SpaceX Starship

As of May 2026, no Starship has flown more than once. Every orbital test has ended in loss—by design. Across seven integrated flight tests (IFT-1 to IFT-7), SpaceX has destroyed $500+ million in hardware to compress decades of development into years. The bet: that full reuse of the largest rocket ever built will cut launch costs by 99% and unlock Mars. That bet remains unproven.

Starship is a two-stage, fully reusable system: Super Heavy (booster) and Starship (upper stage). At 121 meters tall and 9 meters in diameter, it surpasses Saturn V. Its 39 Raptor engines—33 on the booster, 6 on the ship—burn liquid methane and oxygen in a full-flow staged combustion cycle, the first engine of this type ever to fly. Methane was chosen not for performance but because it can be synthesized on Mars via the Sabatier reaction, using atmospheric CO₂ and subsurface water ice. Without in-situ propellant production, return missions from Mars are impossible.

The architecture: stainless steel, radical reuse, chopstick catch

Starship’s material choice stunned aerospace engineers. In 2019, SpaceX abandoned carbon fiber for 301 stainless steel—a metal last used in orbital rockets in the 1960s. Why? At cryogenic temperatures, steel’s strength-to-density ratio exceeds carbon fiber. It also tolerates high reentry heat (up to 1,650°C) and costs $3/kg versus $130/kg for aerospace-grade composites.

Super Heavy launches the stack to ~70 km, then separates. It flips, burns back toward the launch tower, and is caught mid-air by mechanical arms—“chopsticks”—on the launch mount. Starship continues to orbit, then reenters nose-first at Mach 25. It rolls to belly-flop, slows with flaps, then flips and lands vertically. A successful catch reduces turnaround time and refurbishment cost. The first successful chopstick catch occurred on IFT-5 in October 2024.

All 39 engines ignite at liftoff, producing 7,600 tonnes of thrust—double Saturn V’s. This brute-force approach demands a launch pad that can survive it. After IFT-1 obliterated its concrete foundation in April 2023, SpaceX installed a steel deflector plate and 200 water-cooled nozzles. The new system held through IFT-7 in February 2026.

Progress by numbers: milestones and losses

2019: Starhopper prototype hops 150 meters on one Raptor engine.
2020–2021: Ten high-altitude prototypes (SN5 to SN15) fly. Six explode—on ascent, during landing burns, or from fuel slosh.
April 2023 (IFT-1): First orbital attempt. Vehicle tumbles; range safety destroys it at T+4 minutes. Pad devastated.
November 2023 (IFT-2): Reaches stage separation. Booster explodes on relight.
March 2024 (IFT-3): Starship reaches space, relights engines in coast, survives 7 minutes of reentry heating.
June 2024 (IFT-4): First soft splashdown of Starship. Booster nearly catches—misses by 3 meters.
October 2024 (IFT-5): First successful chopstick catch of booster. Starship breaks apart during reentry.
February 2026 (IFT-7): Starship survives full reentry arc. Splashes down under roll control. Data confirms 80% of heat shield tiles intact.

Despite progress, no stage has reflown. Refurbishment timelines remain classified. SpaceX targets 100 flights per year per vehicle by 2030. Falcon 9 boosters average 20 flights; no rocket has exceeded 30.

What's contested: Is full reuse economically viable?

The Space Shuttle flew 135 missions but never achieved its promised reusability. Each orbiter required 50,000+ hours of inspection and $1B refurbishment per flight. Starship aims for 48 hours between flights and $10M per launch—$2M for propellant, $8M for operations.

Critics point to the heat shield. Starship uses 18,000 hexagonal tiles of silica ceramic, each hand-installed. During IFT-5, 30% detached. SpaceX is testing welded metallic heat shield panels for future ships. Whether either system can survive 100+ reentries with minimal maintenance is unknown.

Elon Musk has claimed “a 50% chance of reaching orbit with full reuse by 2026.” As of May 2026, that threshold remains uncrossed. The FAA licenses only one Starship launch per month due to environmental and safety review backlog—12% of the target cadence.

Why this has to do with other realms

Starship’s ambition mirrors the scaling laws in computing. In AI, performance scales with compute, data, and model size—all multiplicative. Starship assumes a similar power law: launch cost per kg falls not incrementally but exponentially with flight rate. At 100 flights per booster, cost could drop below $100/kg to orbit—versus $2,720/kg on Falcon 9. This would make space-based solar power, giant telescopes, and interplanetary logistics viable. It also risks accelerating concept kessler syndrome—if thousands of Starships launch annually, orbital debris management becomes urgent. The same physics that enables access also threatens to clog it.

An open question

If a rocket must fail often to learn fast, at what point does failure become a liability—financial, regulatory, or public?

Key sources

to verify: Raptor engine full-flow staged combustion details – public data limited; based on SpaceX presentations and Sutton’s Rocket Propulsion Elements.
to verify: Sabatier reaction on Mars – NASA Mars Direct architecture studies.
to verify: Starship heat shield tile count – observed from SpaceX launch webcasts and 3D renders.

Abhishek