A disc of reflective foil so vast it could harvest a third of a star’s entire energy output can hold itself in stable orbit for billions of years with no engines, no computers, and no one watching it, which means if a civilisation built one and then went extinct, the structure could still be hanging there right now.
Both concepts under examination sit near the top of the Kardashev Scale, the framework astronomers use to rank civilisations by energy consumption. A Type I civilisation harnesses the full output of its home planet. Type II controls the entire output of a star. The structures at the centre of this new physics are Type II technology, and for decades the core assumption has been that even if you could build them, they would be mechanically useless because any small nudge would send them drifting away or crashing inward. Findings published in Monthly Notices of the Royal Astronomical Society in January 2026 quantify the precise conditions under which two categories of alien megastructure, called stellar engines and Dyson bubbles, can achieve passive stability around a host star. That assumption is now wrong.
A stellar engine, in its simplest form, is a reflective disc hovering above a star. The star’s radiation pressure, the physical push that light exerts on any surface it strikes, pushes the disc outward. The star’s gravity pulls it back. When those two forces balance, the disc hangs in place. Because the disc and the star are gravitationally coupled, the whole system, disc and star together, slowly accelerates through space, physically relocating the star across the galaxy over millions of years. The concept was first proposed in 1987 by Soviet engineer Leonid Shkadov and has sat in the theoretical toolbox of megastructure physics ever since, with one persistent problem: the balance point is always unstable. Push the disc slightly inward and gravity wins; it falls. Push it slightly outward and radiation pressure wins; it drifts away forever. Every previous treatment confirmed the equilibrium was a knife edge.
The knife edge disappears if you change how the disc’s mass is distributed. A uniform flat reflective disc is always unstable, and the reason becomes clear once you drop the simplifying assumption that the disc behaves like a point mass and do the force calculations properly for an extended physical object. A disc large enough to intercept meaningful starlight is so physically wide that photons from the star’s edges strike the disc’s outer rim at an oblique angle rather than head-on, changing the effective radiation pressure force in ways that the point-mass approximation cannot capture. Once you account for the real geometry, a clear path to stability opens up, and it requires nothing more exotic than rearranging where the mass sits. Concentrate the disc’s mass into a heavy structural ring at the outer edge, with a thin reflective film spanning the interior, and the force balance reverses its behaviour entirely. Move the ring-engine closer to the star and radiation pressure now pushes it back outward; move it farther away and gravity pulls it back inward. Physics locks it in place with no active correction required, the same self-correcting property that keeps a ball settled at the bottom of a bowl rather than balanced on top of one.
For our Sun, the reflective film spanning such a ring structure works at a thickness of just 15 nanometres of aluminium, roughly 5,000 times thinner than a human hair, and the structure still functions as a stable engine. The ring must sit at a distance greater than approximately 0.7 times its own radius from the star’s centre, which translates geometrically to the disc subtending a half-angle of up to 55 degrees as seen from the star. That constraint is not merely a theoretical nicety. At that precise stability boundary the structure intercepts exactly one-third of the star’s total light output, which is also the maximum energy fraction physically available before the force balance collapses. A stellar engine at the stability boundary around a Sun-like star intercepts one-third of roughly 3.8 times 10 to the power of 26 watts, more energy than all of human civilisation has consumed across its entire recorded history, delivered continuously, every second.
The second structure the new physics addresses is the Dyson bubble, which is meaningfully different from the Dyson sphere most people picture. A Dyson sphere is a solid enclosed shell surrounding a star, an engineering problem so extreme it makes a stellar engine look modest by comparison. A Dyson bubble replaces the rigid shell with an enormous cloud of individual reflectors, each one hovering in place with radiation pressure balancing gravity, collectively enveloping the star without the need for a monolithic structure. The immediate problem is the same one that plagued the stellar engine: a single reflector in static equilibrium above a star is always unstable, and a cloud of individually unstable reflectors looks, at first glance, like an even bigger mess.
Packing enough reflectors together fixes everything, and the mechanism is straightforward once you see it. In a sufficiently dense cloud, each reflector partially blocks the starlight reaching the reflectors deeper inside it. That mutual shading changes the effective force law for radiation pressure. Instead of dropping off with the square of distance as light normally does in open space, the effective radiation push drops off faster, because some of it never arrives, absorbed by the layers of reflectors above. That steeper falloff creates a restoring force where none existed before, and it activates for any positive cloud density, with no minimum threshold required. Even a moderately populated bubble starts self-correcting rather than dispersing. A dense cloud also carries its own mass, and the cloud’s self-gravity provides a second completely independent stabilising mechanism. A reflector sitting inside the cloud feels a net inward gravitational pull from all the material at smaller radii, which adds to the star’s own gravity on the inward side and steepens the effective force gradient sufficiently to maintain stability even when the shading effect alone is too weak to matter.
The longevity that passive stability enables is the part that sits uncomfortably with the idea that we are alone. A structure requiring constant active control fails the moment the control system fails, and on geological timescales, control systems always fail. A passively stable structure has no such vulnerability. It persists as a function of geometry and physics alone, for as long as the host star keeps burning. A Dyson bubble built by a civilisation that no longer exists retains its equilibrium regardless, sitting in stable configuration around its host star, producing a detectable signal with no one home to generate it. A cloud of reflectors enveloping a star produces an infrared excess, a surplus of heat radiation against the star’s normal optical output, because the reflectors intercept visible light and re-radiate it as warmth at longer wavelengths. That spectral signature is exactly what infrared SETI surveys have been scanning stellar catalogues to find.
Stars age and brighten, which raises an obvious question about whether any equilibrium survives long enough to matter. A star like the Sun gets roughly 10 percent brighter per billion years, shifting the radiation pressure balance and in principle disrupting any fixed hovering position. For a sparse Dyson bubble with reflectors in simple inverse-square equilibrium, any luminosity increase immediately pushes every reflector outward with nothing to stop the drift, destroying the structure gradually but completely. For a dense self-stabilising bubble the situation is different: the equilibrium position migrates outward quasi-statically as the star brightens, tracking the slowly shifting force balance rather than rupturing under it, and the new equilibrium distance remains finite and defined throughout the star’s main-sequence lifetime. The ring stellar engine shows equivalent resilience, with its equilibrium surviving a stellar luminosity increase of up to 73 percent before the configuration finally breaks down, giving structures built around young stars a margin that spans multiple billions of years of stellar evolution.
A Dyson swarm, where reflectors orbit the star rather than hovering stationary, produces a different and separately searchable signature. The orbital motion of individual reflectors transiting the stellar disc generates a flickering pattern in the star’s observed brightness, distinct from the smooth continuous infrared excess of a static Dyson bubble. The stable orbital zones for swarm reflectors form a well-defined band in the parameter space of orbit radius against reflector lightness number, the ratio of radiation pressure force to gravitational force for a given reflector design, and configurations within that band follow closed predictable orbits with no tendency to drift or collide. Two search targets emerge from the same underlying physics: flickering at characteristic periods for swarms, steady infrared excess for bubbles, and the two signatures are mutually exclusive for a given structure type, which means a detection of either one immediately constrains what kind of megastructure is present.
No confirmed stellar engine or Dyson bubble has been detected. Several infrared surveys have returned candidate stars with anomalous heat output, and KIC 8462852 generated years of public controversy over irregular dimming that some researchers tentatively attributed to orbiting megastructures before dust and comet debris gained wider acceptance as explanations. Current infrared survey programs, including archival data from the Wide-field Infrared Survey Explorer, carry sensitivity sufficient to detect a structure intercepting more than a few percent of a Sun-like star’s output within several hundred light-years of Earth. The new stability physics narrows the search geometry: a passively stable stellar engine presents a half-angle of up to 55 degrees and an infrared fraction near one-third of total stellar output, while a self-stabilising Dyson bubble presents a smooth spectral modification consistent with a dense enveloping cloud and no transit signal. Those are precise, testable predictions, and the existing survey data has not yet been systematically searched against them.
Source:
McInnes, C. R. (2026). Stellar engines and Dyson bubbles can be stable. Monthly Notices of the Royal Astronomical Society, 546, 1–18. https://doi.org/10.1093/mnras/stag100
