How will the universe end?

Two independent teams, measuring the brightness of exploding stars of a standard kind at great distances, found in 1998 and 1999 that those explosions were fainter than a decelerating universe predicted.

The reading that survived is that the expansion slowed at first and has been accelerating for the more recent part of cosmic history. It has since been confirmed by the microwave background and by the imprint of sound waves from the early universe on the distribution of galaxies, which are entirely different measurements.

Star formation runs down as galaxies exhaust their gas. Adams and Laughlin’s survey of the long term puts that beyond one to ten trillion years, with the smallest hydrogen-burning stars still going at ten trillion, and significant numbers of stars only out to about a hundred trillion.

Every ending under discussion is remote enough that the difference between them is a difference in what kind of universe this is, not in what anyone should do about it.

The simplest description is a cosmological constant: a fixed energy per unit volume that does not dilute as space expands. That has fitted the data well since 1998.

In 2024 and again in 2025, baryon acoustic oscillation measurements from a large spectroscopic survey, combined with the microwave background and supernovae, preferred a dark energy whose strength changes with time, at a significance the collaboration reported in 2025 as 2.8 to 4.2 standard deviations depending on the supernova sample.

Corrections to the supernova samples since have narrowed that range: a 2026 reanalysis puts all three at 3.2 to 3.4, and the team behind the sample that gave 4.2 now reports 3.2, which it calls a weak preference. Whether the signal is dark energy at all is disputed: George Efstathiou argued that a small offset between supernova samples could bring the most discrepant one into line with a constant.

That is interesting rather than settled. It is also the single measurement that most changes the answer to this question.

Some theories that unify the forces predict that protons decay, on timescales enormously longer than the present age of the universe. If they do, ordinary matter eventually evaporates, by a process that does not depend on dark energy.

It has been looked for. A detector in Japan watched a large volume of water from 1996 to 2018 and saw no candidate events in the decay channel many of those theories favour, which puts the partial lifetime for that channel above 2.4 times 10 to the 34 years, at 90 per cent confidence.

Searches like this one have put the simplest supersymmetric unified theory under serious strain and leave others alive. It does not establish that protons are stable, which is not a thing an experiment can establish.

If dark energy is not merely constant but strengthens, the acceleration eventually tears apart every bound structure, at smaller and smaller scales: clusters, then galaxies, then solar systems, then planets, then atoms. A 2003 paper worked out the sequence and the timing.

Current measurements do not favour this. It remains inside the error bars of what has been measured, which is the only reason it is here.

The measured mass of the Higgs boson sits close to the boundary between a stable vacuum and a metastable one. In the metastable case the vacuum state of the universe is not the lowest available, and could in principle decay, with the new state expanding at the speed of light.

A 2018 calculation, which its authors describe as the first complete one, puts the expected lifetime at around 10 to the 161 years, with an uncertainty of many orders of magnitude. It assumes no physics beyond the Standard Model, and new physics at high enough energies could make the vacuum stable or destabilise it, which is a substantial assumption.