Big Rip vs Big Crunch vs Big Freeze

In the contest of Big Rip vs Big Crunch vs Big Freeze, the Big Freeze (heat death) is still the likeliest ending: the universe expands forever and fades. The Big Rip needs dark energy that grows stronger; the Big Crunch needs it to turn negative. DESI results since 2024 hint that dark energy is weakening, which disfavors a rip but has not overturned the freeze.

All three endings come down to one quantity, dark energy, and one number that describes it, the equation of state w. This page compares them side by side. For the full timeline, see our guide to the future of the universe.

The comparison table

	Big Freeze (heat death)	Big Rip	Big Crunch
What dark energy does	Stays constant (w = −1) or fades but stays positive	Grows denser as space expands (w below −1, “phantom” energy)	Fades and turns negative, or there is a negative cosmological constant
What happens to expansion	Continues forever	Accelerates without limit, reaching infinite rate in finite time	Slows, stops, reverses
How it ends	Stars burn out, black holes evaporate, everything goes cold	Clusters, galaxies, planets, then atoms are torn apart	Everything is crushed into a hot, dense state
When	Last stars around 1014 years, last black holes around 10100 years	22 billion years from now in the w = −1.5 example	About 20 billion years from now in one 2025 model
Key paper	Adams and Laughlin 1997	Caldwell, Kamionkowski and Weinberg 2003	Luu, Qiu and Tye 2025
Status as of October 2026	Leading scenario	Disfavored by current data	Allowed by some models, not required by data

What sets the ending: dark energy and w

Dark energy is the unknown component, about 68% of the universe’s energy today (Planck 2018), that has made cosmic expansion speed up for the last several billion years. Its discovery came from supernovae in 1998 and 1999.

Its equation of state, w, is the ratio of its pressure to its energy density. Three cases matter:

• w = −1 exactly: a cosmological constant. Its density never changes as space grows, and the universe approaches steady exponential expansion. Result: Big Freeze.
• w below −1: phantom energy. Its density increases as space grows, so the push gets stronger without limit. Result: Big Rip.
• w above −1, rising over time: dark energy thins out. If it stays positive, expansion slows its acceleration but continues (Big Freeze). If the total dark energy turns negative, gravity can win (Big Crunch).

The Big Freeze

The Big Freeze, or heat death of the universe, is the default forecast of the standard ΛCDM model. Galaxies beyond our Local Group drift out of view within about 100 billion years, star formation ends around 10¹⁴ years, and the last black holes evaporate around 10¹⁰⁰ years. It is the ending traced by the five eras of the Deep Time explorer.

The Big Rip

The Big Rip was worked out by Robert Caldwell, Marc Kamionkowski and Nevin Weinberg in 2003. If w is below −1, the density of dark energy becomes infinite after a finite time. In their example (w = −1.5, Hubble constant 70 km/s per megaparsec), the end comes 22 billion years from now, in this order:

1. Galaxy clusters are erased about 1 billion years before the end.
2. The Milky Way is torn apart about 60 million years before.
3. The Solar System comes unbound a few months before.
4. Earth explodes about 30 minutes before.
5. Atoms are ripped apart about 10^-19 seconds before the end, followed by nuclei.

Measurements so far put w very close to −1. Planck combined with supernovae gives w = −1.03 ± 0.03, and DESI’s results favor w above −1 today, the opposite of phantom energy now.

The Big Crunch

In a Big Crunch, the universe stops expanding and falls back on itself. Before 1998 this was the main alternative to eternal expansion, and it depended only on whether there was enough matter. There is not: matter is about 31.5% of the critical density. So a crunch now requires dark energy itself to change sign.

In 2025, Hoang Nhan Luu, Yu-Cheng Qiu and S.-H. Henry Tye fitted a model with an ultralight axion field plus a cosmological constant to DES and DESI data. Their fits favor a negative cosmological constant. With their best-fit values, the universe expands for about 11 more billion years, then collapses, ending about 20 billion years from now for a total lifespan of about 33 billion years. The paper is peer reviewed (JCAP, 2025), but the date is only as good as the model, and the uncertainties are large.

What DESI changed, and what it did not

The Dark Energy Spectroscopic Instrument (DESI) maps the positions of millions of galaxies and quasars to measure baryon acoustic oscillations (BAO), a fixed ruler imprinted in the early universe. Comparing that ruler at different distances traces how expansion changed over the last 11 billion years.

Result	Date	Preference for evolving dark energy
DESI first year (DR1) + CMB + supernovae	April 2024	2.5, 3.5 or 3.9 sigma (Pantheon+, Union3, DES-SN5YR)
DESI DR2 + CMB	March 2025	3.1 sigma
DESI DR2 + CMB + supernovae	March 2025	2.8, 3.8 or 4.2 sigma (Pantheon+, Union3, DES-SN5YR)
DES supernovae recalibrated (DES-Dovekie) + DESI DR2 + CMB	November 2025	3.2 sigma, down from 4.2
DESI DR2 Lyman-alpha full shape + CMB, + supernovae	July 2026	2.7 sigma; 3.2 sigma with supernovae

The pattern in every combination is the same: w above −1 today (w₀ > −1) and decreasing into the past (w_a < 0). In plain words, dark energy looks like it is weakening now.

The caveats matter as much as the numbers:

• No result reaches 5 sigma, the threshold physicists use for a discovery. A 3-sigma result happens by chance about once in 370 tries; real experiments also have systematic errors that statistics do not capture.
• The answer depends on the supernova sample. The highest figure, 4.2 sigma, used the original DES five-year supernovae (DES 2024). After the DES team recalibrated that sample, its combined significance dropped to 3.2 sigma. With Pantheon+, the preference is 2.8 sigma.
• The CMB input matters. For DESI plus CMB, DESI reports that dropping CMB lensing lowers the preference from 3.1 to 2.7 sigma, and using only the most basic CMB information lowers it to 2.4 sigma.
• The best fits imply a “phantom crossing”. In the fitted curves, w dips below −1 at redshifts above about 0.4, in the past, before rising above −1. Simple physical models of dark energy struggle to do that, and it is unclear whether nature does.
• The formula is not built for the future. The standard parametrization, w(a) = w₀ + w_a(1 − a), is a fitting tool for the past. Extending it billions of years ahead is not a prediction; specific physical models, like the axion model above, are needed for that.

So what does this mean for the ending? Weakening dark energy makes the Big Rip less likely, because the data prefer w above −1 today. It keeps the Big Freeze as the default, because positive but fading dark energy still expands the universe forever. And it opens a window for the Big Crunch, but only in models where dark energy eventually turns negative. More detail is in is dark energy weakening?

What will settle it

DESI finished its planned five-year map in April 2026, with more than 47 million galaxies and quasars, and expects dark energy results from the full data in 2027 (details). Euclid, the Vera C. Rubin Observatory and the Nancy Grace Roman Space Telescope will test the same question with weak lensing and far larger supernova samples.

The other endings

Two more scenarios sit outside this three-way contest. In a Big Bounce, a crunch rebounds into a new expansion, as in the cyclic model of Steinhardt and Turok (2002); no observation supports it yet. And vacuum decay could end the universe as we know it regardless of dark energy, though almost certainly not for an extremely long time.