Why a hub
One Vehicle Carries Four Destination Pages
The Moon, cislunar space, Earth orbit and Mars pages each lean on a different Starship variant. This page holds the shared record once.
Starship is a two-stage, stainless-steel, methalox launcher: a 33-engine Super Heavy booster under a six-engine ship that is also the spacecraft. Both stages are meant to return to the tower that launched them and be caught by its arms — the “chopsticks” on the tower SpaceX calls Mechazilla. Anything bound beyond low Earth orbit refuels there first, from tanker flights and an orbital propellant depot.
That architecture ties the site together. The Moon page depends on Starship HLS and its cargo derivative. The Cislunar Space page depends on the depot and ship-to-ship propellant transfer, which has not yet flown. The Earth Orbit page depends on the Starlink V3 deployer ship and on Starmind, the orbital data-center constellation Starship is meant to launch. The Mars page depends on one-way cargo landers and the hulls they leave behind. The Chinese counterparts — Long March 9, Long March 10 and LandSpace’s Zhuque-3 — are tracked here beside it. Building
Vehicles and variants
Block 3 Is Flying; Every Beyond-LEO Variant Waits on Propellant Transfer
One airframe, many jobs. Variants differ in what the payload bay holds and whether the ship ever comes back through Earth’s atmosphere.
| Vehicle | Key figures | Status and role | Rated |
|---|---|---|---|
| Super Heavy Block 3 booster | 72.3 m · 3,650 t propellant · 33 Raptor 3 | Flying. Interstage now integrated with the methane tank; three enlarged grid fins that double as tower catch points. All 33 engines can relight. No Block 3 booster has yet landed intact | Observed |
| Starship Block 3 ship | 1,600 t propellant · 3 sea-level + 3 vacuum Raptors · 100 t to LEO | Flying. Adds docking ports and a redesigned quick disconnect for refueling, plus a revised heat-shield tile design | Observed |
| Block 4 stack | ~142 m · ~200 t to LEO · booster 4,050 t, ship 2,300 t | Planned. Ship grows to nine engines — three more Raptor Vacuums. Waits on the Gigabay integration buildings | Projected |
| Starlink deployer (“PEZ”) | 26 Starlink V3 per flight planned · ~1 Tbps each | Flying. A slot door and a rack that pushes satellites out one at a time. The first commercial workload and the one that pays for the program | Observed |
| Tanker | Payload is propellant | In development. Flies as the “chaser” in the ship-to-ship transfer demonstration, which slipped from March 2025 to March 2026 and is still unscheduled | Building |
| Propellant depot | Stays in LEO · >10 tanker loads per lunar mission | In development. No heat shield or flaps. Filling begins more than 200 days before a crewed HLS flight, per NASA’s Office of Inspector General | Building |
| Starship HLS | ~100 t to lunar surface · crew 2–4 · two 13 m³ airlocks | In development under NASA contracts of $2.89B (2021) and $1.15B (2022). Never re-enters Earth’s atmosphere. Artemis III is now a 2027 LEO docking test with Orion; first crewed landing moved to Artemis IV, early 2028 | Building |
| Large Cargo HLS | Dedicated cargo, no crew cabin | Ordered by NASA in November 2023; expected in service by Artemis VII. Framed as a delivery vehicle — no disposition of the landed hull is published | Projected |
| Crewed Starship | Circumlunar pass within 200 km · two-year Mars flyby | Announced June 2026: a week-long lunar flyby with Dennis and Akiko Tito and Chun Wang, then Chun Wang’s Mars flyby. Polaris III with Jared Isaacman is also on the manifest | Projected |
| Mars cargo lander | Five ships were planned for the 2026 window | Deferred. SpaceX told investors in February 2026 it would put the Moon first and dropped the uncrewed 2026 Mars landing. Italian Space Agency experiments remain contracted for the first flight | Projected |
| One-way lunar and Mars cargo hulls | Tens of tonnes of steel, engines, tanks per landing | Original analysis, not a SpaceX or NASA plan — see One-way cargo | Modeled |
Raptor is the common engine: full-flow staged combustion on subcooled methane and oxygen. Raptor 3 moves the plumbing and sensors inside the engine body, which removed most of the heat shielding around the engine bay and saves over a tonne per booster engine. Production and testing run at McGregor, Texas. Flight reliability is the open issue — on Flight 13 only 10 of 13 booster engines lit for the landing burn.
Flight record
Block 3 Ships Reach Their Targets; Block 3 Boosters Have Not Yet Landed
Thirteen flights since April 2023: eight successes, five failures, three booster catches — all by Block 1 and Block 2 boosters.
| Flight | Date · vehicles | Outcome | Rated |
|---|---|---|---|
| Flight 12 — first Block 3 | 22 May 2026 · Booster 19, Ship 39 | First launch from Pad 2. Boostback failed and the booster was lost in the Gulf of Mexico. The ship lost one Raptor Vacuum on ascent, still reached engine cutoff, and splashed down in the Indian Ocean. FAA grounded the program until 13 July | Observed |
| Flight 13 — first Starlink V3 | 24 July 2026 · Booster 20, Ship 40 | Twenty working Starlink V3 satellites deployed on a suborbital path. The booster hit the Gulf hard after a partial landing burn. The ship was the first to survive tipping over after splashdown, and was recovered aboard the heavy-lift vessel Boskalis Forte | Observed |
| Flight 14 — first orbital flight | 28 September 2026 · Booster 21, Ship 41 | Ship 41 placed 26 Starlink V3 satellites in LEO and flew two truncated orbits at 275 km over roughly three hours, then splashed down in the Pacific north of Hawaii. Booster performed a soft splashdown offshore in the Gulf. The ship catch was removed from this flight during FAA review | Building |
| Flight 15 — catch candidate | Communications permit window from 28 September 2026 | Could attempt a booster catch, a ship catch, or both. A ship catch would be the first tower recovery of an orbital upper stage | Projected |
Production
Two Gigabays Quadruple Integration Capacity by the End of 2026
Starbase builds today; Florida is being set up to build its own ships and boosters rather than receive them by barge.
| Facility | Scale | Role and status | Rated |
|---|---|---|---|
| Starfactory, Starbase | Parts and subassembly | Operating. Rings, domes and tile work feed the stacking bays next door | Observed |
| Mega Bay 1 and Mega Bay 2, Starbase | Six work stations combined | Operating. Current booster and ship stacking | Observed |
| Gigabay, Starbase | ~116 m tall · ≥24 work stations | Under construction on the former High Bay and STARGATE footprint; construction began 1 July 2025, cladding under way, completion targeted December 2026. Needed for Block 4 stretch | Building |
| Gigabay, Roberts Road, Kennedy Space Center | 380 ft · 815,000 ft² · 24 cells · 400-ton cranes | Structural steel finished by July 2026; roof and cladding near complete. Integration and refurbishment for Space Coast flights. Already building booster stands while Starbase supplies ship hardware | Building |
| Starfactory, Florida | Footprint ~50% larger than Starbase’s | Planned at Roberts Road. Until it opens, finished boosters and ships travel from Starbase by barge | Projected |
| McGregor, Texas | Horizontal and vertical Raptor stands | Raptor 3 production and test since August 2024; also tests header tanks and other components | Observed |
| Massey’s test site, Starbase | Cryogenic proof and ship static fire | Every booster and ship is proof-tested here. A new rig squeezes the payload section and tests the refueling mechanism. Rebuilt after Ship 36 exploded in June 2025 | Observed |
| Cocoa and Cape Canaveral, Florida | Heat-shield silica | Cocoa purifies silica into slurry; tiles are produced at a Cape Canaveral facility | Observed |
Launch sites
From One Active Pad to Fifteen: Texas, Florida, Then Pecan Island
Every pad is also a catch site, so the pad count is the recovery count.
| Site | Pads | Status | Rated |
|---|---|---|---|
| Starbase Pad 2, Texas | 1 | The only active Starship pad; first used on Flight 12. Flame trench and hardened service bunker from the start | Observed |
| Starbase Pad 1, Texas | 1 | Rebuilt from the ground up with a new flame trench, refurbished tower arms and shielding for ship catches. Two working Starbase pads are hoped for in early 2027 | Building |
| LC-39A, Kennedy Space Center | 1 | FAA approval February 2026 for up to 44 launches and 88 landings a year. Deluge testing has begun; first Florida Starship launch is possible before the end of 2026 | Building |
| SLC-37, Cape Canaveral Space Force Station | 2 | Launch mount under construction at SLC-37A. Owned by the US Space Force, which makes it the likely pad for national-security missions | Building |
| Starbase Louisiana, Pecan Island, Vermilion Parish | 10 | Announced 25 August 2026: about 125,000 acres of former Exxon Mobil land, $100 billion committed, five launch complexes of two pads each, propellant production, a power plant, a deep-water port, an airport and housing. Construction from 2027, first launch as early as 2029. Musk’s stated goal is more than 30 flights a day. Southward launches reach polar orbits for Starmind | Projected |
Propellant
Oxygen Moves On Site; Methane Stays Trucked Until Louisiana
A Starship launch loads about 5,000 t of liquid oxygen and liquid methane, roughly 78% of it oxygen by mass.
| Element | What it does | Status | Rated |
|---|---|---|---|
| Starbase orbital tank farm | Stores liquid methane, liquid oxygen, liquid nitrogen for subcooling, and deluge water; subcoolers chill propellant below boiling for density | Operating; two new methane tanks added in 2026 for dual-pad work | Observed |
| Starbase air separation unit | Separates oxygen, nitrogen and argon from air beside Highway 4 | Under construction since 23 July 2025. Replaces most of the 200-plus tanker trucks each launch needs — about a 75% cut | Building |
| Linde plant, Brownsville | Industrial-gas supplier producing oxygen, nitrogen and argon | Expected online in 2026; cuts delivery from about 10 hours to under 2 | Building |
| Methane (LNG) supply | Commercial liquefied natural gas, trucked in | SpaceX dropped an on-site gas pretreatment plant and liquefier in 2022 after Raptor 2 proved it could burn commercial LNG | Observed |
| Florida tank farms | LC-39A tank farm and flame trench built alongside the pad | Nearing completion with the pad | Building |
| Louisiana propellant production | On-site plant and power generation next to Gulf Coast natural gas | Planned. Local gas supply was a stated reason for the site | Projected |
| Orbital depot | Propellant storage in LEO, filled by tankers | Not yet flown — ship-to-ship cryogenic transfer is the gate for every Moon and Mars variant | Building |
Recovery and landing
Towers Catch Boosters; the Ship’s Return Corridor Is the Real Constraint
Starship has no landing legs and no drone ship. Recovery is a tower catch at the launch site, and the limit is where a returning ship may fly.
The booster flips after staging, burns back, and reaches the tower about six minutes after liftoff — a short, fixed path over water. The ship returns from orbit: its ground track has to cross a catch site, and its final reentry leg crosses land or crowded airspace. SpaceX’s proposed orbital paths cross Florida, where the FAA estimated hazard areas would affect between 10 and 200 commercial flights an hour depending on time of day. Each return corridor needs its own FAA environmental review. The Flight 14 assessment added a new reentry path back to Starbase and a contingency landing area in the northeast Pacific for Florida-to-Starbase returns — and the same review took the ship catch off Flight 14.
| Method | Record | Notes | Rated |
|---|---|---|---|
| Booster tower catch | 3 catches: Boosters 12, 14, 15 | First on 13 October 2024. None yet with Block 3, whose grid fins now serve as the catch points | Observed |
| Ship tower catch | 0 attempts | Pad 1 rebuild adds shielding for the ship’s three sea-level Raptors during the flip-and-burn. First attempt no earlier than Flight 15 | Projected |
| Ocean splashdown | Gulf of Mexico, Indian Ocean, Pacific | Current test practice for both stages. Ship 40 was lifted intact from the Indian Ocean aboard Boskalis Forte and sent to the Port of Brownsville | Observed |
| Barge transport | Starbase → Florida | Moves finished boosters and ships to the Space Coast until the Florida Starfactory opens | Observed |
| Offshore platforms Phobos and Deimos | Abandoned 2023 | Two oil rigs bought in 2020 for sea launch and catch towers, on Musk’s view that high-cadence spaceports would sit about 20 miles offshore for noise. Both sold. Starbase Louisiana’s remote coastline fills the same role on land | Observed |
| Net-catch ship (China) | 1 recovery, 10 July 2026 | CALT caught the Long March 10B booster with a hook-and-net system on a vessel in the South China Sea — the only working sea recovery of an orbital booster, and the alternative model to tower catch | Observed |
Tower count is not the shortage: LC-39A’s license already counts 88 landings against 44 launches, and Louisiana adds 10 towers. Three things are: no Block 3 catch has been shown, each tower both launches and catches so a recovery occupies the pad until the vehicle is cleared, and missions too energetic for a booster flyback have no downrange landing option. SpaceX answers the last with orbital refueling, which keeps boosters on short return paths.
One-way cargo · original analysis
A Cargo Ship That Stays Is Worth More as Parts Than as a Return Flight
NASA and SpaceX describe cargo landers as delivery vehicles. Neither has published what happens to the hull after it lands.
Returning a lunar cargo Starship means refueling it on the surface or in lunar orbit. Until propellant is made locally, every tonne held back for the trip home is a tonne not delivered, so one-way delivery maximizes cargo. The landed ship is then the cheapest hardware on site: welded stainless pressure volume, cryogenic tanks, Raptors, solar arrays and avionics, already paid for. A Mars lander also brings heat-shield tiles usable as furnace refractory and four flaps driven by high-power electric actuators. The tanks store propellant, the hull holds pressure, the engines become spares — the same salvage row carried on the Mars matter inventory.
The nearest public statement is Musk’s remark to Popular Mechanics that a Mars cargo ship could be outfitted as a propellant plant and landed as a working plant. The Blue Moon cargo lander is the only lunar vehicle quoted both ways: up to 20 t reusable, 30 t one-way. Modeled
Adjacent entities
Starship Now Sits Inside a Launch, Satellite and AI Company
The 2026 xAI merger and IPO made Starship the launch arm of an orbital-compute business.
| Entity | Relationship | Detail | Rated |
|---|---|---|---|
| xAI | Acquired 2 February 2026 | All-stock deal; brought the Grok models, the Colossus data centers in Memphis and the X social network | Observed |
| SpaceX, Nasdaq SPCX | Public since 12 June 2026 | The S-1 records Starlink at $11.4 billion 2025 revenue — the cash that funds Starship | Observed |
| Starlink V3 | First Starship payload | About 1 Tbps per satellite; only Starship can carry them | Building |
| Starmind · AI1 | Orbital data centers | FCC filing for up to 1 million satellites (30 January 2026). AI1 design of 4 August: 75 m solar array, 30 m radiators, 175 kW average, one Vera Rubin NVL72 rack. Two prototypes due early 2027 | Building |
| Terafab | Chip supply for Starmind | Tesla, SpaceX and Intel fab in Grimes County, Texas; construction slated from 1 December 2026. The S-1 calls the Tesla arrangement a “general framework” with no binding deal | Projected |
| Tesla Optimus | Robotic payload | Was to ride the first Mars cargo ships; now tied to the deferred Mars plan | Projected |
| NASA | HLS customer | Artemis lander contracts; Office of Inspector General flags crew rescue and propellant transfer as open risks | Building |
| US Space Force · AFRL | Rocket Cargo | $102 million five-year contract (2022) to study point-to-point delivery of up to 100 t | Building |
| Commercial payloads | Manifest | SKY Perfect JSAT Superbird-9, Starlab, Vast Haven-2, Astrolab FLEX rover, Italian Space Agency Mars experiments | Projected |
| Regulators and hosts | Licensing and land | FAA, FCC, City of Starbase, Cameron County, Louisiana Economic Development, Space Florida | Observed |
| Suppliers and services | Operations | Linde (industrial gases), Boskalis (heavy-lift recovery) | Observed |
Chinese counterparts
China Caught a Booster Before Block 3 Did
China logged 92 orbital launches in 2025 against about 193 for the United States, 165 of them SpaceX. Its reusable and super-heavy programs are closing on Starship’s functions one at a time.
| Vehicle · builder | Key figures | Status | Rated |
|---|---|---|---|
| Long March 10B · CALT / CASC | 70.2 m · 7 YF-100K · 16 t to LEO with recovery | First flight 10 July 2026 from Wenchang Commercial: reached orbit, and the booster was caught by a hook-and-net vessel in the South China Sea. China’s first orbital booster recovery; reflight targeted before the end of 2026 | Observed |
| Long March 10A · CALT | Crew-rated | First orbital flight with the Mengzhou crew spacecraft planned for late 2026, leading to a crewed lunar landing with the Lanyue lander by 2030 | Building |
| Long March 9 · CALT | 114 m · 10.6 m diameter · 30 YF-215 methalox · 150 t to LEO | In development. The closest Starship counterpart in size and propellant; takes over super-heavy lift from Long March 10 after the lunar landing | Projected |
| Zhuque-3 · LandSpace | Stainless steel, methalox | Reached orbit; first booster recovery attempt failed | Observed |
| Long March 12A · SAST | Reusable first stage | In testing | Building |
Contested
Deliberately Not Claimed
The load-bearing uncertainties, tagged with the confidence they actually earn.
Propellant transfer gates the Moon, Mars and the depot at once. The ship-to-ship demonstration has slipped from March 2025 to March 2026 and is unscheduled as of September 2026. NASA’s estimate for a lunar landing is a tanker sequence in the high teens, flown in quick succession because of boil-off. Until transfer flies, every beyond-LEO row on this page stays in development.
Raptor 3 relight reliability, not structure, is losing boosters. Flights 12 and 13 both lost the booster to engines that failed to light or failed after lighting on boostback and landing. The ships on both flights reached their targets.
Cadence targets are stated, not demonstrated. Musk has cited up to 1,000 and later 10,000 ships a year and more than 30 flights a day from Louisiana. Two Block 3 flights have flown in 2026. The factory and pad buildout is real; the rate is not yet measured.
One-way cargo hulls are an inference, not a plan. The salvage value follows from the propellant arithmetic, but no SpaceX or NASA document assigns a landed cargo hull to habitation, storage or parts.
Record
Provenance and Data Sources
- Confidence
- Observed for flown vehicles, operating facilities and completed flights — Building for facilities under construction and variants in development — Projected for announced plans — Modeled for original analysis
- Status date
- Q3 2026, ahead of Flight 14
- Vehicles and flights
- SpaceX updates; Starship flight records for Flights 12 and 13; NASASpaceFlight pad and Massey’s coverage
- Facilities
- SpaceX Gigabay announcement (March 2025); FAA environmental assessments for LC-39A and Flight 14; Louisiana Economic Development and the 25 August 2026 Starbase Louisiana announcement
- Propellant
- Cameron County air separation unit approval (July 2025); FAA 2022 Starbase programmatic assessment
- Programs
- NASA Office of Inspector General report IG-26-004 on HLS (March 2026); SpaceX S-1 (May 2026)
- China
- CALT Long March 10B first flight and booster recovery (10 July 2026); Long March 9 and Long March 10 program statements
- Weakest link
- Ship-to-ship cryogenic propellant transfer, not yet flown, and Block 3 booster landing reliability