Megastructures: Ringworlds and the Engineering of Impossible Scales

Larry Niven's 1970 novel Ringworld described a band wrapped around a star at the distance of Earth's orbit, 1.6 million kilometers wide, with the habitable area of about three million Earths. Megastructures like it, along with O'Neill cylinders, Bishop rings, and Banks Orbitals, share one design rule that physics imposes: spin supplies the gravity, and the spinning material has to hold itself together. The numbers below show where the real limit sits and why a ringworld fails on strength long before it fails on anything else.

Key takeaways

• A spinning habitat gets its gravity from rotation, so the rim must move at v = √(g × r). At Earth's orbital distance that is about 1,211 km/s, and one turn takes roughly nine days.
• The rim material has to carry its own weight. The largest self-supporting 1 g hoop has a radius equal to the material's specific strength divided by g: about 16 km for steel and about 4,900 km for the best nanotube figure quoted.
• O'Neill cylinders and the Stanford torus were studied seriously by NASA-backed teams in the 1970s. Bishop rings need nanotube-class materials that exist only in laboratory samples.
• A Niven ringworld needs strength about 30,000 times beyond the best nanotube figure. No amount of extra material fixes that, because the extra material adds its own load.
• The hard limits are strength first, then mass, then stability. None of them is a law of nature, which is why the idea keeps surviving.

What counts as a megastructure, and what actually limits one

An engineer would call something a megastructure when its size is measured in thousands of kilometers, or when it exists to house millions of people. The label covers a wide range. At one end sits a 1.8 km wheel that a summer-study team of professors and students sketched in detail. At the other sits a ring larger than Earth's entire orbit, which no one has sketched in any detail that survives arithmetic.

Three constraints sort the whole family. Strength asks whether the material can carry the load. Mass asks whether there is enough matter in reach to build the thing. Stability asks whether the structure stays where you put it. Strength is like the rope in a tug of war: it does not matter how many people pull if the rope snaps. Mass is the shipping bill, and stability is the problem of balancing a pencil on its point.

The table below lays out the four habitat designs this article compares, from smallest to largest. The Dyson swarm gets its own treatment later, because it solves a different problem.

Design	Proposed by	Rough size	Status
Stanford torus	1975 NASA/Stanford summer study	1.8 km wide wheel, 10,000 residents	Detailed paper design; never built
O'Neill cylinder (Island Three)	Gerard O'Neill, 1974 and 1976	Two cylinders, about 8 km wide and 32 km long	Paper design; never built
Bishop ring	Forrest Bishop, 1997	About 1,000 km radius, 500 km wide	Concept; needs materials not yet in bulk form
Banks Orbital	Iain M. Banks, Culture novels	About 3.7 million km across	Fiction; relies on forcefields
Niven Ringworld	Larry Niven, 1970	1 AU radius, 1.6 million km wide	Fiction; fails the strength test

The spin equation that sets every habitat's size

Nothing in these designs relies on exotic physics for gravity. A rotating ring pushes its occupants against the inner surface, just as a bucket swung over your head pins the water to the bucket's bottom. The required rim speed follows from the same relation that governs a car on a curve: the inward acceleration equals speed squared divided by radius. Set that acceleration to g, 9.81 m/s², and the rim speed is v = √(g × r). One full turn then takes 2π × r / v.

Spin has a price, which is tension. Picture a hammer thrower: the athlete feels the ball trying to fly outward, and the wire carries that pull. A spinning ring is a wire made of its own material, and every slice of it tugs outward on its neighbors. For a thin ring the stress in the material is its density times the rim speed squared, σ = ρ × v². Divide both sides by density and you get a clean test: the material's specific strength (strength per unit of density, in joules per kilogram) must be at least v².

That test gives a hard ceiling. Because v² = g × r, the largest radius a uniform hoop can reach at 1 g, supporting nothing but itself, is R = S / g, where S is the specific strength. A chain hanging from a hook behaves the same way. Past a certain length the top link carries more than any link can hold, and no thicker chain helps because the extra links add their own weight.

For the numbers, this article uses two reference materials. High-carbon steel has a specific strength of about 154 kN·m/kg, which is 154,000 J/kg. Carbon nanotubes have a best-case specific strength quoted at up to 48,000 kN·m/kg, which is 48 million J/kg. That nanotube figure comes from measurements on single tubes. Bundles and bulk fibers fall to a few gigapascals because the tubes slip against each other, so the figure is a ceiling rather than a product you can buy by the ton.

Steel limits a 1 g spinning hoop to a radius of about 16 km. The best nanotube figure stretches that to about 4,900 km. A ringworld at 1 AU is about 30,000 times beyond even that.

The table applies the spin equation to each design. Rim speeds and spin periods come from v = √(g × r) at g = 9.81 m/s²; the strength column is v² in kJ/kg. The last column compares that to the 48 million J/kg nanotube ceiling, so values below 1 are inside the limit and values above 1 are outside it.

Design	Radius	Rim speed	One turn	Strength needed (kJ/kg)	Multiple of nanotube ceiling
Stanford torus (1 rpm)	0.9 km	94 m/s	60 seconds	8.8	0.0002
O'Neill Island Three	4 km	198 m/s	127 seconds	39	0.0008
Bishop ring	1,000 km	3.1 km/s	33 minutes	9,800	0.2
Banks Orbital	1.85 million km	135 km/s	24 hours	18,100,000	380
Niven Ringworld	150 million km	1,211 km/s	9 days	1,470,000,000	30,600

Two patterns jump out. The first three rows live comfortably below the ceiling, and Bishop's ring sits at one fifth of it. The last two rows are not near the edge at all. They are hundreds to tens of thousands of times past it.

O'Neill cylinders and the Stanford torus: the designs NASA's teams actually studied

Physicist Gerard O'Neill laid out the idea of large rotating settlements in a 1974 Physics Today article and expanded it in his 1976 book The High Frontier. His "Island Three" design is a pair of counter-rotating cylinders, each about 8 km across and 32 km long (some descriptions use 6.4 km for the diameter). Two cylinders spinning in opposite directions cancel each other's gyroscopic effect, so the pair can keep pointing at the Sun while still spinning.

Each cylinder spins about 28 times an hour to produce Earth-normal gravity on the inner surface, which matches the 127-second turn in the table. The inside is divided into six equal-area strips running along the length. Three are land and three are windows, with large hinged mirrors behind the windows that reflect sunlight in and swing shut to make night. By simple geometry the whole interior surface is about 800 km², so roughly half of it, near 400 km² per cylinder, is habitable ground. The Island Three atmosphere in O'Neill's description holds oxygen at about 20 percent of Earth's sea-level pressure, with nitrogen adding about 30 percent more, so the total is roughly half of Earth-normal and the air load on the structure is lower than a full Earth atmosphere would impose.

The size is not arbitrary. Astronomical-engineering summaries note that O'Neill picked the scale by estimating the largest 1 g cylinder that could be built from steel. The table shows why that works: the 4 km radius needs about four times less specific strength than steel offers, which leaves room for a safety factor and for the load of soil, air, and people.

The 1975 summer study went further. It was a ten-week engineering program at Stanford University and NASA Ames Research Center, with participants from physical science, architecture, engineering, and social science. Its report, Space Settlements: A Design Study (NASA SP-413, edited by Richard D. Johnson and Charles Holbrow, published in 1977), compared spheres, cylinders, and a torus, and chose the torus because it needed the least mass for the same living area and spin radius. The result, the Stanford torus, is about 1.8 km across, turns once per minute for roughly 0.9 to 1 g, and houses 10,000 permanent residents at the Earth-Moon L5 point. The Lagrange points article explains why those spots in space hold objects in place.

The most surprising number in the torus design concerns weight. The total mass is about 10 million metric tons, and about 95 percent of that is the radiation shield: a layer of raw lunar soil 1.7 m thick. The structure people picture, the spinning wheel with its spokes, is the small part. The bulk of the cost is a blanket of dirt that stops cosmic rays.

The plan also did not call for lifting most of that mass from Earth. It proposed mining the Moon, firing material with a mass accelerator to a catcher at the L2 point, and carrying it to L5 for processing, with only materials missing from the Moon coming from Earth. A modern version of the same logic leans on asteroids, a business the asteroid mining economics article treats with proper skepticism.

Nobody has built one. That is the honest status of the whole family: the O'Neill and Stanford designs passed their own paper tests, and no space agency has funded the first mile of construction.

Bishop rings and Banks Orbitals: the middle of the ladder

Forrest Bishop of the Institute of Atomic-Scale Engineering proposed the Bishop ring in 1997, in a text titled "Open Air Space Habitats." His ring has a radius of about 1,000 km and a width of about 500 km. Multiply the circumference by the width and you get about 3 million km² of inner surface, which Wikipedia's summary compares to Argentina or India. It would have no roof. The air is held in by the artificial gravity and by retention walls about 200 km high at the rims, much as water stays in a trough on a spinning platform.

The structural material in Bishop's proposal is carbon nanotubes instead of the steel O'Neill assumed. The table explains the choice: the ring needs about 9,800 kJ/kg, which is 64 times more than steel can give and about one fifth of the best nanotube figure. In plain terms, Bishop's ring is inside the limit only if you can make bulk material that performs near the best single-tube measurements. Today's bulk nanotube fibers do not. A ring of this size is a research goal for materials science, not an engineering project.

Bishop suggested lighting by mirrors that reflect sunlight inward, or by a central light source, with the ring at the Sun-Earth L4 or L5 points, closer to the Sun, or in the asteroid belt. The ring would spin once every 33 minutes by the arithmetic above. The air alone is a serious mass. At Earth-normal pressure there are about 10,300 kg of air above every square meter, so 3 million km² of floor holds roughly 3 × 10^16 kg of atmosphere before you add soil, water, or the ring itself.

Iain M. Banks's Orbitals in the Culture novels are the next rung up. Wikipedia's summary gives the diameter as about 3.7 million km, nearly five times the diameter of the Moon's orbit. An Orbital turns once about every 24 hours, which gives the inhabitants an Earth-length day. That detail checks out in the table: a radius of 1.85 million km at 1 g gives a 24.0-hour turn, so Banks's numbers are internally consistent. The strength needed, though, is about 380 times the nanotube ceiling, and in the novels forcefields hold the structure together. Each Orbital also has its own Mind that runs its systems, and the fiction describes Orbitals as habitats for billions of people.

For a natural comparison, the planetary rings article covers Saturn's rings, which are flat structures held up by orbital motion instead of material strength. A ring built to spin faster than orbital speed, as every habitat above does, has to pay for that choice in tension.

Niven's Ringworld: where the arithmetic breaks

Niven published Ringworld in October 1970 through Ballantine Books. It won the Nebula Award for Best Novel for 1970 and the Hugo and Locus Awards in 1971. In the book the ring orbits a star at roughly Earth's distance, is about 1.6 million km wide, and produces about 99 percent of Earth's gravity through rotation. The inhabited inner surface is equivalent to about three million Earths.

You can check the area in one line. The circumference at 1 AU is about 940 million km, and multiplying by the 1.6 million km width gives about 1.5 × 10^21 m². Earth's surface is about 5.1 × 10^14 m² (computed from a mean radius of 6,371 km, per NASA's Earth fact sheet), so the ratio is close to 3 million. The ring would also catch light. Spread over a sphere at 1 AU, the ring's area covers about half a percent of the sky seen from the star, so it intercepts roughly 0.5 percent of the star's output. If the star radiates the 4 × 10^26 W of a Kardashev Type II benchmark, that is about 2 × 10^24 W landing on the floor.

The spin follows from the same equation. At 1 g the rim moves at about 1,211 km/s, which is about 4.4 million km/h, and one turn takes about nine days. The star's own gravity barely helps: at 1 AU the Sun pulls with only about 0.006 m/s², roughly 0.06 percent of 1 g, so the ring has to supply nearly all of the inward force itself.

The material problem is that stress. The stress in the ring is σ = ρ × v² with v² of about 1.5 × 10^12 J/kg. For a material as dense as carbon nanotubes, 1,300 kg/m³, the stress comes to roughly 2 × 10^15 Pa, or about 2 million gigapascals. The best single-tube nanotube measurements are 63 to 100 GPa. The requirement scales with density, so a denser material would need even more.

The important feature is that more material does not fix the problem. The specific-strength test is independent of thickness. A ring whose material cannot reach v² cannot hold itself up at any thickness, because the extra mass adds extra outward pull in exact proportion. For steel the shortfall is about 9.5 million to one. For ideal nanotubes it is about 30,600 to one. Niven's answer was a fictional material he called scrith, which is another way of saying the required strength was left to the reader's imagination.

Niven's novel needed night, so it added shadow squares: an inner ring of panels, joined to one another by thin ultra-strong wire, that pass in front of the star. It also needed the ring to stay centered, which it does not do on its own. After the book came out, fans pointed out that a rigid ring is not truly in orbit around its star and would drift into it. MIT students at the 1971 Worldcon chanted, "The Ringworld is unstable!" and Niven wrote The Ringworld Engineers in part to address the engineering problems.

The instability is easiest to see as a ball resting on top of a hill. In the plane of the ring, the star's pull is strongest on whichever side is closer, so any small sideways drift gets amplified. A real ring would need active station-keeping, with something pushing against every drift for as long as it exists. Freeman Dyson, who proposed enclosing a star with collectors in 1960, said that a solid shell around a star is mechanically impossible, and preferred a loose swarm of independent objects for that reason.

Where the mass would come from

Strength is the first wall, and mass is the second. The mass of a megastructure is the area times the average mass per square meter, and the second figure is easy to underestimate because a habitat floor has to carry air, soil, and shielding. The Stanford torus is the best-documented case, at about 10 million metric tons, or 10^10 kg.

The table gives order-of-magnitude figures. The Bishop and ringworld rows come from the arithmetic in this article, not from published designs, so treat them as scale markers.

Structure	Mass estimate	Basis
Stanford torus	About 10^10 kg (10 million tonnes)	NASA/Stanford summer-study design
One Island Three cylinder	At least 10^12 kg	Air alone: about 1.6 × 10^12 m³ of interior at roughly half Earth-normal pressure
Bishop ring	At least 3 × 10^16 kg	Air at Earth-normal pressure over 3 × 10^12 m² of floor
Ringworld at 1 tonne per m²	1.5 × 10^24 kg, about a quarter of Earth's mass	1.5 × 10^21 m² times 1,000 kg/m²
Ringworld at 1,300 tonnes per m²	About Jupiter's mass (1.9 × 10^27 kg)	Same area, much heavier floor and walls

For the cylinder row, the volume is π × (4 km)² × 32 km, about 1.6 × 10^12 m³. At about half of Earth-normal air density, roughly 0.6 kg/m³, that is nearly 10^12 kg of air, so the whole cylinder must weigh more than that.

The ringworld rows show why dismantling a planet comes up in the literature. One tonne of material per square meter is only half a meter of rock, thinner than the soil on many hillsides, and it already costs a quarter of Earth's mass. A floor that carries oceans, walls hundreds of kilometers tall, and real structural thickness lands at around Jupiter's mass, which NASA's fact sheet lists as 1,898 × 10^24 kg. Disassembling and relocating that much matter is an energy project on a scale only a Kardashev Type II civilization would contemplate. A Dyson swarm runs into the same energy logic from the other direction, as the final sections below explain.

Closer to home, the mass problem is the argument for building from the Moon and asteroids instead of Earth. Lifting from the Moon takes much less energy than lifting from Earth, and the 1975 study built its construction plan around that idea. The Artemis program is a long way from mass-driver construction, but it is the same direction of travel.

Would anyone build one, and could we see it?

The reason these ideas keep returning is energy. Nikolai Kardashev's 1964 paper "Transmission of Information by Extraterrestrial Civilizations," published in Astronomicheskii Zhurnal, grouped hypothetical civilizations by the power they control. A Type I civilization uses the energy reaching its home planet, a Type II uses a star's full output at about 4 × 10^26 W, and a Type III uses a galaxy's at about 4 × 10^37 W. Carl Sagan's interpolation puts humanity at about 0.73 on the scale, based on a world energy consumption of about 595 exajoules, or about 18.9 TW, in 2021.

Freeman Dyson's 1960 paper in Science argued that a civilization with growing energy needs would eventually capture much of its star's output, and that the resulting waste heat would show up as infrared light. He later described a swarm of independently orbiting collectors, not a rigid shell. The swarm solves the strength problem by removing it: nothing is attached to anything, so nothing carries tension. The what if we built a Dyson sphere article follows what that would do to the Solar System. Other proposed stellar-scale projects include the Shkadov thruster, which pushes a star by reflecting its light on one side, and star lifting, which removes part of a star's matter for use elsewhere.

Searches for the signatures have produced candidates and no confirmations. A Fermilab search flagged 17 ambiguous infrared-heavy sources, none confirmed as engineered. The star KIC 8462852, which made headlines for irregular dimming after citizen scientists flagged it in 2015, was later attributed to dust. In 2024, Suazo and colleagues, using the Gaia, 2MASS, and WISE surveys under the name Project Hephaistos, listed seven M-dwarf candidates within 1,000 light-years; the authors avoided strong claims, and dust-obscured background galaxies are one alternative explanation.

Whether any civilization has built anything like this is the Fermi paradox in a sharper form. If building at this scale were easy and common, the sky should show it. The silence is data, though nobody agrees on what it says.

For humanity, the nearer question is whether a settled Solar System could ever use designs like these. The Mars settlement engineering article covers the problems of a first home beyond Earth.

The longer view is a thought experiment about an interstellar civilization, which pushes the timeline out by thousands of years and asks what would have to change first.

What people get wrong about megastructures

"A ringworld breaks the laws of physics." It does not. Rotation, tension, and orbital mechanics all work as written. What fails is the table of known materials, which is a statement about what we can make, not about what nature allows. The strongest materials article lists what actually exists, including nuclear-density matter that no engineer can handle.

"The ring's gravity comes from its mass." A ringworld's mass would produce a pull far too weak to matter. For a floor averaging 1,000 tonnes per square meter, a flat-sheet estimate (2πGσ) gives about 0.0004 m/s², roughly one twenty-thousandth of 1 g. The felt gravity is the inward push of the floor on anyone standing on it, which is purely a rotation effect. This is why you cannot live on the outside: there, the same rotation would throw you off into space.

"Rotating habitats are science fiction." The physics is ordinary, and the 1975 study treated the Stanford torus as an engineering design with mass budgets and a construction logistics plan. What does not exist is the launch industry, the lunar mining, and the political will. Those are large gaps, but they are not material limits.

"A Dyson sphere is a solid shell." Dyson himself rejected that idea. The versions in fiction are solid because they are easier to draw. The version he proposed is a swarm of independent orbiters.

"Bigger is just harder." Past a certain size the difficulty is not smooth. The first three designs in the table need ordinary or near-term materials, and the last two need strengths that increase by factors of hundreds or more. The scale ladder has a cliff, and it sits between the Bishop ring and the Orbital.

Frequently asked questions

Is a ringworld physically possible? A: No law of physics forbids one, but no known material can build it. At 1 g and 1 AU the ring would need a specific strength about 30,600 times beyond the best nanotube figure, and adding more material does not help. It would also be unstable in the plane of the ring and need continuous active correction.

How fast would a ringworld spin? A: About 1,211 km/s at the rim for 1 g at a radius of 1 AU, so one turn would take about nine days. The star's own gravity supplies only about 0.06 percent of the needed inward force at that distance.

What is the difference between a ringworld, a Bishop ring, and a Banks Orbital? A: They are the same idea at three scales. A Bishop ring has a radius of about 1,000 km, a Banks Orbital about 1.85 million km (3.7 million km across), and a Niven ringworld 150 million km. Bigger rings need stronger material, and the jump from Bishop to Orbital is about 1,900 times in strength demanded.

Did NASA really study space settlements? A: Yes. In summer 1975, a ten-week program at Stanford University and NASA Ames Research Center produced Space Settlements: A Design Study (NASA SP-413, 1977). It chose a 1.8 km torus for 10,000 residents at the Earth-Moon L5 point, with about 95 percent of the 10 million tonne mass in a lunar-soil radiation shield. Nothing like it has been built.

What is the biggest spinning habitat steel could support? A: With no safety margin, steel's specific strength of about 154 kJ/kg limits a 1 g hoop to a radius of about 16 km. O'Neill's 4 km radius leaves a margin of about four, which is why his cylinder fits steel, while Bishop's ring requires nanotube-class material.

Could astronomers spot a megastructure? A: In principle, yes. Dyson's 1960 paper pointed to excess infrared light as the signature. Searches have found ambiguous candidates, but the best-studied dimming star, KIC 8462852, was later attributed to dust, and the seven 2024 Project Hephaistos candidates have alternative explanations.

The takeaway

The spin equation turns a vague debate into a ladder. Torus and cylinder designs from the 1970s sit near the bottom, within reach of ordinary engineering materials. Bishop's ring sits on the nanotube edge, where the answer depends on whether bulk fibers ever approach the best single-tube numbers. Orbitals and ringworlds are not hard versions of the same project. They are past the point where any known material, in any thickness, can hold itself together.

That is a more interesting result than a flat "impossible." It says which part of the problem is solved (gravity from spin works), which is open (strength), and which is the real price of admission (mass and energy at stellar scale). It also explains Dyson's quiet correction: if the shell cannot hold, break it into a swarm and let orbital mechanics do the work.

Sources

• Ringworld, publication, awards, area, shadow squares, and the stability debate
• O'Neill cylinder, Island Three dimensions and layout
• Stanford torus and the National Space Society's host copy of NASA SP-413, Space Settlements: A Design Study
• Bishop ring (habitat) and Orbital (The Culture)
• Carbon nanotube, tensile and specific strength figures
• Dyson sphere, Kardashev scale, and astronomical engineering
• NASA planetary fact sheets for Earth and Jupiter

Related Reading

• What Are Dyson Spheres and Could We Actually Build One?
• How Civilizations Could Survive the Death of the Universe
• The Mysteries of Time: Why Aging Slows Down in Space
• What Is the Kardashev Scale, and Could Humanity Ever Reach Type III?
• Why Space Junk Could Lock Us Out of Space Forever