SpaceX Starship Explained — Rocket Built for Mars

Starship is the largest and most powerful rocket ever flown. At 121 meters (397 feet) stacked, it is taller than the Statue of Liberty including its pedestal, and its 33-engine booster produces roughly twice the liftoff thrust of the Saturn V. What makes it different from every heavy-lift rocket before it is not just size. SpaceX is trying to make the whole vehicle, booster and ship, fly, land, and fly again the same day, the way an airliner does. On July 24, 2026, the redesigned "V3" version flew for the first time carrying next-generation Starlink satellites, an important but still incomplete step toward the orbital refueling and Mars architecture the entire program depends on.

What Starship actually is — two reusable stages, not one rocket

Most rockets are built once and used once. Starship is designed as a stack of two independently reusable vehicles that happen to launch together.

Super Heavy is the booster. It is 71 meters (233 feet) tall, 9 meters (30 feet) wide, and carries 33 Raptor engines arranged in three concentric rings. At full throttle, those 33 engines together produce roughly 74 meganewtons (about 16.7 million pounds) of thrust, which SpaceX and independent trackers have described as close to double the Saturn V's 35-meganewton liftoff thrust (NASA Saturn V fact sheet).

Starship, called "Ship" internally, is the upper stage and the spacecraft itself. It stands about 50 meters (164 feet) tall on the V3 design, uses the same 9-meter diameter as the booster so the two stages stack cleanly, and carries six to nine Raptor engines depending on the version — a mix of sea-level engines for landing and vacuum-optimized engines for the coast to orbit.

Both stages are welded from stainless steel rather than the carbon-fiber composites or aluminum-lithium alloys used on most modern rockets. That was a contrarian call when SpaceX announced it in 2018, but the logic holds up: steel keeps its strength at the high temperatures of atmospheric re-entry, it is dramatically cheaper per kilogram than aerospace-grade carbon fiber, and it can be welded outdoors in a tent in South Texas rather than in a climate-controlled autoclave facility. Think of it as the difference between building a bridge out of forged steel girders you can weld in a parking lot, versus custom carbon-fiber panels that need a clean room. The steel version is heavier per unit strength, but it is fast and cheap to iterate on, which matters enormously for a program that has intentionally destroyed dozens of prototypes to learn faster.

Super Heavy's 33 Raptor engines ignite in under a second on liftoff, and by the time the booster separates from the ship roughly two and a half minutes later, the vehicle has already burned through more propellant than an entire Falcon 9 carries at launch.

How the Raptor engine actually works

The engine is where Starship's real engineering bet lives. Raptor is a full-flow staged-combustion engine burning liquid methane and liquid oxygen (commonly abbreviated "methalox"). Understanding why that combination of words matters requires unpacking both halves.

Full-flow staged combustion is a way of running the turbopumps — the machinery that forces propellant into the combustion chamber at extreme pressure — without wasting any fuel or oxidizer to do it. Most rocket engines burn a small fraction of their propellant separately, in a "preburner," just to spin the turbopumps, then dump that partially-burned exhaust overboard or into the main chamber at low pressure. Full-flow staged combustion instead runs two preburners, one fuel-rich and one oxidizer-rich, and routes both full flows of propellant through turbopumps before they ever reach the main chamber together. Nothing is wasted, and because neither preburner burns at extreme temperatures (each is starved of either fuel or oxidizer), the turbine blades run cooler and last longer. It is like the difference between an engine that redirects a little bit of raw gasoline to spin a fan before throwing it away, versus one where every drop of fuel eventually contributes to actual thrust.

The catch is that full-flow staged combustion is brutally hard to build. Before Raptor, the only engine to reach flight-testing on this cycle was the Soviet RD-270, developed in the 1960s and never flown operationally (Encyclopedia Astronautica RD-270 entry). SpaceX is the first organization to fly the cycle successfully, first in the Falcon-derived test articles and then in Raptor.

Why methane instead of kerosene or hydrogen?

Methane sits in a sweet spot other propellants miss. It burns cleaner than RP-1 kerosene, leaving less soot ("coking") inside the engine, which matters enormously if you intend to refly the same engine dozens of times without a full teardown. It is denser than liquid hydrogen, so tanks can be smaller and lighter for the same amount of energy stored, and it does not require the extreme sub-70-kelvin insulation hydrogen does. Most importantly for Mars, methane and oxygen can both be manufactured on the Martian surface from local resources, which no other rocket propellant combination in wide use can claim.

What is Raptor's thrust today?

The production Raptor 2 engine generates roughly 230 tonnes-force at sea level. The newer Raptor 3, which debuted on Flight 12 in 2026, produces around 280 tonnes-force at sea level and roughly 306 tonnes-force in its vacuum-optimized variant, while being about 7 percent lighter than Raptor 2 — a meaningful jump in thrust-to-weight ratio and a design SpaceX says needs no external heat shield around the engine bay (SpaceX Raptor 3 announcement, X/Twitter, 2024).

Making fuel on Mars — the Sabatier reaction

Raptor's methane and oxygen were chosen specifically because both can be produced from what is already on Mars: carbon dioxide, which makes up about 95 percent of the thin Martian atmosphere, and water ice, which exists as subsurface deposits and at the poles. The chemistry is called the Sabatier reaction, discovered by French chemist Paul Sabatier in 1897 and long used industrially on Earth:

CO2 + 4H2 → CH4 + 2H2O

Carbon dioxide plus hydrogen, run over a nickel catalyst under heat and pressure, yields methane and water. The water can then be split by electrolysis into oxygen (for the rocket, and for breathing) and more hydrogen, which gets fed back into the Sabatier reactor to keep the cycle going. This entire approach is called In-Situ Resource Utilization, or ISRU: instead of hauling every gram of return propellant from Earth, you manufacture it on-site from local raw materials.

The reason this matters is not a technical curiosity, it is the mission's central constraint. A Starship returning from Mars needs roughly the same amount of propellant it took to land. Hauling that mass all the way from Earth, on top of the outbound propellant, the crew, and the cargo, would gut the payload capacity to something impractically small. NASA's own Mars ISRU research, including the MOXIE experiment that flew on the Perseverance rover and successfully produced oxygen from Martian CO2 in a series of 2021–2023 tests, exists precisely because every serious Mars architecture depends on some version of this idea (NASA MOXIE mission overview).

Reusability — landing a skyscraper, twice

The single hardest engineering problem in the Starship program is not the size of the rocket, it is bringing every part of it home intact, over and over, fast enough that reuse is actually cheaper than building a new one each time.

Super Heavy returns to the launch site after stage separation, executes a boostback burn to reverse its trajectory, and comes down tail-first toward the launch tower rather than to a drone ship or a separate landing pad. Two enormous mechanical arms mounted on the tower — nicknamed "chopsticks," with the tower itself called Mechazilla — close around pins near the top of the booster and catch it in mid-air. SpaceX achieved this for the first time on Integrated Flight Test 5 (IFT-5) on October 13, 2024, a moment that drew comparisons across the aerospace industry to catching a launched airliner returning to the gate under its own power (SpaceX Flight 5 mission summary). It is worth pausing on how unusual that is: no orbital-class rocket booster in history had previously been caught by the launch infrastructure itself rather than landing on legs.

Starship, the upper stage, faces a harder re-entry problem because it comes back from orbital or near-orbital velocity carrying far more heat. It re-enters belly-first in a horizontal "bellyflop" orientation, using its body as a lifting surface and its four flaps for control, shedding velocity through atmospheric drag while a layer of hexagonal ceramic tiles on the windward side absorbs peak temperatures that can exceed 1,400°C (2,550°F). Seconds before touchdown, it flips vertical and relights its engines for the final landing burn. SpaceX's eventual goal is for Ship, like the booster, to be caught by the tower rather than landing on legs, cutting turnaround time further.

The target SpaceX has stated publicly is a full turnaround, refurbishment where necessary, and reflight of both the booster and the ship within the same day — closer to how an airline manages a jetliner between flights than how any rocket has operated before. That target has not yet been met. As of the Flight 13 mission in July 2026, individual boosters had been caught and reused on subsequent flights, but daily-cadence turnaround remained a stated goal rather than an operational reality.

The test flight record — a rapid, expensive learning curve

SpaceX built the Starship program around a philosophy sometimes called "hardware-rich development": build many full-scale prototypes, fly them aggressively, expect early failures, and let real flight data replace years of ground simulation. It is a deliberate trade of individual vehicle cost against speed of learning, closer to how early aviation pioneers iterated on airframes than how traditional aerospace primes like Boeing or Lockheed typically develop a new rocket.

Table 1. Starship integrated flight test history, 2023–2026

Flight	Date	Key outcome
IFT-1	April 20, 2023	Multiple engines failed; vehicle lost roughly four minutes in; destroyed by flight termination system
IFT-2	November 18, 2023	First successful hot-stage separation; booster and ship both lost afterward
IFT-3	March 14, 2024	Reached space on a suborbital trajectory; ship lost on re-entry, booster lost on splashdown
IFT-4	June 6, 2024	Ship survived re-entry and soft-splashed in the Indian Ocean; booster completed a controlled Gulf splashdown
IFT-5	October 13, 2024	First Super Heavy booster catch by the launch tower's mechanical arms
IFT-6	November 19, 2024	Ship completed a controlled splashdown; booster diverted to a sea landing rather than a catch
IFT-7	January 16, 2025	First flight of the redesigned V2 ship; ship broke apart over the Caribbean
IFT-8	March 6, 2025	V2 ship, lost again during ascent
IFT-9 through IFT-11	2025	Continued V2 iteration; IFT-10 (August 26, 2025) deployed Starlink simulator satellites successfully but the booster splashed down hard in the Gulf rather than being caught
IFT-12	May 2026	First flight of Raptor 3 engines on a substantially redesigned booster; booster's engines failed to relight for its planned Gulf water landing
IFT-13	July 24, 2026	First flight of the V3 vehicle stack; carried 20 next-generation V3 Starlink satellites, with all 33 booster engines igniting cleanly after two earlier scrubbed attempts

Sources: Wikipedia, Starship flight test 10, Wikipedia, Starship flight test 11, Spaceflight Now, Flight 13 launch coverage, News 13 Orlando, Flight 13 coverage.

Two patterns stand out in that table. First, the early flights (IFT-1 through IFT-4) were about surviving ascent and re-entry at all; the later flights are about landing precision and engine reliability on a specific, already-proven airframe shape. Second, SpaceX changed the vehicle itself twice in this window, moving from the original design to "V2" after IFT-6 and then to "V3" with Raptor 3 engines starting around IFT-12. Each redesign reset some of the reliability gains from the previous version, which is a direct cost of iterating this fast, but also the point: SpaceX is choosing to find structural and engine problems in flight rather than spend years simulating them on the ground.

What is Flight 13 and why does it matter?

Flight 13, launched July 24, 2026, was the second flight of the redesigned Block 3 (V3) Starship and the first to successfully deliver a payload of next-generation V3 Starlink satellites to space. All 33 Super Heavy engines ignited cleanly, a milestone after two earlier launch attempts were scrubbed by engine ignition problems and weather.

How Starship compares to other heavy-lift rockets

Raw numbers only tell part of the story, but they show why Starship is treated as a category shift rather than an incremental upgrade.

Table 2. Heavy-lift launch vehicle comparison

Rocket	Payload to LEO	Height	Reusability	Status
Saturn V	140 tonnes	110 m	None (expendable)	Retired, 1973
Space Shuttle	27 tonnes	56 m	Partial (orbiter and boosters)	Retired, 2011
Falcon Heavy	64 tonnes (expended configuration)	70 m	Partial (side boosters and core)	Active
SLS Block 1	95 tonnes	98 m	None (expendable)	Active, Artemis program
New Glenn	45 tonnes	98 m	First stage only	Active since January 2025
Long March 9 (planned)	150 tonnes	110 m	Planned, full reuse	Targeting around 2030
Starship (V3, target)	100–200 tonnes	121 m	Full, both stages	In flight testing

Sources: NASA Saturn V fact sheet, NASA Space Shuttle overview, Blue Origin New Glenn overview.

Starship is the first vehicle in more than fifty years designed to exceed the Saturn V's payload capacity, and it is the first ever designed from the outset for full reuse of both stages at this scale. Falcon Heavy and New Glenn reuse their boosters but expend or discard their upper stages; Starship's bet is that reusing the upper stage too, despite the much harder re-entry heating problem that stage faces, is what finally makes the cost math work.

Orbital refueling — the step nothing else has been tested against

This is the part of the Starship architecture that has received the least public attention relative to how much the entire Mars and lunar plan depends on it. No spacecraft has ever transferred a large quantity of cryogenic propellant between two vehicles in orbit. SpaceX's plan requires it repeatedly, on every single deep-space mission.

Here is why. A Starship that launches to low Earth orbit arrives with its tanks mostly empty, because most of the propellant was spent getting there. To go anywhere beyond LEO — the Moon, Mars, anywhere requiring a large additional burn — that Starship needs to refill its tanks in orbit. SpaceX's plan is to launch a series of tanker Starships, essentially the same vehicle with cargo space replaced by extra propellant tanks, dock them to the target ship, and transfer cryogenic methane and oxygen between the two while both are in freefall. Estimates for how many tanker flights a single crewed or cargo mission requires have ranged from roughly 8 to over 15 depending on mission profile and which version of the vehicle is used.

SpaceX has been building toward this incrementally. A dedicated Starship Propellant Transfer Demonstration mission, intended to prove ship-to-ship transfer of cryogenic propellant in orbit, is targeted for later in 2026 (Wikipedia, Starship Propellant Transfer Demonstration). Until that demonstration succeeds, every downstream milestone, the Artemis III lunar landing, any Mars cargo mission, is provisional. This single unproven step is arguably the largest schedule risk in the entire program, larger than any individual engine or heat-shield problem, because none of the harder mission profiles can even begin without it.

Orbital refueling is not a Starship-specific curiosity. It is the mechanism that lets a rocket sized for Earth orbit reach the Moon or Mars without being built impossibly large. If it works reliably, it changes what "payload to deep space" means for the entire launch industry, not just SpaceX's own missions.

NASA Artemis — Starship as the Moon lander

NASA selected a modified Starship, called Starship HLS (Human Landing System), as the lander that will carry astronauts down to the lunar surface for Artemis III, which is intended to be the first crewed Moon landing since Apollo 17 in 1972. NASA's contract with SpaceX for the lunar lander work is worth $4.05 billion combined across Artemis III and Artemis IV (NASA HLS award announcement).

The lunar variant differs from the Earth-to-Earth vehicle in a few important ways. It has no atmospheric flaps or heat shield tiles, since the Moon has no atmosphere to survive re-entry through, but it carries side-mounted reaction control thrusters positioned high on the vehicle to avoid kicking up lunar dust during the final landing approach, and it needs the same orbital-refueling capability described above just to have enough propellant left for the descent, ascent, and return to lunar orbit. For the fuller picture of how Starship fits into the broader Artemis architecture alongside NASA's own Space Launch System and Orion capsule, see our explainer on the Artemis Moon Program.

The Mars architecture, step by step

A crewed Mars mission built around Starship, as SpaceX has described it in public presentations since 2016, looks roughly like this:

1. Launch a crewed Starship into low Earth orbit, largely empty of propellant after reaching orbit.
2. Launch a series of tanker Starships to rendezvous with the crewed ship and transfer propellant, refilling its tanks in orbit.
3. Once full, the crewed ship fires its engines for Trans-Mars Injection, the burn that sends it on a roughly six- to nine-month coast to Mars.
4. On arrival, the ship aerobrakes through the thin Martian atmosphere to shed velocity, then performs a powered vertical landing.
5. On the surface, the crew deploys solar or nuclear power, extracts water from subsurface ice or the atmosphere, and runs a Sabatier reactor to manufacture methane and oxygen return propellant from local CO2 and water.
6. Once the return tanks are full, the same Starship launches back off the Martian surface for the trip home.

Every step after step 2 depends on technology that either has never been demonstrated at scale (orbital refueling), has only been demonstrated in miniature (MOXIE produced oxygen at a rate of about 6 grams per hour, nowhere near what a return-propellant plant would need), or has never been attempted at all (crewed landing on another planet using powered descent alone, without parachutes for the final phase). This is not a criticism unique to SpaceX; it reflects how far beyond anything previously flown a Mars surface mission actually is. Whether Mars can sustain permanent human habitation once astronauts get there, independent of how they arrive, raises a separate set of biological and terraforming questions covered in our piece on whether humans can live on Mars.

What could actually stop or delay Starship

It is easy to treat an active flight-test program as inevitability. It is more useful to look at where the real risks sit.

• Orbital refueling has not been proven. Every deep-space mission profile depends on it, and the dedicated demonstration mission has not yet flown as of this writing.
• Heat shield durability remains a recurring problem. Early V1 ships shed tiles repeatedly during ascent and re-entry; V2 and V3 use revised tile attachment methods, but a fully reusable heat shield that survives dozens of re-entries without inspection or replacement has not been demonstrated.
• Engine reliability at high flight rate is unproven. Raptor 3 is meant to simplify manufacturing and reduce the parts count that has historically caused engine-out events, but it is new hardware, and IFT-12's booster engine relight failure shows the transition is not yet finished.
• Regulatory pacing is a real constraint, not a footnote. Each Starbase launch requires FAA licensing tied to an environmental review, and repeated anomalies, including debris falling outside designated hazard areas on some V2 flights, have triggered additional agency scrutiny and temporary pauses between flights.
• The promised cost per launch is far from demonstrated. SpaceX has floated launch costs as low as a few million dollars once the vehicle reaches high flight rate and full reuse, but current per-flight costs, driven by ongoing hardware losses and a still-developing production line, remain far higher, and SpaceX has not published verified figures.

None of these risks are exotic or unprecedented in rocket development; they are the same categories of problem every new launch vehicle has faced. What is unusual is the scale at which Starship needs to solve all of them simultaneously to fly people to Mars, rather than solving them one at a time across separate, smaller vehicles the way most space programs have historically staged their progress.

What people get wrong about Starship

A few misconceptions come up often enough to be worth addressing directly.

The first is treating every uncrewed test flight failure as evidence the program is failing. SpaceX's own stated methodology treats early prototype losses as expected data collection, not as setbacks in the conventional sense — the goal on IFT-1 was survivable telemetry through max dynamic pressure, not orbit. Judging the program purely by "did the rocket blow up" misses what each flight was actually testing for.

The second is assuming Starship reaching orbit and Starship being ready for Mars are close to the same milestone. They are not. Reaching a stable orbit with a functioning heat shield is a necessary but far from sufficient condition; the orbital refueling demonstration, ISRU hardware validation, and life-support systems for a six-to-nine-month transit are separate, largely unstarted problems that do not automatically follow from a successful orbital flight.

The third is expecting SpaceX's own announced dates to hold. The company has a well-documented pattern of publicly stated timelines slipping by one to several years, going back to the original 2018 target of an uncrewed Mars cargo mission by 2022. That pattern does not mean the milestones will never happen, but any specific year attached to a crewed Mars landing should be read as aspirational rather than as a schedule.

The takeaway

Starship is a genuine engineering departure from every rocket flown before it, not because of its size alone but because it is the first vehicle designed from day one for full reuse of both stages at orbital scale, using an engine cycle almost no one else has successfully flown. The July 2026 Flight 13 mission shows the vehicle can reach space reliably and deliver real payloads on the redesigned V3 airframe. But the harder, less visible milestone, transferring propellant between two ships in orbit, still has not been proven, and until it is, the Moon landing, Mars cargo missions, and everything downstream remain provisional rather than scheduled.

---

Read next:

• Can Humans Live on Mars? Terraforming and the Body's Limits
• Artemis: NASA's Plan to Return Humans to the Moon — Explained
• Asteroid Mining — Trillion Dollar Space Industry
• Complete Space Guide 2026

FAQ

Q: Is Starship the largest rocket ever built? A: Yes. At about 121 meters (397 feet) stacked, it exceeds the Saturn V's 110 meters (363 feet), and its 33-engine Super Heavy booster produces roughly twice the Saturn V's liftoff thrust.

Q: Has Starship reached orbit yet? A: Starship has reached space on suborbital trajectories across multiple flights, most recently Flight 13 on July 24, 2026, which successfully deployed Starlink satellites. It had not, as of that flight, completed a full stable orbital insertion and controlled deorbit on a standard operational profile.

Q: How does the Super Heavy booster catch actually work? A: The booster performs a boostback burn to reverse course toward the launch site, then descends tail-first toward the launch tower, where two mechanical arms nicknamed "chopsticks" close around pins near the top of the vehicle and arrest its fall. SpaceX achieved this for the first time on October 13, 2024, during IFT-5.

Q: What is orbital refueling and why does Starship need it? A: Orbital refueling means transferring propellant from one Starship to another while both are in orbit, refilling the tanks of the ship that will continue on to the Moon or Mars. Starship reaches low Earth orbit with its tanks mostly empty, so without a refill in orbit it cannot carry enough fuel for a deep-space burn. A dedicated demonstration mission for this capability is targeted for later in 2026.

Q: Why does Starship burn methane instead of kerosene or hydrogen? A: Methane burns cleaner than kerosene, reducing engine wear from soot buildup on a vehicle meant to be reflown dozens of times, and it is denser and easier to store than hydrogen. Most importantly, methane and oxygen can both be manufactured on Mars from atmospheric CO2 and water ice using the Sabatier reaction, which no other common rocket propellant combination allows.

Q: What is the Sabatier reaction? A: It is a chemical reaction, discovered in 1897, in which carbon dioxide and hydrogen combine over a catalyst to produce methane and water. On Mars, that reaction converts local atmospheric CO2 into rocket fuel, which is central to any plan for a Starship to return from the Martian surface without carrying its return propellant from Earth.

Q: Will Starship carry the Artemis III Moon landing? A: Yes. NASA selected a modified Starship, called Starship HLS, as the lander for Artemis III under a contract worth $4.05 billion combined with Artemis IV. That lander still depends on the same orbital refueling capability used for Mars missions.

Q: When might Starship reach Mars? A: SpaceX has publicly discussed an uncrewed cargo mission in the second half of the 2020s and crewed missions in the 2030s, but the company's timelines have slipped by years before, and the orbital refueling demonstration that every Mars mission depends on had not yet flown as of mid-2026.

Q: Will Starship replace Falcon 9? A: SpaceX has said it eventually wants to retire Falcon 9 and Falcon Heavy once Starship reaches a reliable, high-cadence flight rate, but Falcon 9 remains the company's primary operational vehicle, and no retirement date has been set.

Q: How much does a Starship launch cost? A: SpaceX has stated an eventual goal of a few million dollars per launch once the vehicle is flying at high cadence with full reuse, but the company has not published verified current per-launch costs, and independent estimates put present-day costs well above that target given ongoing hardware losses during testing.

Sources

• NASA Saturn V fact sheet
• NASA Space Shuttle era overview
• NASA MOXIE mission overview
• NASA Human Landing System award announcement
• SpaceX Flight 5 mission summary
• SpaceX Raptor 3 announcement, X/Twitter
• Spaceflight Now, Flight 13 launch coverage, 2026
• Wikipedia, Starship flight test 10
• Wikipedia, Starship flight test 11
• Wikipedia, Starship Propellant Transfer Demonstration
• Blue Origin New Glenn overview