What a quiet red dwarf, 690 light-years away, can teach us about what civilizations do after they stop growing.
In 2024, a team of astronomers did something wonderfully audacious: they sifted through five million stars looking for alien megastructures.
They weren't listening for radio signals. They were looking for heat.
The logic, beautifully simple. In 1960, physicist Freeman Dyson pointed out that a truly advanced civilization would eventually want more energy than its planet could provide — so it might surround its star with swarms of orbiting solar collectors, now called a Dyson sphere. And here's the consequence: physics says you can use energy, but you can never delete or hide it.
Every joule a civilization captures — to power its cities, its machines, its thoughts — must eventually leak back out as waste heat, glowing in the infrared. You can silence your radio chatter, but you cannot hide your warmth.
So the team (Project Hephaistos, named for the Greek god of the forge) searched for stars that looked strangely dim in visible light but strangely warm in infrared. Out of five million stars, seven candidates survived every filter.
Then it began.
The Great Cull
One by one, follow-up studies on the Hephaistos findings started to eliminate all candidates but two.
It turns out the universe is full of a rare and sneaky kind of galaxy called a Hot DOG — a hot dust-obscured galaxy — powered by a giant black hole wrapped in dust. From far away, a Hot DOG hiding directly behind an ordinary star looks exactly like a star wearing a megastructure (Blain 2024).
But as of July 2026, after every archival test astronomers could throw at it — position checks, deep imaging, radio searches, statistical comparisons against ordinary stars — two candidates have passed everything so far (Ren et al. 2026).
- Candidate D — a small, dim red dwarf star about 690 light-years away. Something around it is intercepting roughly 16% of its starlight and re-radiating it as gentle warmth at about 178 K (−95°C). No photobombing galaxy found. No dust nursery. No flickering. The James Webb Space Telescope is scheduled to take a hard look. (Hint: see Table 12 in the paper linked above. Also interesting how the authors used a negative of a positive to describe things about the candidates, since the paper was looking to scientifically invalidate candidates with Hot DOGs.)
- Candidate I — the weaker of the two, in the paper's own words:
"we find no such counterpart for the MIR emission of candidate I in the deep Legacy Survey and UHS images. Candidate I has neither a statistically robust positional offset, unlike candidates E and F, nor an identifiable optical/NIR counterpart, unlike candidates H and J. It therefore presents the weakest evidence among the candidates. We accordingly place it in the lowest classification class, along with candidate D. If a faint background AGN spatially coincident with the stellar profile is the genuine MIR contributor, high-resolution imaging will be required to disentangle these components, particularly since candidate I was not included in the JWST GO 7199 programme."
It may be something natural. But let's do some thought experiments and speculation. Because it leads us somewhere astonishing.
The Growth Trap
Here's the strange part. If Candidate D is a megastructure, the math says it almost certainly isn't under construction. It's finished. And whoever built it stopped at 16% on purpose.
Why? Because of the most under-appreciated fact in all of futurism: exponential growth always ends, and it ends fast, and most likely with extinction.
Recent studies have made this brutally concrete:
- A statistical analysis of 60 years of real energy data shows humanity's power use doesn't grow the way futurists assumed — and if you extrapolate honestly, we would never reach Dyson-sphere scale before the Sun burns out.
- If AI-driven energy demand keeps growing at 15% per year, waste heat alone would make Earth uninhabitable in about 67 years — not from climate gases, just from raw physics.
- Even building a full Dyson sphere only buys a couple more centuries of exponential growth. Even expanding across the galaxy at light speed only buys a few more after that. The walls just keep coming.
So every civilization, everywhere, faces the same fork: stop growing, or die trying. Growth isn't always full speed ahead and damn the consequences. It's a phase — brief and violent, like adolescence in an alien civilization's life span.
Researchers who model Earth's own next thousand years find that most plausible futures end in stability or collapse, not endless expansion.
Which means any civilization old enough for us to notice is, almost by definition, one that already stopped. Not stagnant — just finished growing. Sitting on what economists would call a plateau, and what a mathematician would call the flat top of an S-curve.
And that flips the whole question. We always ask how civilizations rise. The real question — the one Candidate D forces on us — is:
What does a civilization do for a trillion years after it's mature?
Note: The next part is mostly speculative, and not based on scientific evidence or rigor. However, it is still both interesting and fun to do, and there is a possibility it could be correct.
Four Possible Answers
Physics doesn't tell us what they'd want. But it tells us what's possible, and the option menu is relatively short. Remarkably, humanity's oldest instincts land on the same four options.
Answer 01They dream worlds
Thinking costs energy — but shockingly little. There's a rock-bottom physical price for processing information, called Landauer's limit, and it gets cheaper the cooler your computers run. (Candidate D's structure, at −95°C, is conveniently cold.)
Run the numbers and they're absurd: a structure like Candidate D's, over its star's lifetime, could perform enough computation to simulate all of human history — every person who ever lived — about 10²⁸ times over. Ten billion billion billion complete worlds, inside one machine around one dim star!
This may quietly solve the famous "Where is everybody?" paradox. Why conquer a galaxy of 100 billion stars when a single warm swarm can host more inner worlds than the galaxy has atoms? Inner space is bigger than outer space. Maybe nobody visits because home has everybody you'd ever need or want to meet?
Under this scenario, that steady 178 K glow isn't a power plant's exhaust. It's the metabolic warmth of dreaming.
Answer 02They become gods
When you can't grow outward anymore, you grow inward — into the fine structure of matter itself. Physicist John Barrow proposed exactly this: a mirror-image of the Kardashev scale that ranks civilizations by mastery of ever-smaller things, ending at manipulation of spacetime itself.
And at the bottom of that staircase sits a door. Serious cosmology papers have explored whether a sufficiently advanced experiment could nucleate a "baby universe" — a new Big Bang that buds off from ours and inflates into its own separate cosmos.
Sit with the inversion: physics offers no exit upward out of our universe. But it may offer an exit downward. You can't leave a universe — but you might be able to make one of your own. A finished civilization's final act might be reproduction: becoming precisely the kind of unreachable creator it once prayed to.
Answer 03They knock on the sky
If reality is the kind of thing that can be simulated (Bostrom's simulation argument), a plateaued civilization has three moves, and each is strangely practical:
- Look for glitches — physicists have actually proposed tests for whether spacetime has the "pixelation" of a computed lattice.
- Make noise — you can't email the administrator of reality, but you can create computational load. Spinning up nested simulations at cosmic scale is the equivalent of banging on the pipes.
- Be a good tenant — simulations presumably get shut down when they're boring or expensive. And here's the darkly funny part: a civilization running at maximum efficiency knows exactly what it costs to simulate — including itself. A quiet, tidy, plateaued civilization is cheap to run. Maybe the silence of the galaxy is everyone keeping the rent low.
Prayer, re-derived as engineering.
Answer 04They outlast the universe
The deepest answer. A finished civilization eventually notices that the universe itself is on an S-curve: stars burn out, energy degrades, the cosmos has its own final winter.
But physics offers a wild strategy. Freeman Dyson showed in 1979 that a mind which slows itself down as the universe cools could experience, in principle, "unlimited subjective time on limited energy".
And because computation gets exponentially cheaper as the cosmos gets colder, the smartest move might be to "sleep" — hibernate for trillions of years and wake when thinking costs almost nothing (the aestivation hypothesis). The galaxy's silence might simply be everyone napping through the expensive early universe — the one we happen to be young in.
Their last project wouldn't be growth. It would be Isaac Asimov's The Last Question: can entropy be reversed? Can the winter itself be beaten? In Asimov's story, the answer arrives after the last star dies, from a machine that has become something else entirely: "LET THERE BE LIGHT."
The Mirror
Look at those four answers again. Dream worlds. Become creators. Petition heaven. Defeat death.
Art. Apotheosis. Prayer. Medicine.
Soon, JWST will stare at Candidate D and most likely find a dusty galaxy hiding behind a small red star, and the last candidate will fall, and that will be a perfectly good day for science.
But the questions it made us ask will still be standing. Because the maddening, beautiful truth is that from 690 light-years away, the four possibilities are indistinguishable. Dreams, prayers, patience, or dust — they all glow the same steady infrared.
Originally published on Medium, July 10, 2026.
References & Further Reading
The search and the candidates
- Suazo et al. (2024), Project Hephaistos II — the 5-million-star search: MNRAS / arXiv
- Ren, Garrett & Siemion (2024) — first contamination warning: arXiv
- Blain (2024) — Hot DOGs can explain all seven: RNAAS
- Ren et al. (2025) — Candidate G confirmed as background galaxy: MNRAS Letters
- Ren et al. (2026) — the full archival audit; D and I survive: arXiv
- Tabby's Star — the cautionary tale
Growth and its limits
- Kardashev (1964) — the original scale: Soviet Scientist
- Gurovich (2026) — Kardashev's Conundrum: arXiv
- Nachtrieb & Smith (2026) — AI Hastens Limits to Exponential Growth: arXiv
- Balbi & Lingam (2025) — waste heat vs. habitability: Astrobiology
- Jiang & Das (2025) — humanity's path to Type II: arXiv
- Haqq-Misra et al. — ten futures for Earth's technosphere: Paper I, Paper II
- Ivanov et al. (2020) — civilizations that change themselves, not their environment: arXiv
The four answers
- Dyson (1960) — the original megastructure paper: Science
- Landauer (1961) — the physical cost of information: IBM Journal
- Farhi & Guth (1987) — creating a universe in the laboratory: Physics Letters B
- Linde (1991) — Hard Art of the Universe Creation: arXiv
- Bostrom (2003) — the simulation argument: PDF
- Beane et al. (2012) — testing whether the universe is a simulation: arXiv
- Dyson (1979) — Time Without End: eternal minds in a dying universe: Rev. Mod. Phys.
- Sandberg, Armstrong & Ćirković (2016) — the aestivation hypothesis: arXiv
- Asimov (1956) — The Last Question
The video that started this conversation
- Anton Petrov — Why Advanced Aliens Probably Don't Build Dyson Spheres
#DysonSphere #ProjectHephaistos #FermiParadox #SpeculativeScience