto them by small load-bearing pins near its top. The model replays the last 25 seconds: a 71 m, roughly 250-tonne stage arriving at about 300 m/s.
Touch or mouse: drag on the view. Left and right tilt the booster, up and down set the throttle. Keyboard: 1 / 2 / 3 choose engines off, 3 or 13, arrows tilt and throttle. Match the descent rate to the guidance target and arrive within 2.5 m of centre at under 6 m/s.
Try 35 t of propellant, 8 failed engines, or a 300 m starting offset. With abort logic on, the booster checks its own prediction and stays offshore, as the real one does. Turn it off to see why the rule exists.
Legs strong enough for this stage would weigh many tonnes and fly on every launch. Moving that hardware to the ground makes the rocket lighter and puts the booster back on its launch mount, which is where it needs to be for a fast turnaround.
Catching the upper stage is the harder problem. The booster comes home from about 7,000 km/h a few minutes after launch. The Ship comes home from orbit at about 28,000 km/h, has to survive re-entry, falls on its belly with no wings, and must swing upright and stop at the arms in the last 18 seconds. On 14 November 2026 Ship 42 did exactly that. The model starts 1.5 km above the arms with the Ship falling broadside at 72 m/s.
Touch or mouse: drag on the view. Far left is flat, three-quarters across is upright; up and down set the throttle. Keyboard: 0 to 3 choose engines, arrows pitch and throttle. Light three engines near 530 m, pitch upright, then drop to two and to one as the guidance target falls. Arrive upright, within 2.5 m of centre, under 6 m/s.
Try the flip 150 m late, or 8 t of propellant. A late flip leaves too little height to cancel the sideways kick from the engines.
| Mission clock | Event |
|---|---|
| T+22:32:16 | Landing burn starts |
| T+22:32:18 | Flip to vertical |
| T+22:32:34 | Caught, about 21:20 UTC |
Eighteen seconds from ignition to catch. The model above takes 18.2 s.
| Super Heavy | Ship | |
|---|---|---|
| Comes back from | Staging, about 7,000 km/h | Orbit, about 28,000 km/h |
| Time away | About 7 minutes | Hours to days |
| Heat protection | Bare steel | Ceramic tile shield |
| Steering while falling | Grid fins, tail first | Four flaps, belly first |
| Final manoeuvre | Landing burn, slide to tower | Landing burn, flip, slide to tower |
| Engines in the burn | 13 → 3 | 3 → 2 → 1 |
| Held by | Pins near the grid fins | Fittings below the forward flaps |
| Caught so far | Flights 5, 7, 8 | Flight 15 |
It is a spacecraft and a second stage at once. Most rockets throw their upper stage away. The Ship carries its own heat shield, flaps and landing propellant so that it can come back, which costs payload on every flight.
Header tanks. Landing propellant sits in small separate tanks. During the flip the main tanks are nearly empty and their contents slosh; the header tanks stay full so the engines never swallow gas.
Why the energy matters. Each kilogram in low orbit carries about 30 megajoules of kinetic energy, roughly seven times its own weight in TNT. Re-entry turns almost all of it into heat in the air around the Ship.
Planned variants. A cargo Ship with the slot door flown today. A tanker and an orbital depot for refuelling. The lunar lander for Artemis, which never returns to Earth and so has no heat shield or flaps. A crewed Ship later.
A Ship reaches orbit nearly empty. This applies the rocket equation to a Ship that tankers have refilled in low Earth orbit, and shows which destinations come within reach.
Δv = Isp · g₀ · ln(m₀ / m₁), with a blended Isp of 370 s. Destination markers are textbook values from low Earth orbit and assume the atmosphere does the braking at Mars.
Each row scores three things separately: the launch, the booster's return and the ship's return. Read down the columns to see the programme learn.
Recovered / success Controlled splashdown Lost Not reached
Dates below are published targets. In this programme they have moved often, and always later.
The next Starlink satellites are too large for Falcon 9. Flight 14 put the first 26 into orbit and Flight 15 added 26 more. This is Starship's first paying job.
Operational nowA Ship reaches orbit with nearly empty tanks. Going beyond needs tanker flights to a depot. Flight 3 moved propellant between tanks inside one ship; ship-to-ship transfer is the next proof. NASA has estimated about 16 launches per lunar landing.
Ship-to-ship demo pendingNASA chose a Starship variant, the Human Landing System, under a 2021 contract worth $2.89 billion. A docking test with Orion in Earth orbit comes first, then a crewed landing.
Docking test 2027 · landing 2028Booked cargo includes the Superbird-9 communications satellite, lunar rovers, and modules for the Starlab and Haven-2 private space stations.
2027 – 2030Flights so far have all left Starbase in Texas. The next is planned from Launch Complex 39A at Kennedy Space Center, the pad that sent Apollo 11 to the Moon.
Flight 16 · no earlier than Dec 2026The founding goal. An uncrewed landing attempt with about five ships is on the manifest without a date, and SpaceX has put the Moon ahead of it. Mars launch windows open every 26 months.
No date setSteel, not carbon fibre. Stainless steel gets stronger at cryogenic temperatures, tolerates re-entry heat better than aluminium, and is cheap enough to build and scrap prototypes quickly.
Hot staging. The Ship lights its engines while still attached, blasting through a vented ring. The booster never stops pushing, which adds payload.
The belly flop. The Ship re-enters broadside, using four flaps as air brakes, then swings upright in the last seconds with a short engine burn.