Dinosaurs became extinct because they didn’t have a space program.

Larry Niven

Asteroid 2024 YR4, with its February 2025 estimate of having ~3% chance of striking Earth, was a sharp reminder to heed Stephen Hawking’s warning: that we must become a spacefaring species to survive extinction-level events. He believed we had ~200 years to do so but others estimate the window is under 40. Beyond survival, large crewed spaceships would enable Gerard O’Neill’s vision: shifting heavy industry into orbit — supplied by asteroid resources — to relieve pressure on Earth’s environment. Startups like Varda are already proving space’s material value, producing pharmaceuticals in orbit that outperform Earth-based processes. Such large pressurised volumes wouldn’t just serve human habitation—they would enable in-space manufacturing industries beyond pharmaceuticals and might house space data centers in the near term.

Yet no major institution is currently building spaceships.

Engineering spaceships far larger than the International Space Station (ISS) as insurance for our survival or as engines for economic growth remains speculative — asteroid deflection could be just as important a mitigation tactic but we might also be wrong about economic advantages of industry-scale orbital manufacturing. But one outcome is virtually guaranteed: expanding human presence in space will deliver adventure and consciousness expansion — astronauts speak of the overview effect — for future generations, who will look back at the first spaceships with the wonder that we still feel about the Pyramids of GizaPerhaps it’ll inspire a new classification for Wonders of the Universe. .

Curiosity and the pursuit for meaning are not necessarily rational reasons for doing anything but underly our desire to explore and inhabit the physical universe — it may very much be the only reason to invest in space research at this point.

But what is a spaceship?

In addressing “Where is my von Braun Wheel?”, I’d created a chart visualising crew sizes against a comfort metric using data on real, planned, and conceptual space stations. Over repeated examinations of this chart while writing proposals and discussing it with others, I noticed myself falling into a habit of using “space station” as a catch-all for all these orbital pressure vessels. But just as we distinguish between kayaks, boats, and ships based on their scale and purpose, I sensed that a similarly higher precision nomenclature is needed for different classes of space structures as we aim to scale from modern stations like the ISS towards the space superstructures like O’Neill Cylinders.

While it may seem over-indulgent to do so, I genuinely feel that this isn’t merely semantic — terminology shapes engineering thinking. When every habitable vehicle in orbit is called a “space station,” we end up defaulting to proven station-like construction methods. This lexical imprecision reflects deeper issues in how we approach engineering these systems and is my theory on why almost every idea on making large stations continues to adopt Apollo-era modular architectures, which is how we got the ISS.

The Romans and Vikings didn’t build vessels for hundreds by lashing together rowboats — yet this is essentially how we built the ISS. Even more telling: 11th-century longships carried crews of 30-40, while the modern benchmark for crewed space vehicle carries seven on average and thirteen on a good day. Yet today’s space station designs still rely on this flawed modular design which offer nowhere near the same scale of millenia-old watercraft.

To define a more precise taxonomy, I have since revised the earlier plot to identify three distinct regions corresponding to three orbital structural classes. This reveals that spaceships could be seen as a separate structural class from either smaller space stations (green region) — like the 7-person ISS — or futuristic space superstructures (purple region) — such as the Stanford Torus or O’Neill cylinders housing thousands to millions.

Space stations refer to everything flown so far under that label, setting an upper limit of 25 crew (arbitrarily chosen). Next, I placed outlandishoutworldish? conceptual designs like the Stanford Torus and O’Neill Cylinders to fall under the Space Superstructures category.Note that this revised chart contains more concepts, such as the Bernal Sphere and Kalpana One. The Spaceship territory then is the region between these two categories.

However, the distinctions aren’t merely about size — they reflect fundamentally different construction philosophies. Spaceships occupy the crucial middle ground: large enough to be economically transformative, yet buildable entirely from Earth using current launch systems.

Spaceships would be manufactured on Earth—like stations, but unlike superstructures, which typically require in-situ resource utilisation of lunar and asteroid material. In the near-term, relying on space resources for spaceship building need not be a requirement and, to keep things grounded in practical current capabilities, should be avoided. However, spaceships should be seen as platforms from where a variety of missions to the asteroids are launched.

Further, with this distinction, we can see that the early engineering prototypes of “unitised” spaceships were, in effect, pioneering a field that ought to be called spaceship engineering, a discipline distinct from spacecraft or satellite engineering. Monolithic hulls allowed them to target crews from tens to hundreds from the outset, a scale they believed achievable in single-digit launches. Contrast this to the $150+ billion ISS, built like space-Ikea from eight small “tin-can” modules, hosting just seven astronauts but needing over 40 rocket launches for assembly. Unlike Ikea, this approach has proven costly, complex, and cumbersome. More broadly, spaceship engineering should be seen as the discipline that approaches large-scale orbital structure construction differently — avoiding the trap of starting with modularity in the hull design.

I anticipate that more distinctions in orbital structure classes will emerge within each of these classes but also beyond the superstructures class — for example, a Dyson sphere is classified as a megastructure, which is grander than the O’Neill cylinder that in turn exceeds the Kalpana One; the latter two are often classed space settlements. Thus, I could already modify this plot to contain space settlements, superstructures and megastructures. But I am eschewing this for now — this might matter when we get around to actually seriously building them and develop deeper perspectives — in lieu of focusing on spaceships as I believe their engineering is the critical next step to be taken.

Next, to address how we might go about building such spaceships, I begin by defining a target specimen called the Ideal Spaceship.

The Ideal Spaceship: the next benchmark for crewed spacecraft

Thus far, I have asserted that spaceships should not be built with modular hulls. Instead, monolithic hulls must be developed to sustain large crews and sensitive equipmentAdverse effects of vibrations, radiation, etc. in hostile environments for long-duration stays. This core property suggests that spaceship construction will share more in principle with shipbuilding traditions especially submarines? — than with space stations. But where the centuries-old practice of shipbuilding was only formalised as naval architecture in the 19th century, space’s inaccessibility demands that theoretical engineering principles for spaceships be defined before building and deployment in orbit. In principle, this is what the engineers at Langley were pursuing but with no prior station benchmarks and subsequent shift in priorities to Apollo, the theoretical foundations were never laid and the necessary technology development roadmap to realising such an Ideal Spaceship remains undeveloped. The early prototyping data also is not easily accessed.

Towards laying stronger foundations, the chart identifies an ideal spaceship as a 70-person system and 90-m³ of habitable volume per person. This results in an overall habitable volume of 6300-m³. This is an order of magnitude increase in crew size to the ISS’s 7 while offering the comfort of Skylab, the best in class for a space station as per the chart. It is worth noting that the Ideal Spaceship does not yet have a defined geometry but its volume is remarkably close to that of Wernher von Braun’s wheel-shaped space station. In his design, the shape was the starting point and the corresponding dimensions were derived for spin rates low enough to generate artificial gravity without inducing space motion sickness to its crew.

This “ideal spaceship”, if shaped like a wheel, could also have some artificial gravity but this is not mandatory. The primary intention is to create a monolithic hull with a target volume of 6300-m³ that, when pressurised, is spacious enough for human comfort but this is also large enough for other microgravity applications, like space data centers and in-space manufacturing. Any addition of gravity might only be downstream of these use cases. Though the remainder of the discussion is constrained to zero-gravity applications of a monolithic hull, addition of thrusters to its outer rim to generate gravity would have important ramifications on the hull’s materials due to thruster plume impingement.

While von Braun also assumed modular robotic assembly of the hull from small rigid sections, the aforementioned Langley engineers sidestepped hyper-modular designs. Their implemented solutions were wheels made from large inflatable “bicycle tubes” and pipes arranged as a hexagonal ring. These structures promised not only uninterrupted interior volume but also dramatically lower cost and complexity: even in the 1960s, their estimated costs were two orders of magnitude cheaper when adjusted for 2025 ($1 billion ) while requiring roughly ten times fewer launches than the ISS. If a monolithic hull is indeed possible then one can imagine how we can treat these hulls as modules themselves which could be assembled into something grander; the above chart indicates this concept as 10 Ideal Spaceships. In a sense, the goal is not to eliminate the modularity but to use it only when necessary.Pete Lynn suggests that I re-examine my claim that hulls be monolithic or inflatable; rigid elements are mechanically easier to deal with than inflatables. His perspective is vastly more informed than mine, given his work at OtherLab, whereas I am trying to provoke a different solution to the hyper-modularity I sense is blighting space infrastructure. My hope is we will collectively tease out the best feasible solution over time.

Inflatable hulls, in particular, offer compelling advantages: once deployed in orbit, they can expand to multiple times their launch volume to enable large-scale living and working space. However, the early concept had a high rupture risk with catastrophic consequences; high strength fabrics were yet to be invented in the sixties. So, this idea lost out to the hexagonal station at the time.

The case for returning to inflatables was reinvigorated in 2016, when NASA tested Bigelow’s Expandable Activity Module (BEAM) on the ISS. Constructed from layered Kevlar-like fabrics, BEAM was found to be stronger than the ISS’s own metallic hulls. Its success has, in my opinion, re-opened the doors to revisiting inflatables as technically viable solutions but scaling issues must be addressed — BEAM is a mere 16-m³ when inflated and it is also capsule shaped.

The engineering challenge now is to create an inflatable structure that can compress 6.3:1 (or better) while using materials light enough that the entire Ideal Spaceship (or its hull with some basic motion control systems) weighs under 100 tons when deployed to fit within Starship’s payload bay when stowed. While this could unlock the imagined economic potential of space and the technological insurance policy for humanity’s survival, its near-term technical feasibility remains unclear.

The next section summarises the state-of-the-art open literature on the research and development into inflatable high-strength fabrics — while the scale of the structures is smaller than the Ideal Spaceship, it helps tease out some of the first set of requirements needed for an Ideal Spaceship’s inflatable hull.

How to (maybe) make them?

As per Ref. 1, BEAM is made from a multilayer inflatable shell structure developed within a NASA project, from the nineties, called TransHab or Transit Habitat. Intended to replace the ISS, its pressurised volume was 339.8-m³ — not only close to the ISS’s volume but over 10 times that of BEAM. Its multilayer inflatable shell structure design comprises five parts (also shown in the figure below alongside a concept image of TransHab); from what I understand, something close to this was eventually flight-proven on BEAM. The five layers, from inner habitat wall to outer later exposed to the space environment, are:

1. inner liner layer
2. bladder layer
3. restraint layer
4. micrometeoroid/orbital debris (MMOD) protection layer, and
5. thermal protection (MLI) layer

The remainder of the discussion specifies the desirable properties for a next generation multilayer shell and its two load-bearing parts — the restraint and MMOD protection layers — for the Ideal Spaceship.

Multilayer Shell properties

The table is presented below, from Ref. 1, which benchmarks various inflated habitat modules (in its deployed state) against the metallic systems on the ISS and other relevant crewed spacecraft.

The densities of the shell layers of BEAM and TransHab could be used to estimate the mass of a 5000-m³ von Braun wheel. So, a 5000-m³ von Braun wheel would weigh about 195 tons if made from TransHab’s materials, which would need only 2 Starship launches; if made from BEAM-style materials, 5 launches would be needed. Alternative calculations for the Ideal Spaceship using BEAM’s surface area density suggest its fabric shell alone could weigh as little as 28-75 tonnes depending on orbital debris environment.See detailed methodology, where the von Braun wheel volume is calculated to be 6650-m³ which is close to that of the Ideal Spaceship’s 6300-m³. Volume-based estimates are less optimistic than more simplified fabric mass estimates that are assumed to scale with surface area.Thanks, again, to Pete Lynn on catching poor phrasing for me earlier.

It is important to note that BEAM’s restraint layer is made from Vectran, which is five times stronger than steel and ten times stronger than aluminium while being half as dense (1400 kg/m³) as aluminium. If made purely from Vectran, the 5000-m³ von Braun wheel would weigh 7000 tons; this translates to 70 Starship launches! But the multilayer shells developed so far are a more optimised material. Further, as Vectran’s density is lower than either of the metals, it’s easy to see that a spaceship made from it would always be lighter than a comparable metallic one.

Though much other public data on these systems cannot be found, I suspect a next-generation multilayer inflatable design must be developed to can fit in one Starship launch. I identify this materials aspect as the first challenge.

It is also important for us to consider packaging properties in the deflated state. If we assume BEAM packs as a cylinder when deflated (these cylindrical dimensions are public), its packed density works out to 157 kg/m³. TransHAB’s theoretical packaged density of 122 kg/m³ is even more impressive. From these, we can determine their compression ratios — the ratio of pressurised to unpressurised volumes — for BEAM and TransHab as 1.78 and 3.12, respectively. So, the Ideal Spaceship should aim for an even more improved packaging density — this is identified as a second challenge.

Restraint Layer properties

The restraint layer is the main load-bearing layer of the multilayer shell on account of the post-inflation loads and stresses. The chart below shows different shell designs at varying scales. In the blue region to the left are simple single-layer bladders with lower loads (e.g., balloons, beachballs). The chart progresses onto higher pressure systems needing restraint layers. The first is the green region which has a broadcloth restraint layer that is either coated or contains a separate bladder to act as a combined restraint with bladder layer (e.g., blimps, basketballs).

Then, we come to the specific space applications at higher loads (e.g., an inflatable radomes, airlocks); these have a separate bladder with additional loose webbing or cordage reinforcing the restraint layer. Even greater loads on full-scale space modules (e.g., TransHab, BEAM) are managed using a bladder and restraint layer with tight webbing.

From a weave perspective, the authors state that NASA JSC’s designs fall within the tight webbing category; here, high-strength webbings are woven in a tight basket weave pattern in the axial and hoop directions of the module. The authors further state that Bigelow Aerospace patented another restraint layer design, which results in a lighter habitation module with fewer longitudinal, or axial, straps. This restraint layer is “made of hoop webbings that are abutting and sewn end-to-end lengthwise, which reduces the potential stress points on a strap weave”.I need to learn what this means.

BEAM’s restraint layer, made from Vectran, was designed for an operational pressure of nearly 60-psi; this is determined from the product of the factor of safety of 4 and operational pressure, which on the ISS atmospheric pressure (14.7 psi). The factor of safety is dictated by NASA structural design standard NASA-STD-5001Ultimate Design Factor in Table 6. . As per these requirements, it is also mandated that the habitat’s survival lifetime at the operating pressure of 14.7 psi exceed the intended duration by the same factorSee Section 4.4, page 24, on Fatigue and Creep. , i.e., 60 years survival if the intended mission duration is 15.

Note that the chart above tells us that the loads experienced per unit length of the restraint layer increases with the diameter of a cylindrical shell — these loads are calculated from cylindrical hoop stress. What this behaviour looks like in practice for larger toroidal structures is unknown.This claim is based on discussions with co-authors of Ref. 1, Doug Litteken at NASA Johnson and Tom Carno at Langley. Further, the chart tells us that TransHab’s loads were managed by weave patterns that scaled to a certain diameter of about 8.2-m at the specified loads. New restraint layer designs were patented by Bigelow for its 2250-m³ BA-2100, which within sits the grey region. Whether new materials were needed here are unclear to me but the Ideal Spaceship sits within this region — the wheel-shaped Ideal Spaceship is about 75-m diameter in the first chart of this blogpost. If the claims made by Ref. 1, that novel restraint materials are needed beyond 9-m inflatable cylindrical structures due to higher loads are true, then this also applied to our Ideal Spaceship as we are also dealing with a different geometry — what works on cylinders, which is what has been documented so far, might not necessarily apply to toruses. So I separate this into two challenges below (based on scale and geometry).

I believe both challenges might be feasibly addressed computationally. Specifically, I believe AI-led materials discovery and novel restraint layer weave pattern designs could help identify promising solutions to large inflatable shell structures.

MMOD layer

Ref. 1 states that 68% of TransHab’s mass was allocated for MMOD protection due to higher debris density in its proposed operation in LEO but drops to 14% for deep space operations. This MMOD mass fraction has significant architectural implications. For example, if the Ideal Spaceship operates in LEO with integrated MMOD protection, ~132 tonnes of shielding is needed if we use 195 tonnes as hull mass from our volume-based estimate. A solution where an inflatable structure is launched with minimal protection in one Starship launch followed by a separate mission for robotic assembly of MMOD panels around it might be another solution.

The Ideal Spaceship’s chosen orbit would most likely be LEO — to take maximal mass to orbit in a single Starship — which makes me believe that there may be a need for such in-orbit assembly of Whipple shield panels for MMOD around the restraint layer. This might be achievable with the current state of robotics but would still need clever mission architectures and structural design for robot locomotion — another benefit here is that we might invent a new inflatable material (independent of MMOD shielding) that can be used in other environments or in other orbits with lesser debris. The counter-argument to this is that we would need to integrate hard-points onto the fabric for attaching the MMOD protection panels which might not be straightforward. This leads to the fifth (and for now last) challenge.

I will stop here for now and will likely come back and revise this blogpost in future while writing separate pieces to address other challenges pertaining to design and development of the Ideal Spaceship.

Conclusion

Here, I’ve answered the main question in the title — a spaceship is a distinct large orbital structure class sitting between today’s modular space stations and tomorrow’s superstructures. Developing an ideal spaceship capable of hosting nearly a hundred crew members within a single monolithic hull was identified as a target. NASA Langley’s 1960s large rotating wheel concepts — abandoned in the Apollo pivot — were the first and only real examples of spaceship engineering with large monolithic hulls of scale to date. Several such hulls could be interlinked to enable crews of a thousand or more. Specifically, their inflatable approach remains a promising path forward.

But to make spaceships a reality, we must identify a set of concrete engineering challenges and solve them. Towards this end, I have outlined a handful of specific challenges with creating lighter, stronger multilayer shells, scalable restraint designs, advanced packaging, and modular debris protection were identified.

These aren’t just technical hurdles — they’re invitations. The next wave of spaceship engineers can decide whether humanity’s future in orbit remains cramped in tin cans or grows into vast, city-sized habitats.

Acknowledgements

Thanks to Pete Lynn for the many emails we have exchanged on this topic and also for reading an early draft of this essay. All errors of interpretations of our exchange remain my own but this post may continue to be revised as I learn more.

References and Further Readings

[1] Paper System Integration Comparison Between Inflatable and Metallic Spacecraft Structures from NASA Langley Research Center and NASA Johnson Space Center.

[2] Slide deck TransHab Materials Selection from NASA-Johnson Space Center.

[3] Paper Design of a Microgravity Hybrid Inflatable Airlock from NASA Johnson Space Center.

[4] US Patent Flexible structural restraint layer for use with an inflatable modular structure held by Bigelow Aerospace