Zhuque-3 and Starship start from the same basic idea: build a stainless-steel rocket, fuel it with methane and liquid oxygen, and fly the first stage back for a landing instead of throwing it away. But they pursue that idea at very different scales.
LandSpace’s Zhuque-3, a Chinese medium-lift rocket that has landed once, completed China’s first leg landing of an orbital-class booster in August 2026. SpaceX’s Starship, the largest rocket ever flown, is still working through a multi-year test campaign toward full reuse and is attempting something much harder: recovering both stages of a vehicle roughly ten times Zhuque-3’s size. The two programs share an engineering playbook, but little else when it comes to scale, budget, or flight history.
The Comparison at a Glance
| Measure | Zhuque-3 | Starship |
|---|---|---|
| Builder | LandSpace (private, China) | SpaceX (private, US) |
| Height | About 65.9 meters | About 120 meters, stacked |
| Structure | Stainless steel | Stainless steel |
| Propellant | Methane and liquid oxygen | Methane and liquid oxygen |
| Payload to low Earth orbit, expendable | 11,800 kg | Far larger, no single figure confirmed as the design iterates |
| Stages designed for reuse | First stage only, on legs | Both stages, booster catch and ship recovery |
| First orbital flight | 3 December 2025 | 2023 |
| Landing record | One successful landing, 18 August 2026, booster later toppled | Multiple booster catches through 2025 and 2026, full ship recovery still in testing |
Why Two Companies an Ocean Apart Made the Same Material Choice
Zhuque-3 and Starship both trade weight for cost, and the logic behind that trade is identical at both companies. Stainless steel is heavier per unit of strength than the aluminum-lithium alloys most rockets use, which costs payload capacity. It is also far cheaper per ton, tolerates the heat of reentry better than aluminum, and can be welded in an ordinary industrial shop rather than a specialized aerospace clean room.
That choice matters most to a company trying to land boosters routinely rather than fly them once. A stage built from commodity steel is cheap enough to risk on a landing attempt and cheap enough to write off if the attempt fails. SpaceX made this trade first, at a much larger scale, and Zhuque-3’s own explainer notes LandSpace reached the same steel-and-methalox conclusion independently on a much smaller vehicle.
The fuel choice follows the same reasoning at both companies. Methane and liquid oxygen burn more cleanly than the kerosene that powers Falcon 9 and most other operational rockets, leaving less soot in the turbomachinery between flights. Methane is also denser than the liquid hydrogen some rockets use, which keeps the tanks smaller for a given amount of propellant. Less soot and smaller tanks both point at the same goal: turning a landed stage around for another flight faster.
The Scale Gap Is the Whole Story
Sharing a material and a fuel does not make Zhuque-3 and Starship comparable machines. Starship stacked with its Super Heavy booster stands roughly 120 meters tall, against Zhuque-3’s roughly 65.9 meters, a difference of nearly two Falcon 9s. Zhuque-3 carries about 11,800 kilograms to low Earth orbit when expended, dropping to roughly 8,000 kilograms when the first stage flies back for recovery. Starship’s payload target has moved as SpaceX iterates the design from one version to the next, and SpaceX has not settled on one figure it stands behind for the vehicle’s current configuration, so treat any specific tonnage quoted for Starship as provisional until SpaceX’s own flight data confirms it.
The gap shows up just as clearly in engine count. Zhuque-3’s first stage clusters nine TQ-12A engines. Starship’s Super Heavy booster clusters roughly 33 Raptor engines, more than three times as many, to lift a vehicle that dwarfs Zhuque-3 in mass and volume. Lighting that many engines at once raises the odds that one fails during ascent, which is one reason SpaceX’s early Starship test flights focused heavily on proving the booster could survive an engine-out event.
Starship Aims to Recover Both Stages. Zhuque-3 Recovers One.
The harder difference between the two programs is not size. It is what each one is trying to bring home. Zhuque-3 recovers only its first stage, the same scope as Falcon 9 and most other reusable rockets flying today. The upper stage that carries the payload the rest of the way to orbit is expended on every Zhuque-3 flight.
Starship is built to recover both stages. The Super Heavy booster separates and flies back to be caught by arms on the launch tower, a step SpaceX has repeated across multiple test flights through 2025 and 2026. The Starship upper stage has the harder job: it reenters from orbital speed, several times faster than a booster that never leaves the lower atmosphere, and has to survive that heating behind a shield of ceramic tiles before attempting its own landing. Recovering that upper stage, not the booster catch, is the piece of the Starship program still being proven out flight by flight.
That distinction is why the two vehicles are not really chasing the same prize. Zhuque-3 proving out a leg landing puts LandSpace roughly where SpaceX stood after Falcon 9’s first booster landing in December 2015. Starship’s remaining test work targets a capability no operational rocket has yet demonstrated at all.
The Landing Record So Far
Zhuque-3 has flown twice and landed once. Its first flight, in December 2025, reached orbit but lost the booster during the landing attempt. Its second flight, on 18 August 2026, delivered a satellite to orbit and set the first stage down upright on legs in Gansu province, the first Chinese orbital-class booster ever to land that way. The stage later tipped over during a post-landing fire while ground crews offloaded leftover propellant, a setback covered in full on the Zhuque-3 explainer.
Starship’s booster has been caught by the launch tower on repeated flights through its 2025 and 2026 test campaign, a sequence that has grown more consistent as SpaceX works through each vehicle iteration. The upper stage’s return to a controlled landing is the harder milestone still being chased. Neither company has yet reflown a recovered orbital-class stage on a second mission, which is the step that would turn either program’s landing into proven reuse rather than a demonstrated one-off.
Who Builds Each Rocket
LandSpace is a private Chinese launch company founded in 2015, and Zhuque-3 is its third orbital rocket after the solid-fueled Zhuque-1 and the methane-fueled Zhuque-2, which reached orbit in 2023. LandSpace filed for a listing on the Shanghai Stock Exchange’s STAR Market at the end of 2025. Its funding and its customer base sit almost entirely inside China’s commercial launch market, which is racing to fill two approved broadband constellations, a demand side covered on the China reusable rockets page.
SpaceX is a private US company that already operates the world’s most-flown rocket in Falcon 9 and derives much of its cash flow from its own Starlink constellation. That existing revenue base funds a Starship test campaign that has absorbed years of flight-test failures without threatening SpaceX’s core launch business, a financial cushion LandSpace does not have to the same degree while it works to reach a similar pace of testing.
What Would Change This Comparison
A second Zhuque-3 landing, without the post-landing fire that toppled the first one, would show LandSpace’s leg-landing method is repeatable rather than a single good outcome. A confirmed, stable payload figure for Starship’s current configuration, published by SpaceX rather than estimated from a prior version, would sharpen the capacity gap in either direction. Both companies reflying a previously landed stage on a second orbital mission would be the clearest signal yet, since neither program has done that as of this comparison.
Readers who want each vehicle’s full mechanics can start with the Zhuque-3 explainer and the Starship explainer. The reusable rockets guide covers the landing, refurbishment, and cost math both companies are chasing at very different scales.