The universe has several possible endings, and they could hardly be more different, down to when they happen. The end might come a second from now, or hold off for a googol years (a one with a hundred zeros after it). Maybe it simply fades into cold and dark. Maybe every galaxy and atom is torn to pieces. Maybe everything falls back together into a single blazing point. Or maybe the strangest fate of all: the universe dies at the speed of light, with no warning, as the laws of physics quietly rewrite themselves from somewhere far out in space.

Which one actually happens isn’t a matter of taste. It comes down to physics we haven’t measured yet. Two unknowns do almost all the work: what dark energy will do over the next hundred billion years, and whether the vacuum we live in is the true bottom of things or just a ledge we happen to be resting on.

Here’s the list, and what each one needs to be true.

1. Dark energy and the vacuum: the two unknowns

The fate of the universe hangs on two open questions: how dark energy behaves over deep time, and whether empty space is really sitting at its lowest possible energy or only a temporary one. Almost every ending in this post is just a different answer to one of those two.

Start with dark energy; since it drives most of the endings. It’s the name we gave to whatever is causing the expansion of the universe to accelerate over time, and it makes up about 68% of everything in the cosmos. We know what it does, and roughly how much of it there is. What it actually is? We have no idea, and it earned its placeholder name from exactly that gap. The acceleration it drives was first spotted in 1998, in the light of supernovae going off in distant galaxies. For the question of how the universe ends, only one thing about dark energy really matters: a single number called its equation of state, written $w$.

That number is the ratio of dark energy’s pressure to its density:

$$w = \frac{p}{\rho}$$

and it controls how the density of dark energy changes as the universe grows. For a fluid with a constant $w$, that density follows a simple scaling law against the scale factor $a$, a single number that tracks how stretched-out space is and climbs as the universe expands:

$$\rho \propto a^{-3(1+w)}$$

Plug in three different values of $w$ and you get three different futures.

• $w = -1$ exactly. The exponent becomes zero, so the density never changes. Empty space holds a fixed amount of dark energy no matter how much it stretches. This is a true cosmological constant, and it drives a steady, eternal acceleration. It leads to heat death.
• $w < -1$. The exponent turns positive, so the density grows as space expands. The acceleration runs away with itself, faster and faster, without limit. This is “phantom” dark energy, and it leads to the Big Rip.
• $w > -1$ (but still below $-1/3$, which is where acceleration starts). The density slowly thins out as space grows. The expansion still accelerates, but gently, and if dark energy keeps weakening or eventually turns attractive, gravity can win the long game and pull everything back. That opens the door to a Big Crunch.

The second unknown is subtler. Empty space carries an energy of its own, set by the quantum fields that fill it, and nothing guarantees that the value we live with is the lowest one on offer. If it isn’t, the vacuum itself is unstable, and that opens an ending that owes nothing to dark energy at all. More on this later.

So what have we actually measured? Two things. Space comes back flat, and $w$ comes back within a few percent of $-1$. Both readings point straight at the cosmological constant ($w = -1$), and so at heat death. But “within a few percent” is not “exactly”, so we can’t yet confidently rule out the other cases. The DESI survey has even hinted, in its 2024 and 2025 results, that $w$ might not be a fixed number at all but one that drifts over cosmic time. If that holds up, it throws the door back open to the endings the flat-and-constant picture had closed.

In essence, we can’t make measurements accurately enough to draw a firm conclusion, and the very thing we are measuring seems to change over time which only complicates things.

2. Heat death: the long fade

If dark energy is a true cosmological constant, the universe expands forever and slowly drains of usable energy until nothing can happen anywhere. This is heat death, and it’s the ending today’s data points to most directly.

Imagine a time when our Sun had long since died, and the Milky Way had joined with Andromeda (~10 billion years from now). Over the hundred billion years that follow, steady acceleration drags every galaxy outside our Local Group across the cosmic event horizon, one by one, until the sky holds nothing but our own merged island of stars. And that emptying is only the beginning. The expansion keeps going, the lights keep going out, and the universe sinks through a run of eras so long they make those first hundred billion years look like a rounding error.

The first is the one we live in, the stelliferous era, the age of stars. It runs from now until somewhere around a hundred trillion ($10^{14}$) years from now. Stars need cold gas to form, and the supply is finite; once it runs dry, no new ones light up. The last stars still burning will be the smallest red dwarfs, which sip their fuel so slowly they can shine for trillions of years. When the final red dwarf gutters out, the age of starlight is over. And measured against the full span, we are absurdly close to the beginning, alive in the brief window when the sky is still full of fire.

Then comes the degenerate era, from about $10^{14}$ to $10^{40}$ years. By now the universe is a graveyard of stellar remnants: white dwarfs, neutron stars, cold brown dwarfs and black holes drifting in the dark. Whether even these last depends on a physics question we still haven’t settled, whether protons are truly stable. If protons decay, as some theories predict, with a half-life of maybe $10^{34}$ years or more, then given enough time the remnants themselves dissolve into radiation and ordinary matter simply stops existing. That “if” is a real one. We’ve never caught a proton in the act of decaying; we’ve only managed to show that if it happens at all, it’s staggeringly rare.

After that comes the black hole era, roughly $10^{40}$ to $10^{100}$ years. With ordinary matter gone, black holes are the last large things left, and even they don’t last. They leak energy as Hawking radiation and slowly evaporate, the smallest first. The largest supermassive ones, anchoring the cores of the biggest galaxies, take something like $10^{100}$ years to fade away entirely. When the last black hole evaporates, the universe’s final concentrated lumps of energy go with it.

What’s left is the dark era, everything past $10^{100}$ years: a thin, near-uniform, near-freezing soup of low-energy photons, neutrinos and stray particles, smeared more and more thinly across more and more space. No temperature differences are left worth the name, so there are no energy gradients, so nothing can drive anything to happen. Matter doesn’t burn, structures don’t form, and nothing changes anywhere, ever again. The universe has reached maximum entropy and quietly stops having a story.

The name “heat death” is a little misleading. The end state isn’t hot at all. It’s the coldest thing imaginable, a hair above absolute zero. The “heat” is the thermodynamic kind: the universe dies not by burning up but by settling to one single temperature everywhere, with no hot and no cold left to play off each other. The equilibrium is the death.

3. The Big Rip: if dark energy strengthens

If dark energy is phantom energy, with $w < -1$, its density grows as the universe expands instead of holding steady, and that runaway eventually tears apart everything held together by any force, ending the universe in a finite time. Unlike heat death, the Big Rip has a date.

It’s the scaling law from section 1 read in its most violent register. With $w < -1$, more space means more dark energy packed into every volume, and more dark energy means faster acceleration, which makes still more space. The loop feeds itself and runs away. The cleanest way to picture it is through the cosmic horizon, the distance beyond which space is receding faster than light can cross it. Under phantom energy that horizon doesn’t sit still. It shrinks, drawing inward toward every observer. And as it closes past the size of a bound structure, that structure can no longer hold itself together, because the space inside it is now stretching faster than its own forces can pull back.

So the universe comes apart from the largest scales down, in order:

• First the galaxy clusters are pulled apart, their member galaxies scattered.
• Roughly 60 million years before the end, galaxies themselves unbind, including our own.
• A few months before the end, planetary systems fly apart as planets drift off their stars.
• About half an hour before the end, stars and planets are torn open.
• In the final fraction of a second, around $10^{-19}$ seconds before the end, atoms are ripped apart, then their nuclei.
• At the final instant, the expansion rate goes infinite and spacetime itself is shredded in a singularity, a rip rather than a collapse.

How far off is “the end”? That depends on just how negative $w$ is. For an illustrative value of $w = -1.5$, the calculation that first laid out this scenario put the rip about 22 billion years from now, less than twice the present age of the universe. Push $w$ back toward $-1$ and the date slides off into the future; the closer dark energy sits to a plain cosmological constant, the longer the reprieve.

Which is also why the Big Rip is the long shot of the bunch. The measured $w$ hugs $-1$ very tightly, and a phantom rip needs it to sit below $-1$. The error bars do reach a little way onto the phantom side, so nobody can rule it out, but the data leans the other way.

4. The Big Crunch: if expansion reverses

If dark energy weakens, switches off, or turns attractive, gravity can regain the upper hand, halt the expansion, and drag everything back into a hot, dense end-state. That reversal is the Big Crunch, the Big Bang run backward.

For most of the twentieth century this was the leading guess, and the logic was simple: pack enough matter into a universe and its own gravity should put the brakes on the expansion, slowing it, stopping it, then reeling everything back in. Two discoveries killed it as the default. The expansion turned out to be speeding up, not slowing down, and space turned out to be geometrically flat. A universe like ours won’t recollapse on its own. For a crunch to happen now, dark energy would have to change its behavior, and some models let it do exactly that. In the quintessence picture, dark energy is a field slowly rolling down a slope of potential energy, and that slope can dip below zero, flipping the cosmic push into a pull. Once it does, the expansion slows, stops, and goes into reverse.

From there the collapse is the expansion film run backward. Distant galaxies stop redshifting away and start blueshifting toward us. The relic glow of the Big Bang, today a faint cold microwave hum, blueshifts back up into a furnace. Structures that took billions of years to drift apart come slamming together again, the sky climbs toward infinite heat, and everything ends crushed into a singularity that looks a lot like the one it all began in.

4.1. The bounce and cyclic universes

If the crunch doesn’t bottom out in a true singularity but rebounds instead, you get a Big Bounce, and maybe an endless chain of universes behind it. The hope is that whatever quantum-gravity physics rules that final, densest instant produces not a hard stop but a turnaround, with the collapse feeding straight back into a new expansion. One universe’s crunch becomes the next one’s bang.

Stretch that into a pattern and you get a cyclic universe: bang, expansion, slowdown, crunch, bounce, and around again. There are a few flavors of this. The ekpyrotic and cyclic models of Steinhardt and Turok drive the cycles with a dark-energy-like field that takes turns accelerating and contracting the universe; “ekpyrotic” comes from the Greek for “out of fire,” the picture of each new expansion igniting from the ashes of the last collapse. Roger Penrose’s conformal cyclic cosmology takes a stranger road altogether: it argues that the empty, mass-free, clockless end state of heat death is geometrically indistinguishable from the smooth, low-entropy state a fresh Big Bang begins in, so the dead universe quite literally is the next one’s beginning, in an endless run of what Penrose calls “aeons.”

Every one of these models has to answer a classic objection, one that goes back to Richard Tolman in the 1930s: entropy should pile up from one cycle to the next, so a naive cyclic universe can’t have been running forever, and each cycle would come out looking different from the last. The modern versions get around it by letting the expansion phase dilute the built-up entropy before the next contraction, wiping the slate clean. None of this has the direct evidence behind it that heat death does. But the math still hangs together, and a bounce stays a real possibility for whatever happens at the bottom of a crunch.

5. Vacuum decay: if the ground isn’t the bottom

There’s one fate that doesn’t depend on dark energy at all, and it could begin anywhere, at any instant, including this one. We’d never see it coming, because the only warning travels at the speed of light, arriving at the very same moment the thing itself does. It’s called vacuum decay, and whether it can happen at all comes down to a single question: is empty space already at its lowest possible energy, or are we perched on a higher rung, a so-called false vacuum?

Quantum fields fill all of space, and the vacuum is just those fields settled into their lowest-energy arrangement. The catch is that “lowest one nearby” and “lowest there is” aren’t the same thing. A field can come to rest in a dip in its energy landscape that’s a local minimum without being the deepest one, the way a ball can settle into a hollow partway down a hillside instead of rolling all the way to the valley floor. A resting spot like that is a false vacuum: stable against small nudges, but not the true bottom.

If our vacuum is false, quantum mechanics hands the universe an escape hatch. Somewhere, purely by chance, quantum tunneling can pop a tiny bubble of true vacuum into being, a speck of space that has dropped to the genuine lowest energy. Let that bubble clear a certain critical size and the energy it releases drives its wall outward, accelerating it to nearly the speed of light. Inside, the constants and laws of physics take on different values, ones that may not allow atoms, or chemistry, or any structure at all. Everything the wall sweeps over is rewritten and erased. And since the wall moves at almost the speed of light, there’s no advance notice to be had. The first sign of it would also be the last: the wall arriving, carrying the news of its own existence.

And here’s the part that should give you pause: this isn’t just a thought experiment. The measured mass of the Higgs boson, about 125 GeV, together with the mass of the top quark, puts the Standard Model right on the knife-edge between a stable vacuum and a metastable one. The best calculations we have come down on the metastable side, which means our vacuum is probably a false one. Before you lose any sleep over it, those same calculations put the expected wait vastly longer than the present age of the universe, well past even heat death’s $10^{100}$ years. And physics we haven’t discovered yet could nudge the answer back to fully stable. But as far as the equations we currently trust can tell, the ground under our feet may not be the real floor.

6. The verdict: heat death, narrowly

Line the contenders up against the measurements and they sort themselves out fairly cleanly. Flat space and an equation of state within a few percent of $w = -1$ favor the slow freeze, push a near-term Big Crunch out to the margins, and leave the Big Rip waiting on a phantom value the data just doesn’t show. Vacuum decay, for its part, runs on a clock so slow it would only arrive long after the freeze anyway, if it ever arrives at all.

So heat death wins, but only on points. And the margin is thinner than it looks. Dark energy is more than two-thirds of everything there is, and we honestly don’t know what it is; the whole verdict leans on the assumption that its $w$ stays pinned at $-1$ forever. Let that assumption wobble and the picture shifts, and it has already started to wobble: DESI’s recent results hint that $w$ may be drifting rather than holding still. If dark energy is fading, a future slowdown, even a reversal, climbs back onto the table. If it’s strengthening, the Rip does. And the physics that would settle what happens in a bounce, or whether our vacuum is truly stable, is quantum gravity, which nobody has solved.

The slow freeze is the front-runner. But the freeze, the rip, and the crunch all hinge on the same missing number, and the last ending needs no number at all. The most honest answer is that the universe’s final page hasn’t been deciphered yet, because we don’t yet have the physics to read it.

There’s something fitting in that. We happen to be reading from very near the opening of the story, with most of the pages still ahead of us and the ending genuinely undecided. We can make out the shape of the question well enough to lay the possibilities side by side. We just can’t turn to the last page yet.