"Rashomon" originally referred to a city gate, but over time, it has evolved into a metaphor for the complexity and ambiguity of truth.

In Akira Kurosawa's iconic film, multiple characters recount the same event, each with a complete and seemingly convincing narrative. Yet, these accounts contradict one another, making it impossible to determine the absolute truth. The issue is not merely deception; rather, each individual selectively presents facts that support their perspective, resulting in partial truths that obscure the full reality.

The capabilities of commercial rockets are similarly ensnared in such a "Rashomon" scenario.

The same rocket can exhibit varying capacities depending on whether it is used for single-use, in-flight recovery, or return recovery missions. Companies may announce current capabilities based on initial flight configurations or showcase long-term goals with complete configurations. Even within the realm of "Low Earth Orbit (LEO)," some specify altitudes such as 200 km or 450 km, while others simply state LEO without further detail. While these figures may not be false, they often contradict each other when compared directly.

Each company presents its own version of rocket capability, each with conditions that justify its claims. However, once orbital parameters, configurations, recovery methods, and verification statuses are stripped away, the remaining tonnage becomes less of a comparable engineering parameter and more of a solitary promotional label.

The true "Rashomon" in commercial space launch capabilities lies not in outright deception, but in the selective presentation of partial truths that favor each company's interests.

01

Rocket capability is often perceived as a fixed value, akin to a car's maximum horsepower. In reality, it is more akin to a performance curve that varies with target orbit, launch site, rocket configuration, fairing specifications, recovery plan, and mission margin.

LEO is not a singular orbit but a vast region extending from near-Earth space to approximately 2,000 km in altitude. Numerous spacecraft operate at various altitudes within this range: space stations at around 400 km, large-scale low-orbit communication constellations at 450 to 600 km, and remote sensing and Sun-synchronous orbit missions distributed between 500 and 800 km. The 200 km, 450 km, and unspecified LEO figures in rocket parameter tables, although all falling under the LEO umbrella, represent different mission conditions.

Beyond orbital altitude, inclination also influences capability. Sun-synchronous orbits, typically LEO, are near-polar and require different velocity costs for the rocket, considering launch site latitude, launch azimuth, and flight zone restrictions. Launching a satellite into a 200 km low-inclination orbit versus a 500 km or 700 km Sun-synchronous orbit presents different challenges, despite both being low-orbit launches.

When domestic rockets announce their capabilities, there are indeed variations in metrics and levels of disclosure. The Long March 12 announces an LEO capability of no less than 12 tons and a 700 km Sun-synchronous orbit capability of no less than 6 tons, without specifying the LEO altitude in this set of public parameters. The Zhuque 3 uniformly labels its single-use, in-flight recovery, and return recovery capabilities under 450 km LEO conditions. Both the Lijian 2 and Tianlong 3 clearly announce their 500 km SSO capabilities, while their LEO capabilities listed on official websites do not specify altitude.

This lack of a common reference is evident. Some announce 450 km LEO, some 500 km SSO, some use 700 km SSO, and others only write LEO without altitude. All are termed capability, but the disclosed boundary conditions are incomplete.

Therefore, what the industry truly needs to standardize is not a single capability figure for all rockets but rather that each figure must be accompanied by its orbital conditions. A tonnage specified with 450 km orbital conditions cannot be directly compared to another without a specified LEO altitude, even if the numbers are close, to determine which has greater capability.

Beyond orbital metrics, recovery methods further diverge capabilities. According to the complete configuration specifications published on LandSpace's official website, under 450 km LEO conditions, the Zhuque 3's single-use capability is 21.3 tons, its in-flight recovery capability is 18.3 tons, and its return recovery capability is 12.5 tons. On the same orbit and with the same rocket, merely changing the first stage's recovery method can result in an 8.8-ton difference in maximum capability.

This is the capability cost that rocket recovery must pay, encompassing propellant needed for return, deceleration, and landing, as well as the structural weight added by landing legs, grid fins, and related recovery systems. Single-use rockets do not need to reserve fuel for the first stage's return, usually having the highest capability; in-flight recovery lets the first stage fly downstream to a recovery area along the launch trajectory; return recovery requires the first stage to increase speed in reverse and return near the launch site, thus sacrificing more payload capability. Therefore, the statement "a reusable rocket has a capability of 20 tons" is inherently incomplete: it may refer to the maximum capability without recovery or a design value under a certain recovery state.

Configuration changes add another layer of complexity. The Zhuque 3 Y1 uses an initial flight transitional configuration, with its body length, engine thrust, and propellant loading scale not yet reaching the planned complete state. After the Y1 flight, its internal personnel stated in public interviews that the current version's single-use capability is about 13 to 14 tons, and its reusable capability reaches the 10-ton level, depending on the orbit.

This does not simply represent a true-false relationship with the official website's 21.3 tons and 18.3 tons. The former addresses what the current transitional configuration can do, while the latter describes what the complete configuration plans to achieve, merely both using the name "Zhuque 3". The company discusses the upper limit of the complete configuration, while flight verification tests the reality of the transitional configuration. If the two configurations are not distinguished in communication, the long-term capability of the complete configuration can easily be superimposed with the initial flight facts, giving the impression that it has already been achieved.

The Lijian 2 plans to cover a wider capability range through configurations with no boosters, dual boosters, and quadruple boosters, following the same logic. It represents the capability boundary of a rocket family, not meaning that a single fixed configuration can freely switch across the entire range. The model name remains unchanged, but the body, number of engines, and takeoff mass have already differed.

Often, the capability figures are not wrong; what is quietly changed is the object of comparison.

02

Capability figures also have their own histories and contexts.

Some are derived from overall plans and trajectory calculations, some have undergone engine tests, structural trials, and full-system verifications, some are supported by real flight telemetry, and others have already been delivered in client missions. All can be termed capability, but they clearly should not have the same credibility level.

A rocket yet to have its initial flight can form a credible design capability through engineering calculations and ground tests, without needing to hang an equivalently massive weight to prove all performance. Similarly, a rocket carrying only a few hundred kilograms in its initial flight does not mean it can only transport a few hundred kilograms. Commercial launches depend on client orders, and rockets rarely fill their remaining space precisely like a truck loading sand each time.

Therefore, not being fully loaded does not prove false labeling, nor does a single light-load orbit insertion prove alone that the maximum capability has been achieved. Whether the design capability is credible requires comprehensive evidence, including overall design, ground tests, flight telemetry, propellant margin, and orbit insertion accuracy. What truly warrants caution is writing design goals in the present tense, portraying long-term complete configurations as current service states, and then interpreting a single light-load success as the maximum capability having been flight-verified.

At this point, capability is no longer just an engineering parameter but also a rhetorical device that can stretch or shrink as needed for communication.

Multiple satellites per launch represent another capability easily misled by quantity.

On June 7, 2023, the Lijian 1 Y2 completed a "26-satellite" launch, setting a new Chinese record for multiple satellites per launch. Eight days later, the Long March 2D sent 41 satellites into their respective orbits, raising the record to "41 satellites per launch". Globally, the Falcon 9 sent 143 satellites into orbit in a single mission during Transporter-1 in 2021, still holding the world record for the number of satellites launched by a single rocket.

The records are striking but cannot be used to rank rocket capabilities. When the Falcon 9 performs different generations of Starlink missions, the number of satellites carried per mission drops from about 60 to over 20. The Lijian 1 Y2 launched 26 satellites, while the Yinli 1 Y4 launched 9. Judging by the number, the rockets seem comparable, but their actual capabilities are not on the same level. Conversely, Transporter-1 launching 143 spacecraft does not mean it was the Falcon 9's mission with the highest payload mass, as this rideshare mission mainly consisted of numerous light and small payloads.

The relationship is often reversed: high capability usually raises the upper limit of multiple satellites per launch, allowing more satellites, but more satellites do not necessarily prove greater rocket capability.

The number of satellites in a multiple-satellite launch is constrained not only by the total payload mass but also by single-satellite mass and volume, fairing space, satellite-rocket adapters, separation mechanisms, and target orbits. Twenty satellites may weigh only a few hundred kilograms or several tons; a two-ton payload can be one large satellite or dozens of small ones. The same number of satellites does not mean the same capability; more satellites do not necessarily mean greater deployed mass.

What a multiple-satellite launch truly proves is not just that the rocket can "carry more" but also includes layout within the fairing, multi-satellite adaptation, separation sequencing, attitude control, satellite-rocket joint testing, and multi-client mission organization. The challenge of the Long March 2D's "41 satellites per launch" lies not in simply adding 41 satellites but in utilizing limited space for multi-layer layout and then avoiding mutual interference through batch and timed separations.

From a commercial perspective, multiple satellites per launch also mean the rocket has "rideshare" capability, organizing multiple clients and different-sized satellites into the same mission, improving capacity utilization and sharing launch costs. However, more passengers only indicate stronger seating arrangement, scheduling ability, and operational efficiency, not necessarily a greater vehicle payload capability.

Multiple satellites per launch count deployed objects, while rocket capability measures deployed mass. The former proves mission adaptability and commercial organization ability but cannot replace tonnage, orbit, and delivery quality as another, more lively capability unit.

03

The confusion in capability metrics ultimately affects another, more sensitive figure: cost per kilogram of launch.

Assuming a rocket's body manufacturing and basic launch organization costs total 300 million yuan, with a designed maximum capability of 20 tons, dividing these figures yields a cost of 15,000 yuan per kilogram. This figure is attractive but assumes a premise that may not exist in reality: every mission can launch at full capacity corresponding to the maximum capability's configuration and orbit.

If a mission actually deploys 12 tons of payload, the average cost per kilogram based on deployed mass rises to 25,000 yuan. This calculation does not yet include costs for tracking, control, insurance, and payload adaptation, nor does it distinguish between rocket manufacturing cost, whole-rocket launch price, rideshare price, and the final contract price paid by the client.

Low-orbit constellations are important not just for providing numerous satellites but also because batch production, unified interfaces, and continuous batching can improve rocket payload utilization. Without stable orders, even large design capabilities may take off with empty seats; without achieving recovery and reuse, using future mature-state costs to explain today's single-use launch prices is also prematurely consuming engineering progress not yet completed.

If the denominator uses the maximum capability of the long-term complete configuration and the numerator uses the target cost after mass production and reuse maturity, an extremely low cost per kilogram can be calculated. Mathematically, this is correct, but the rocket state corresponding to this price may not have flown yet, nor may clients truly be able to purchase it.

The nominal capability supports the nominal cost, which in turn supports order expectations and enterprise valuations. The entire narrative seems interconnected, but its starting point may be just a number stripped of orbit, configuration, and verification status.

In April 2026, the China National Space Administration (CNSA) and the State Administration for Market Regulation released the Commercial Space Standard System (1.0), setting up 6 primary branches, 32 secondary branches, and planning over a thousand standard projects. The previously released Action Plan for Promoting High-Quality and Safe Development of Commercial Space by the National Space Administration (2025-2027) also explicitly proposed improving the standard specification system for commercial space.

The industry has initiated the construction of a foundational framework of standards. However, when it comes to capacity—the most prevalent yet frequently misconstrued parameter—there remains a notable absence of market-oriented, publicly accessible guidelines that facilitate direct, side-by-side comparisons.

What truly requires standardization in the future may not necessarily be a mandatory capacity metric itself, but rather a comprehensive set of conditions underlying that metric. These conditions encompass: the specific rocket version, the launch site utilized, the orbital altitude and inclination achieved, the type of fairing employed, whether recovery operations are conducted, the location of recovery, and whether the capacity pertains to a design objective, a value supported by ground testing, a value derived from flight data, or an actual delivered performance already accomplished.

New energy vehicles, at the very least, specify CLTC or WLTC conditions alongside their range figures, as range loses its comparative significance in the absence of standardized test conditions. Naturally, rocket capacity is even more intricate than vehicle range, lacking unified test conditions. Yet, this very complexity underscores the heightened necessity for disclosing all boundary conditions employed in calculations. Commercial rockets navigate more complex orbital dynamics yet often present only a solitary tonnage figure.

The commercial space sector demands not only greater capacity but also a requirement for every capacity figure to be accompanied by its respective conditions and historical context. Design values and long-term configurations can be deliberated upon, but not under the guise of the same model name, prematurely adorned with the attire of current service capability.

Various public sources offer their own interpretations of capacity, each backed by conditions that justify their stance. Nevertheless, once orbital parameters, configurations, recovery methods, and verification statuses are systematically stripped away, the remaining tonnage ceases to be a comparable engineering parameter and instead becomes a solitary promotional label.

The true "Rashomon effect" in commercial space launch capacity does not necessarily stem from false statements, but rather from truthful assertions that neglect to mention their conditions of validity.

This article marks the second installment in the "Demystifying Commercial Space" series. Moving forward, our focus shifts from rockets to their launch pads: with public launch pads experiencing queues and dedicated pads undergoing expansion, does every commercial rocket genuinely necessitate its own dedicated launch platform?